The Impact of Trade Agreements and
Regional Economic Integration
on Trade Flows
― The Case of SAARC Countries ―
Suresh MOKTAN
DOCTORAL PROGRAM
GRADUATE SCHOOL OF INTERNATIONAL DEVELOPMENT
NAGOYA UNIVERSITY
The Setsutaro Kobayashi Memorial Fund
A Research Paper for 2007
Fuji Xerox Co.,Ltd.
EXECUTIVE SUMMARY
This research paper is submitted as a part of work developed with the financial support of the
Fuji Xerox Setsutaro Kobayashi Memorial Fund. It essentially draws from the two main
pillars (chapters) of the author’s Ph.D. dissertation. The paper in effect is divided into three
parts. Part I consists of the manuscript, which is due to be published in the Journal of
International Economic Studies in Volume No. 23 by the end of March 2009. Part II consists of
the material taken from the manuscript that has been recently submitted for publication
consideration in the Journal of Policy Modeling. The conclusions and policy implications of
this research is provided in Part III.
Part I entitled The Impact of Trade Agreements on Intra-regional Exports: Empirical
Evidence from SAARC Countries employs the generalized Gravity Model using a distinctive
panel dataset for over a period spanning 35 years from 1971 to 2005. The model performs a
regression analysis and examines the impact of trade agreements on the volume of exports
amongst SAARC countries, which is the first objective of this research. Part II entitled
Assessing the Economic Impacts and Welfare Implications of SAFTA and SAFTA+5: The
South Asian Experience employs the Global Trade Analysis Project (GTAP) Model – a
multiregion, multisector applied general equilibrium (AGE) model. The GTAP Model
aggregates the original 87 regions to 10 regions, and 57 sectors to 20 sectors. The main aim of
this model is to perform simulation analysis and policy experiments in order to analyze the
second objective, i.e., to analyze the economic impacts of SAFTA as well as the welfare
implications on SAARC countries via FTAs with other countries/regions.
In this research, three hypotheses have been tested. In Part I, we test whether or not the
trade agreements amongst SAARC countries is positively associated with exports. In Part II, the
first hypothesis that is tested is to examine if selective combinations of tariff rates will result in
welfare gains for both contracting parties; the second hypothesis to test is whether the effects of
SAFTA, and FTAs with five observer countries/regions (SAFTA+5) will be welfare improving
to SAARC member countries, causing more trade creation as opposed to trade diversion.
Specifically, the research addresses the following key questions:
(1) What is the actual impact of having trade agreements amongst SAARC members? Have
they affected positively or negatively on the growth of exports?
(2) Did SAPTA play a catalytic role of enhancing intra-SAARC exports?
(3) What are the economic effects of SAFTA and SAFTA+5 on trade flows as a result of the
reduction in tariffs?
(4) What are the welfare implications of FTAs amongst SAFTA members and five observer
countries – China, Japan, South Korea, the United States and the EU?
(5) Which of the contracting parties are likely to have the most feasible FTAs?
As indicated above, this research exploits two different models to tackle these questions.
While Part I answers questions (1) and (2) employing the Gravity Model, Part II answers
questions (3), 4) and (5) employing the Global Trade Analysis Project (GTAP) Model. The
principal reason for implementing these two different models is that they are found to be the
most appropriate to tackle the aforementioned research questions. First, the Gravity Model is a
suitable tool to analyze the impact of FTAs on intra-regional exports. Based on the results
obtained from the Gravity Model, the next pertinent question that arises is about the welfare
implications for individual SAARC countries consequent to ratification of SAFTA, and also by
effectuating FTAs for inter-regional trade with interested observer countries. However, the
Gravity Model can not efficaciously estimate the welfare effects of FTAs or more specifically,
the trade creation and trade diversion effects. In order to answer the latter set of three questions,
the most appropriate tool is the global AGE model, such as the GTAP Model.
Thus, the significance of this research is not only to answer the aforementioned core
questions, but while addressing these questions the supplementary contribution is to show the
efficacy of trade agreements in furthering exports in the SAARC region. The focus, as such, is
to recount the findings after testing the effects of trade agreements before the inception of
SAARC and SAPTA, and comparing with the post-SAARC and post-SAPTA periods.
Secondly, this research provides an impression about the economic effects and welfare
implications resulting from preferential tariffs and FTAs amongst not only SAARC nations per
se (SAFTA effects), but also by way of extending the notion of free trade or aggregating with
other nations and regional blocs such as China, Japan, South Korea, the United States and the
EU (SAFTA+5 effects). Hence, examining the feasibility and potential of South-South
economic cooperation, and then expanding further to North-South economic cooperation via
FTAs with leading nations/regions can become a compelling case study. Thirdly, from the
lessons derived from this research, it is expected to stimulate constructive ideas to comprehend
more about the structure of trade complementarities and comparative advantages that might
help explain why intra-SAARC trade is low, and how trade as well as regional reciprocal
understanding and cooperation amongst SAARC as well as observer countries could be further
enhanced. The understanding of the veiled reasons could also provide incentive to policymakers
of SAARC countries to devise appropriate measures to boost up intra- as well as inter-regional
trade.
The originality of the research may be distinguished in that the estimation on the impact of
trade agreements on intra-regional exports of SAARC countries for over a long span of 35 years
by dividing the pre-SAPTA and post-SAPTA periods is the first of its kind. Furthermore, the
extension to GTAP Model in investigating the welfare implications of FTAs with the current
observers including China, Japan, South Korea, the United States and the EU is another
unexplored area.
Empirical tests based on the Gravity Model find scant evidence of the impact of trade
agreements on exports for the pre-SAARC and pre-SAPTA periods, but statistically significant
and positive impact is observed in the post-SAARC and post-SAPTA periods even amidst
sustained significant negative impact of conflict in all sub-periods. This propensity is
discernible regardless of the estimation methods applied. However, further tests reveal that the
positive impact emanated not expressly owing to SAPTA per se, but it is rather the coalesced
effect arising from the delayed impact of the existing trade agreements amongst SAARC
countries.
Findings from the GTAP Model demonstrate that the largest welfare gains for SAARC
countries materialize from plurilateral FTAs with deeper liberalization in selective sectors.
While the maximum possible FTAs emerge from varying tariff combinations, results
corroborate that selective tariff structure is welfare enhancing for both contracting parties
generating a number of feasible FTAs amongst SAARC and +5 countries. There is also clear
evidence showing that SAFTA and SAFTA+5 are welfare enhancing, resulting in net trade
creation as opposed to trade diversion. Albeit major fluctuations are observed in the industry
output, other economic variables such as household demand, aggregate exports and imports,
terms of trade, GDP, and allocative efficiencies of SAARC as well as +5 countries increase
significantly a propos SAFTA+5 scenario in particular.
ACKNOWLEDGEMENTS
The author is highly grateful to Professor Naoko Shinkai for her excellent advice, support and
insightful comments. Special thanks are due to Professor Mitsuo Ezaki, Professor Shigeru T.
Otsubo and Professor Hiroshi Osada for their constructive suggestions. The author would also
like to thank the participants of the Third Annual APEA Conference, Hong Kong University of
Science and Technology, Hong Kong, China, July 25-26, 2007, particularly Professor Yasuyuki
Sawada and Professor Sven W. Arndt. The paper has additionally benefited from valuable
comments from the participants at the International Conference on Policy Modeling (EcoMod
2008) held in Berlin, July 2-4, 2008.
The financial support from the Fuji Xerox Setsutaro Kobayashi Memorial Fund and Sato
International Scholarship Foundation is gratefully acknowledged. The usual caveat about the
remaining errors applies.
Suresh MOKTAN
Content
PART I
THE IMPACT OF TRADE AGREEMENTS ON INTRA-REGIONAL EXPORTS:
EMPIRICAL EVIDENCE FROM SAARC COUNTRIES ......................................... 1
1.
Introduction ....................................................................................................................... 1
2.
Literature Review.............................................................................................................. 3
3.
Methodology ..................................................................................................................... 6
3.1 Model Specification .................................................................................................... 7
3.2 Correction for Endogeneity Bias and Heteroskedasticity ........................................... 8
3.3 Computation of Variables ........................................................................................... 9
4.
The Data .......................................................................................................................... 10
5.
Empirical Results ............................................................................................................ 11
5.1 Pooled OLS Estimation (Benchmark)....................................................................... 11
5.2 Robustness Checks.................................................................................................... 14
5.3 Effect of SAPTA ....................................................................................................... 17
5.4 Lagged Effects of TRAG .......................................................................................... 18
6.
Summary of Findings ...................................................................................................... 19
PART II
ASSESSING THE ECONOMIC IMPACTS AND WELFARE IMPLICATIONS
OF SAFTA AND SAFTA+5: THE SOUTH ASIAN EXPERIENCE....................... 21
1.
Introduction ..................................................................................................................... 21
2.
Some Issues on Free Trade and Welfare ......................................................................... 22
2.1 Bilateralism or Plurilateralism? ................................................................................ 22
2.2 Free Trade and National Welfare .............................................................................. 24
2.3 SAFTA and Broader Economic Agenda ................................................................... 24
3.
Methodology and Data .................................................................................................... 26
3.1 The GTAP Model and the AGE Framework ............................................................ 26
3.2 Model Calibration and Aggregation Strategy ........................................................... 27
3.3 Import Tariff and Export Subsidy ............................................................................. 28
3.4 The Data .................................................................................................................... 29
4.
Simulation Scenarios and Experimental Design ............................................................. 29
5.
Simulation Results .......................................................................................................... 30
5.1 Effects of SAFTA ..................................................................................................... 30
5.2 Effects of SAFTA+5 ................................................................................................. 35
6.
Summary of Findings ...................................................................................................... 41
PART III
CONCLUSIONS AND POLICY IMPLICATIONS................................................ 43
Endnotes......................................................................................................................................... 45
References ...................................................................................................................................... 47
PART I
THE IMPACT OF TRADE AGREEMENTS ON INTRA-REGIONAL EXPORTS:
EMPIRICAL EVIDENCE FROM SAARC COUNTRIES
1.
Introduction
The issue of free trade and regional integration is becoming synonymous with trade
liberalization and a subject of avid interest in the arena of international trade and politics today.
It has been essentially taking the form of bilateral trade agreements (BTAs), preferential trading
arrangements (PTAs), regional trading arrangements (RTAs) and free trade agreements (FTAs).
The South Asian Association for Regional Cooperation (SAARC) was conceived in Dhaka,
Bangladesh on December 8, 1985 by late president Ziaur Rahman of Bangladesh with the broad
objectives of economic, social, cultural and scientific cooperation amongst seven South Asian
nations, viz., Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka.
South Asia is home to more than one-fifth of the world’s population, and therefore, it is
believed to have the potential to become an area of great prosperity based on the idea of
growing trade amongst the member countries. Increased trade, particularly intra-regional
exports, is expected to be a driving force for economic growth in the region. To this end, the
South Asian Preferential Trading Arrangement (SAPTA) was signed in Dhaka on April 11,
1993 to give boost to regional trade integration, which came into operation in 1995. Four
rounds of exchange of trade concessions have taken place under the SAPTA. A large number of
products have been offered concessions exclusively to Least Developed Countries (LDCs).
India has offered the largest number of concessions, particularly favoring LDCs with tariff
preferences ranging from 50-100 percent (Mukherji 2004).
Inspired by the worldwide trends and successful experience of India-Sri Lanka bilateral
FTA, leaders of SAARC countries have decided to facilitate intra-regional trade with the
ratification and formal launching of the South Asia Free Trade Area (SAFTA) on July 1, 2006.
Based on this, the SAARC members have agreed to bring down the average tariffs on goods
from 25-30 percent to 0-5 percent over the next decade. Separate deadlines have been set for
the developing countries, i.e., India, Pakistan and Sri Lanka, and the LDCs, which are Bhutan,
Bangladesh, Maldives and Nepal. SAFTA calls for reduction in import duties to 20 percent by
2006 and between 0-5 percent by 2013, but allows the LDCs to reduce the tariff rates to 0-5
percent by the year 2016 (SAFTA 2005).
―1―
SAARC countries share a lot of similarity in culture and socio-economic conditions, but as
opposed to the EU or ASEAN countries that have deep connections in political and economic
areas, members in the SAARC bloc are characteristically far less integrated. Many researchers
argue that the volume of intra-regional trade continues to be low, despite SAARC’s efforts to
enhance intra-regional trade through bilateral and preferential trading arrangements. In view of
the downsides, SAARC recognizes the need to move towards a more pragmatic approach to
regional cooperation. Over the years, the seven South Asian nations have signed several trade
agreements (see Table 1). However, there are very limited studies that have attempted to
evaluate the actual impact of trade agreements on intra-SAARC trade. Moreover, most of the
studies have omitted the effect on small countries, such as Bhutan and Maldives in their
analyses. This may be partly due to lack of data and smallness of the economic size, or perhaps
due to lack of researchers’ interest for its negligible influence in the region. Whatsoever the
reasons are, there is an apparent need to fill up this vacuum. Hence, examining the feasibility
and potential of South Asian trading arrangements is a highly desirable case study.
Table 1. Status of Trade Agreements amongst SAARC Countries
Date/Year
Contracting States
Agreement Type/Title
Jan 1972;
Renewed Mar 2, 1995
Mar 28, 1972;
Renewed Mar 21, 2006
Apr 2, 1976
India and Bhutan
Agreement on Trade, Commerce and Transit between the
Govt. of the Rep. of India and the Royal Govt. of Bhutan.
Trade Agreement between India and Bangladesh.
Apr 3, 1979
Nepal and Sri Lanka
1980;
Renewed Sep 2000
Bangladesh and
Bhutan
Mar 31, 1981
India and Maldives
Jul 28, 1982
Pakistan and Nepal
Dec 6, 1991
Nepal and India
Apr 11, 1993;
Operational Dec 7, ‘95
Dec 28, 1998
7 member states
Jun 12, 2005
Pakistan and Sri
Lanka
7 member states
India and
Bangladesh
Nepal and
Bangladesh
India and Sri Lanka
Jan 6, 2004;
Operational Jul 1, 2006
Source: Author’s compilation from various sources.
Trade and Payment Agreement between His Majesty’s
Govt. of Nepal and the Govt. of the People’s Rep. of
Bangladesh.
Trade Agreement between His Majesty’s Govt. of Nepal
and the Govt. of the Dem. Socialist Rep. of Sri Lanka.
Trade and Transit Agreement between the Govt. of the
People’s Rep. of Bangladesh and the Royal Govt. of
Bhutan.
Trade Agreement between the Govt. of the Rep. of India
and the Govt. of the Rep. of Maldives.
Trade Agreement between the Govt. of Islamic Rep. of
Pakistan and His Majesty’s Govt. of Nepal.
Free Trade Agreement between His Majesty’s Govt. of
Nepal and the Govt. of India.
South Asian Preferential Trading Arrangement (SAPTA).
Free Trade Agreement between the Rep. of India and the
Dem. Socialist Rep. of Sri Lanka.
Free Trade Agreement between the Govt. of Islamic Rep.
Pakistan and the Dem. Socialist Rep. of Sri Lanka.
South Asian Free Trade Area (SAFTA).
The objective of this study is, therefore, to investigate whether the trade agreements
amongst seven SAARC countries have actually boosted the volume of intra-regional exports.
The plain hypothesis is that trade agreements amongst SAARC members is expected to be
―2―
positively associated with exports, ceteris paribus, as the principal reason for member countries
to enter into trade agreements is the prospect of enhancing their exports and furthering trade
creation.1 In light of this, the paper specifically addresses the following key questions:
(1) What is the actual impact of trade agreements amongst SAARC countries? Have they
affected positively or negatively on the growth of exports, and to what extent?
(2) Did SAPTA play a catalytic role of enhancing the intra-SAARC trade?
One of the contributions of this paper is to answer the above questions and to provide
further evidence by estimating a Gravity Model, which uses a distinctive panel dataset for over
a long period spanning 35 years from 1971 to 2005. This period is preferred for three reasons.
First, there are very few studies conducted for the SAARC region that seeks to investigate the
trade effects over this long period of time. Second, Bangladesh became an independent country
in the year 1971.2 Third, this long period facilitates the dataset to be divided into sub-periods,
enabling to contrast and examine the impacts of trade agreements on exports in different time
periods. While addressing the above-mentioned questions, this study also tests the efficacy of
free trade agreements in furthering exports, and particularly for the SAARC region, this is the
first of its kind. The focus, as such, is to recount the findings after testing the effects of trade
agreements before the inception of SAARC and SAPTA, and comparing and contrasting with
the post-SAARC and post-SAPTA periods.
The remainder of the paper is organized as follows: Section 2 provides the theoretical
foundation and assessment of regional trade integration in South Asia. The empirical
methodology and sources of data is presented in Section 3 and Section 4, respectively. The
regression results are examined in Sections 5, while Section 6 summarizes the key findings of
the paper.
2.
Literature Review
Although there are some qualitative studies on SAARC and SAPTA, quantitative studies
centering on economic integration in South Asia are limited. Bandara and Yu (2003) point out a
number of possible reasons for this. Firstly, many trade analysts have not given much attention
since this region did not play a major role in the global trade, investment and growth. Secondly,
the data on trade and other variables in this region are scarce. Thirdly, as the volume of
informal trade is very high, published data do not reflect the true picture of trade structure in the
region. Finally, while there is much focus on non-tariff barriers compared to many other regions
in the world, the recognition and quantification of non-tariff barriers are difficult. Despite the
limitations, available studies may be broadly classified into three different views: pessimistic,
optimistic and neutral or cautionary.
―3―
From the pessimistic point of view, Hassan et al. (2002) finds that SAARC countries have
not only reduced trade amongst them but also reduced trade with non-members. Singh (2005: 1)
observes the intra-regional trade low at about only five percent of the total trade. The author
cites the lack of trade complementarities, prevalence of sizeable informal trade, and political
tensions as some of the “culprits”. Chowdhury (2004) examines the issue of convergence of per
capita GDP across SAARC countries. The results from his analysis fail to find evidence of
convergence. The reasons for non-convergence of per capita GDP is explained by low and
falling volume of intra-country trade, weak governance, and low level of growth of individual
SAARC countries. Jayaratne (2004) also points out that some issues that have prevented
effective regional integration of South Asian nations from more rapid development and
benefiting from cross border and global trade and investments are political disputes, macroeconomic instability, policy deficiencies, lack of a common position, and low implementation
capability amongst others. Pitigala (2005: 42) shows that “South Asian countries can be
characterized only moderately as ‘natural trading partners’”, and therefore the trade structures
amongst the South Asian countries may not facilitate a rapid increase in intra-regional trade.
In comparison with the rest of the world, the economic size of the SAARC region is small
in terms of both GDP and share in the world trade. Recent studies show the economic case even
for SAFTA as pretty weak. For instance, Baysan et al. (2006) argues that it is quite unlikely
that trade diversion would be dominant as a result of SAFTA, as it is reinforced by high levels
of protection in the form of restrictive sensitive lists and stringent rules of origin (ROO).
Similarly, Kalegama (2004) asserts that “… not much can be expected from SAFTA. The initial
euphoria that comes with the signing of the SAFTA agreement will soon taper away. The
realities and the geo-politics of the region will once again determine the pace of negotiations in
SAFTA.”3
Conversely, optimists including Bandara and McGillivray (1998) describe signs of
progress in liberalizing South Asian trade regimes in the 1990s, even if most of the programmes
in South Asian countries, with exception of Sri Lanka, have been slow. They find the recent
economic growth in the region quite satisfactory and are likely to continue in the near future.
Perhaps, it is for this reason that Bhargava (1998: 22) suggests the South Asians to learn from
the European experience. He believes that “the coming decades will witness meaningful
cooperation between the two largest configurations of democratic states in the continents of
Europe and Asia in order to build a better world.” The recent move by the EU to become an
observer in the SAARC group indeed forms the basis towards his line of thinking.
―4―
Mohanty (2003) emphasizes that the region has a substantial potential for trade and
investment. He rejects the hypothesis that South Asian countries compete amongst themselves
to export similar kind of products to the world market leading to very low level of regional
trade. His study finds a significant level of trade potential in the region to promote intraregional trade, and estimates the export potential to be more than six times than the present level
of intra-regional trade if it is harnessed completely. Mukherji (2004) reckons that bilateral trade
amongst member countries can be self-sustaining when backed by investment linkages. His
work demonstrates some modalities by which SAPTA could transform swiftly to SAFTA. He
stresses that the Trade Liberalization Programme that will be launched under SAFTA must take
a more consolidated approach by removing a variety of non-tariff barriers, and setting welldefined targets to promote a number of trade facilitation measures. Likewise, CUTS (2005: 3)
notes, “SAFTA would be a vehicle through which all participants can gain by exploring their
competitive advantages. Integration of economies in South Asia would lead to the emergence of
a big market for investors.” In more recent work, Rodríguez-Delgado (2007) estimated the
economic impacts of SAFTA using a gravity model covering the data from 1988-2004.
Studying the impact of tariff reductions on the GDP, he estimates that SAFTA can provide the
highest increase for SAARC countries in terms of trade flows that they could expect from any
RTAs.
Dash (1996) argues from more or less neutral perspective that, given the low level of
mutual trust, effects of ethnic and religious conflicts, and the extent of bilateral disputes in
South Asia, it is unrealistic to believe that any substantial growth in regional cooperation is
possible without easing political tensions. To evaluate the magnitude of preferential trade under
SAPTA, Mukherji (2000) estimated the extent of trade preference under all SAPTA rounds in
terms of trade values and percentages of preferential imports. The estimates show that the
region’s total preferential imports amounted to about US$479.8 million – nearly half of which
went to Pakistan. India’s share of preferential trade out of total regional preferential imports
was about 26 percent, while that of Sri Lanka was about 16 percent. In terms of its total
regional imports, he finds that Pakistan had the highest coverage of preferential imports (about
40 percent), followed by Nepal (35 percent), India (30 percent), Bhutan (17 percent) and Sri
Lanka (12 percent).
Using gravity equation and a panel data for 1996-2002, Hirantha (2004) showed strong
evidence of trade creation in the region with no trade diversion effect as far as trade with nonmembers is concerned. Pattanaik (2006: 140) has more of a cautionary approach to SAARC’s
future. He opines that if SAARC continues to remain “stymied” and the smaller states do not
―5―
actively integrate, even India’s hopes of integrating with other relevant groups will be stifled.
The smaller members that seek to gain from the opening up of a large and growing Indian
economy would also be the losers if SAARC does not prosper. Accordingly, Ghani and Din
(2006: 4) construe that an effective implementation of SAFTA and other regional initiatives
“will require a strong willingness of all members for greater economic integration as well as a
favorable political environment in the region.”
Kemal (2004) describes that the trade patterns of the SAARC group vary sharply from
country to country. For instance, the share of intra-regional imports in total imports in 2000 for
Bangladesh, Nepal and Sri Lanka stood at 11.7 percent, 33.2 percent and 10.1 percent
respectively. During the same year, Pakistan and India met only 2.3 percent and 0.7 percent
respectively of their import requirements from the region. Taneja (2006) points out that South
Asia is the least integrated region compared to the East Asia, Central Asia, Europe, Latin
America, the Middle East and North Africa. Intra-regional trade in South Asia is only 0.8
percent of GDP, one-eighth of the Latin America’s level, and only a fraction of East Asia’s
nearly 27 percent of GDP. However, India’s share in total SAARC trade increased from 38
percent in 1991 to 45 percent in 2004. Therefore, Taneja believes that if India and Pakistan
could tap the region’s trade potential, intra-SAARC trade could undoubtedly reach newer
heights.
3.
Methodology
This paper estimates a generalized Gravity Model or so-called Unilateral Exports Model, which
was applied in earlier works of Mátyás et al. (2000) and Aristotelous (2001), and more recently
by Baak (2004) and Billen et al. (2005). With slight modification from a typical gravity model,
the generalized gravity model assigns not the product of the exports of two trading countries as
in the paper by Dell’ Ariccia (1999), but the exports from one country to another as the
regressand. The advantage of this is that it allows including depreciation of exporting country’s
currency value as one of the regressors that affects the volume of exports. Moreover, as small
countries are expected to export less than big countries, ceteris paribus, the dummies for
exporting countries can be included (Baak 2004: 100). As initiated by Rose (2004: 99), the
gravity equation in this paper is further augmented by controlling for a number of “natural
causes of trade” or “extraneous factors” including in economic, cultural, political and
geographical variables that may affect trade.
Regressions are designed in such a way as to capture the effects of trade agreements with
respect to both Pre-SAARC and Post-SAARC periods. The sample is broken down into five
groups. Table 2 shows the design of regressions conducted in this study. Only five countries are
―6―
included in the Pre-SAARC I Period (1971-1979) owing to missing data for the period before
1980 for two countries, Bhutan and Maldives. All other periods include seven countries.4
Table 2. Regression Design
Regression
Group
Period (Year)
Countries Included
Pre-SAARC Period
1
Pre-SAARC I
(1971 – 1979)
2
Pre-SAARC II
(1980 – 1984)
5 countries: Bangladesh, India, Nepal, Pakistan, Sri
Lanka
7 countries: Bangladesh, Bhutan, India, Maldives,
Nepal, Pakistan, Sri Lanka
Post-SAARC Period
3
4
5
SAARC I
(or Pre-SAPTA)
SAARC II
(or Post-SAPTA)
SAARC I+II
(1985 – 1995)
(1996 – 2005)
(1985 – 2005)
7 countries: Bangladesh, Bhutan, India, Maldives,
Nepal, Pakistan, Sri Lanka
7 countries: Bangladesh, Bhutan, India, Maldives,
Nepal, Pakistan, Sri Lanka
7 countries: Bangladesh, Bhutan, India, Maldives,
Nepal, Pakistan, Sri Lanka
3.1 Model Specification
The gravity equation in this paper is different for two reasons. First, the model does not use the
total trade flows comprising exports and imports, but exports from one country to another as the
regressand. Intuitively, exports of one country are the imports of another country. When both
exports and imports are accounted for in trade flows, and if imports are registered much higher
than exports, the volume of trade may seemingly be inflated. Moreover, exports in a sense tend
to have economic characteristics associated in theory with welfare-enhancing net trade creation
effect. Second, this model does not include the exporting country’s GDP as one of the
regressors to avoid endogeneity problems, as exports form part of the exporting country’s GDP
(Billen et al. 2006).
Accordingly, the Gravity Model takes the following form:
ln( X ijt ) = β 0 + β1 ln(GDPjt ) + β 2 ln( POPN it ) + β 3 ln( DREX ijt )
+ β 4 ln( DISTij ) + β 5 BORDij + β 6 LANGij + β 7CURRijt
+ β 8TRAGijt + β 9CONFijt + β10 MEMBijt + β11 LLOCKij
(1)
+ β12 ILANDij + β13 PORTij + β14 SAPTAijt + δ 2 CD2 + ...
+ δ 7CD7 + φ TD + ε ijt
where i, j and t stands for exporting country, importing country and time, respectively; Xijt
denotes real exports from i to j at time t; GDPjt is the real gross domestic product of j at time t;
POPNit is the population of i at time t; DREXijt is the depreciation rate of the real bilateral
exchange rate of i with respect to j at time t; DISTij is the great circle distance between i and j;
BORDij is a dummy variable which is one if i and j share a common border, and zero otherwise;
―7―
LANGij is a dummy variable which is one if i and j share a common language, and zero
otherwise; CURRijt is a dummy variable which is one if i and j use common currency at time t,
and zero otherwise; TRAGijt is a dummy variable which is one for having trade agreement
between i and j at time t, or zero otherwise; CONFijt is the index value for conflict between i
and j at time t; MEMBijt is a dummy variable for membership in other regional trade group(s),
which is one if i and j are part of a common free trade area at time t, and zero otherwise;
LLOCKij is a dummy variable equal to one if a country is landlocked, and zero otherwise;
ILANDij is a dummy variable equal to one if a country is an island, and zero otherwise; PORTij
is a dummy variable equal to one if a country has access to sea ports, and zero otherwise;
SAPTAijt is a dummy variable for South Asian Preferential Trading Arrangement, which is one
if i and j are part of this agreement at time t, and zero otherwise; CD an TD denote country
dummy and time trend dummy respectively; β , δ and φ are vectors of nuisance coefficients;
and ε ijt is the error term or any other omitted influences.
More specifically, the paper estimates a pooled ordinary least squares (OLS) model, but
for robustness checks it also performs sensitivity analyses using country and time dummy
effects, first differencing, instrument variables (IV), fixed effects and random effects models.
The parameters of interest are β8 , β 9 and β14 , i.e., the coefficients for trade agreement (TRAG),
conflict (CONF), and South Asian Preferential Trading Arrangement (SAPTA), respectively.
Except for dummy variables, all other variables take on log values to narrow the range of
variable and to make estimates less sensitive to outlying or extreme observations on the
regressand and regressors. The novelty of this paper is that the effect of TRAG is measured visà-vis CONF that might likely offset for any frailty of results obtained by using only TRAG as
one of the regressors. This variable is selected particularly as a consequence of its strong
influence in determining the magnitude of trade in the SAARC bloc.
3.2 Correction for Endogeneity Bias and Heteroskedasticity
In order to mitigate the potential endogeneity bias resulting from a possible correlation between
the TRAG and unobserved characteristics, the paper follows corrective procedures as inspired
by earlier researchers. Baier and Bergstrand (2005) distinguish a standard problem in crosssection empirical work is the potential endogeneity of right-hand side (RHS) variables. If any of
the RHS variables in equation (1) is correlated with the gravity equation error term ε ijt , the
variable is considered econometrically endogenous and the OLS may yield biased and
inconsistent coefficient estimates. They argue that the potential endogeneity bias of RHS
variables may arise due to omitted variables, simultaneity, and measurement error. In fact,
―8―
omitted variables are often the major source of endogeneity bias in gravity equations.
Particularly with respect to bilateral trade arrangements, the unobserved heterogeneity in trade
flow determinants is associated with the decision of whether or not to form an FTA. For
instance, in their earlier work, Baier and Bergstrand (2004: 6-8) find strong empirical evidence
that pairs of countries that have FTAs are likely to share similar economic characteristics. That
is, the probability of two countries’ governments seeking to enter into FTA may be high if they
expect a large welfare gain from potential bilateral trade creation further deepening
liberalization beyond tariff barriers and other non-tariff barriers into “domestic regulations”.
This means that TRAGijt and the intensity of domestic regulations could be positively correlated,
but ε ijt and the intensity of domestic regulations could be negatively correlated. As a result,
TRAGijt and ε ijt can be negatively correlated, and the TRAGijt coefficient will tend to be
underestimated. Therefore, they underscore the importance of addressing the endogeneity.
Fortunately, for cross-section data, we can address the problem of omitted variables by
including IV; for panel data, fixed effects and first differencing can be effectively applied to
treat endogeneity biases (see Wooldridge 2003).
Mátyás (1997, 1998), and Harris and Mátyás (1998) also suggest a pooled time-series of
cross-sections or panel data in order to identify these biases and correctly specify the
econometric model. They advocate the panel data to increase degrees of freedom, to enable
identification of business cycle, and correctly account for exporting and importing country
effects. Such effects can be treated as constants and estimated by fixed effects model in which
one is able to identify separately the unobserved effects of those countries that have strong
propensities to export and import, once divergences in other factors such as GDP, population,
distance, etc. has already been accounted for. As recommended in Harris and Mátyás’ work,
this paper also additionally considers a random effects model, where the unobserved effects is
assumed to be uncorrelated with the explanatory variables in each time period. All estimations
employ heteroskedasticity consistent covariance matrix estimator derived by White (1980),
which provides correct estimates of the coefficient covariances in the presence of
heteroskedasticity of unknown form. Serial correlation is not an issue because the samples are
independent across time.
3.3 Computation of Variables
Following Baak (2004) and Billen et al. (2005), some of the key economic variables in this
model have been computed as follows:
―9―
3.3.1 Real Exports
The real exports (Xijt) from country i to country j is defined as
⎛ NX ijt
⎞
× 100 ⎟⎟
X ijt = ln⎜⎜
⎝ USDt
⎠
(2)
where NXijt is the annual nominal exports (in US dollars) from country i to country j in year t,
and USDt is the US GDP deflator5 in year t.
3.3.2 Real GDP
The real GDP of an importing country j (GDPjt) is defined as
⎛ NGDPjt
⎞
GDPjt = ln⎜⎜
× 100 ⎟⎟
⎝ USDt
⎠
(3)
where NGDPjt is the nominal GDP of country j measured by purchasing parity, and USDt is the
US GDP deflator in year t. This variable is a proxy for economic mass or size of the trading
country.
3.3.3
Depreciation of real bilateral exchange rate
The depreciation rate of an exporting country’s currency value (DREXijt) is determined as
DREX ijt = ln (REX ijt ) − ln (REX ijt −1 )
(4)
where REXijt is the real exchange rate, which is measured as
REX ijt = NEX ijt ×
CPI jt
CPI it
(5)
where NEXijt is the average nominal exchange rate between country i and country j in year t,
and CPIit and CPIjt denote consumer price index of country i and country j respectively in year t.
This variable stands as a proxy for prices.
4.
The Data
The data in this paper comes from a wide range of sources. Annual nominal exports data (in
million US$) from the year 1971 to 2005 have been compiled from the IMF’s Direction of
Trade Statistics (DOTS). Missing exports data have been supplemented from the UN Comtrade,
and UNCTAD Handbook of Statistics. Nominal GDP (in constant 2000 US$) and population
data were obtained from the World Bank’s World Development Indicators (WDI). The US
GDP deflator, nominal exchange rates, and consumer price indices have been gathered from the
IMF’s International Finance Statistics (IFS).
The data for distances between two countries were calculated using the Great Circle
Distance Between Capital Cities and Time and Date.com. The geographical distance is the
― 10 ―
theoretical air distance, i.e., the great circle distance. Unless the capital cities of two trading
countries are major hubs or trade centers, this paper considers the distance between the major
trade centers of two countries, on account of the fact that if the distance is measured exclusively
between the capital cities of two countries, it could probably underestimate or overestimate the
actual gravity factor between two trading partners.
The index for conflict variable was taken from Conflictbarometer 2005, Heidelberg
Institute for International Conflict Research, Department of Political Science, University of
Heidelberg. Other country-specific variables, such as border, language, currency, trade
agreements, membership in other regional blocs, landlocked and island and status and seaports
were obtained from the CIA’s The World Factbook, SAARC’s official homepage, and related
websites.
In addition, as evoked by Baier and Bergstrand (2005), the paper also introduces some
political variables as instruments (z variables hereinafter) that might have some correlation in
two governments’ decision to form an FTA, but may not have correlation with their exports.
Past studies such as by Jaggers and Gurr (1999), Mansfield et al. (2002), and Kaufmann et al.
(2003) reflect the fact that two countries are more inclined to form FTAs if their governments
are more democratic. Therefore, three governance indicators have been selected as z variables
from the World Bank’s Governance and Anti-Corruption, as F Test suggests that they are
jointly significant. These are: (i) Voice and Accountability, (ii) Rule of Law, and (iii) Control of
Corruption. The variables are measured in terms of percentile rank (0 to 100) – zero
representing as the lowest and 100 as the highest.
5.
Empirical Results
5.1 Pooled OLS Estimation (Benchmark)
Table 3 presents the estimation results for pooled OLS data. The following findings emerge
from the estimation. The estimated coefficients values are conventional and quite stable across
all sub-periods. For instance, the trade orientation indicators for gross domestic product (GDP),
population (POPN), depreciation rate of real bilateral exchange rate (DREX), distance (DIST),
border (BORD), currency (CURR), trade agreement (TRAG), conflict (CONF) and membership
in ORTG (MEMB) are found to have the expected signs and to register statistical significance.
While countries with seaports (PORT) have positive and significant coefficient, landlocked
(LLOCK) and island countries (ILAND) show negative and/or insignificant impacts on exports
in general, which are also as expected. The negative coefficient for MEMB is as expected and
supports the findings of Pitigala (2005: 42) since SAARC members are not so much
― 11 ―
characterized as “natural trading partners” for most SAARC members demonstrate a tendency
to trade outside the region.
Table 3. Gravity Equation Estimates for Pooled OLS Data Model (Benchmark)
Specification
Sub-period 1
Pre-SAARC I
Sub-period 2
Pre-SAARC II
Sub-period 3
Sub-period 4
Sub-period 5
SAARC I or
SAARC II or
SAARC I+II
Regressors
Pre-SAPTA
Post-SAPTA
(1971-1979)
(1980-1984)
(1985-1995)
(1996-2005)
(1985-2005)
2.23***
1.83***
1.84***
1.74***
ln(GDPjt)
2.51***
(24.48)
(25.10)
(11.86)
(29.28)
(4.68)
1.542***
1.77***
2.37***
1.43***
ln(POPNit)
2.91***
(11.99)
(15.53)
(8.84)
(15.06)
(4.57)
1.19***
0.45
0.65
1.47***
ln(DREXijt)
1.99***
(2.30)
(0.21)
(1.00)
(3.50)
(3.04)
-4.52***
-3.65***
-2.63***
-3.42***
ln(DISTij)
1.59*
(-10.89)
(-9.21)
(-2.73)
(-11.07)
(1.85)
0.37***
0.28*
0.61
0.41***
BORDij
0.77
(2.11)
(1.71)
(1.52)
(3.15)
(1.37)
-1.45
-2.13
-2.17*
-1.55*
LANGij
1.65
(-1.24)
(-1.57)
(-1.72)
(-1.85)
(1.57)
2.40***
2.99***
2.02***
CURRijt
2.14***
(6.89)
(4.40)
(8.61)
(8.99)
0.14
-0.99***
0.59***
-0.39***
TRAGijt
0.57***
(0.73)
(-2.01)
(5.11)
(-3.31)
(3.86)
-0.75**
-0.61***
-0.67**
-0.86***
-0.61**
CONFijt
(-2.10)
(-6.09)
(-2.45)
(-4.02)
(-1.99)
-1.19***
-0.88***
-0.84***
0.03
-0.81***
MEMBijt
(-7.05)
(-2.05)
(-5.71)
(0.04)
(-3.98)
-0.30
-3.01***
-0.38
0.62
-1.06***
LLOCKij
(-0.86)
(-3.72)
(-1.29)
(0.72)
(-2.64)
0.04
0.71
-0.05
0.46
0.14
ILAND ij
(0.14)
(1.05)
(-0.20)
(0.57)
(0.50)
3.10***
-1.88***
2.61***
2.02***
PORTij
(14.49)
(-3.76)
(14.56)
(8.59)
-20.55***
-24.27***
-18.59***
-59.19***
-20.57***
Constant
(-14.31)
(-5.56)
(-14.91)
(-4.80)
(-12.90)
462
210
882
180
420
Observations
0.81
0.67
0.77
0.50
0.80
R2
Notes: The regressand, ln(Xijt) is the log real exports. Numbers in parenthesis are t-statistics. *, ** and *** indicate
significance level at 10 percent, 5 percent and 1 percent, respectively. White Heteroskedasticity-Consistent
Standard Errors & Covariance used.
In contrast to our expectations, the coefficient for DIST is counterintuitive and statistically
significant at 10 percent level of significance in the sub-period 1. One principal reason for this
can be attributed to the fact that the closest neighbors, namely, India, Pakistan and Bangladesh
were hostile to each other during the period on account of the 1971 Bangladesh Liberation War
(Mukti Juddho) or what is commonly known as the Indo-Pakistan War of 1971. Hence, trade
plummeted sharply for these countries during the period. The only two countries that were
engaged in formal trade were Nepal and Sri Lanka, which are the most distant countries in the
region. Therefore, in this case, the result cannot be interpreted in a causal fashion.
― 12 ―
The coefficients for LLOCK and PORT are also more or less consistent to expectations.
However, the negative coefficient for PORT in the sub-period 2 could appear as result of
repercussions of the war amongst three major countries in the earlier sub-period 1, as noted
above. One result that attracts attention is the negative and statistically significant coefficient
for language (LANG) for the sub-period 2 through 5, while our conventional wisdom tells us
that it should have a positive coefficient. The reason goes back to none other than the abovementioned interpretation, wherein most of the trading partners in the region with similar
languages have been exhibiting animosity against each other due to history of war and
contentions, thereby resulting in less trade and nullifying the expected positive impact. Another
parameter of interest is the coefficient for CONF, which is negatively associated with exports at
1 percent significance level against a two-sided alternative across all periods. In sub-period 1,
the coefficient for CONF is -0.86, suggesting that the presence of conflict between two trading
partners decreases exports by about 58 percent ( e −0.86 ).6 Similarly, in sub-period 2, sub-period 3,
sub-period 4 and sub-period 5, the negative impact of CONF is reflected by the decrease in
exports between the trading partners by about 46 percent, 53 percent, 46 percent and 49 percent,
respectively. Given the scenario of hostility and incessant discord amongst SAARC members as
already discussed above, this result is not surprising.
Of special interest in this regression result is the coefficient for TRAG, which is negative
and statistically significant at 1 percent level of significance for the sub-period 1 and sub-period
2, insignificant in the sub-period 3, and then positive and highly significant again in the subperiod 4 and sub-period 5. It is quite evident that before SAARC came into existence, intraregional trade was much lower amongst the South Asian nations. Even after the inception of
SAARC in 1985, the impact is not significant during the sub-period 3. However, the impact of
TRAG can be clearly observed in the sub-period 4 and sub-period 5, i.e., after the SAPTA came
into operation in 1995. For instance, in the case of sub-period 2, which is the period just before
SAARC came into force, the coefficient of TRAG is -0.99, and the coefficient in sub-period 4,
i.e., the period after SAPTA came into operation is 0.57. This implies that even those countries
that did have trade agreements had about 63 percent less exports in the sub-period 2. However,
in the sub-period 4, exports increased by about 77 percent. There is a further increase in exports
in the sub-period 5 by about 80 percent. These results indicate that the impact of trade
agreements is time-dependent or period-specific, which is largely consistent with the findings
by Baier and Bergstrand (2005). Moreover, this strongly supports our hypothesis and the case
for deeper regional trade integration in South Asia.
― 13 ―
5.2 Robustness Checks
Thus far, we have observed strong positive impact of trade agreements on exports, but are the
findings robust? To check the robustness of the benchmark results, some sensitivity analyses
are performed using country and time dummies, first differencing, IV technique, and fixed as
well as random effects models.7
5.2.1 Country Dummy Effects
With the introduction of country dummies, both GDP and POPN are still strongly associated
with exports. The impact of DREX is found to be largely insignificant. This may be because
almost all SAARC countries follow a fixed exchange rate system and so the depreciation rates
amongst these countries are negligible. Following analogous trend in Table 3, the coefficient
for DIST is found to be positive in the sub-period 1, but negative and significant for the most
part, which is typically as expected. The coefficient for BORD turns out to be insignificant and
even negative in the sub-period 1 with the inclusion of country dummies. The estranged
relationship between two major countries in the region, India and Pakistan, could yet again
explain this phenomenon. Besides, although Bangladesh, Bhutan, and Nepal also share a
common border, the gravity effect of these smaller countries have so little impact relative to
huge economies of India and Pakistan.
The coefficients for LANG, CURR, TRAG and CONF show very similar pattern as in the
benchmark results. The interpretations for this are not so different from what has been
deliberated earlier. Nevertheless, a careful scrutiny demonstrates that the positive impact of
TRAG has slightly weakened in the sub-periods 4 and 5. On the other hand, CONF has a further
negative impact on exports, as the exports during the sub-period 4 and sub-period 5 decreased
by almost 59 percent and 51 percent, respectively. Except for the sub-period 2, PORT has a
significant impact on exports in the sub-period 3, sub-period 4 and sub-period 5.
Interestingly, economically larger countries, especially India and Pakistan did not fare well
in sub-period 1, displaying clearly the backlash of the war during the period. Nevertheless,
beginning from the sub-period 3, i.e., soon after SAARC came into being, trade volume of these
three countries picked up momentum with positive and significant impacts on the regressand.
As characteristic to gravity effects, small economies like Bhutan, Maldives and Nepal are losers
in the game, but it is interesting to note that during the sub-period 1 when all major players
were at conflict, only Nepal and Sri Lanka fared well as they continued to have better terms of
trade. During the same period, DIST though positive, is not statistically different from zero,
which further justifies this observation.
― 14 ―
5.2.2 Time Dummy Effects
Adding time dummies has little material effect on the estimated results. However, a few points
worth mentioning are the insignificant impacts of DREX and BORD, and the inclusion of lag for
TRAG has a stronger positive impact especially in the sub-period 5. Inclusion of one lag for
TRAG has an increased impact on coefficient from 0.47 to 0.49. With two lags, the coefficient
value leaped to 0.50. 8 With regard to coefficients of the CONF variable, the trend is very
similar with significantly negative impacts on exports for all sub-periods.
5.2.3 Country and Time Dummy Effects
There is not much variation in the results even after controlling for both country and time
dummy effects. In general, we observe that the TRAG and CONF effects retain similar trend.
However, the intensity of TRAG weakens in comparison to the earlier results.
5.2.4 First Differencing
First differencing is particularly useful when the unobserved factors that change over time are
serially correlated. If ε ijt follows a random walk, meaning that there is very substantial positive
serial correlation, then the difference Δε ijt is serially uncorrelated, and therefore, first
differencing is a good alternative to solve this problem. 9 Focusing again on the TRAG and
CONF variables, the coefficients retain the expected signs as before. TRAG has a highly
significant positive impact on exports in the post-SAARC periods, i.e., in the sub-period 4 and
sub-period 5. The estimates suggest that TRAG increases partner countries’ exports by about
127 percent in the sub-period 4 and about 132 percent in the sub-period 5. CONF, on the other
hand, has a negative impact in all periods, and the coefficients are statistically significant from
the sub-period 3 through sub-period 6, but the impacts are lesser as compared to the earlier OLS
estimates.
5.2.5 IV and 2SLS Estimation
A set of three z variables are used as IV that is likely to influence the formation of FTA and less
likely to be correlated to the error term, ε ijt . The method of IV can be used to solve the problem
of endogeneity of one or more explanatory variables. This method applies two staged least
squares (2SLS or TSLS), which is second in popularity next to OLS for estimating linear
equations in applied econometrics. Six different scenarios were tested using IV technique and
2SLS (see Table 4).
― 15 ―
Table 4. Gravity Equation Estimates using IV
Specification
(1)
(2)
(3)
(4)
(5)
(6)
With no
With
With Time
With
With First
With First
Country
Country
Dummies
Country and Differencing Differencing
Regressors
and Time
Dummies
Time
and Time
and no Time
Dummies
Dummies
Dummies
Dummies
2.37***
2.28***
1.89***
2.37***
1.90***
ln(GDPjt)
2.23***
(23.49)
(23.62)
(6.10)
(23.55)
(6.17)
(24.48)
-10.25***
1.64***
-0.34
-8.77***
-0.32
ln(POPNit)
1.54***
(-2.04)
(11.73)
(-0.51)
(-3.55)
(-0.47)
(11.99)
0.27
-1.64
6.10***
-0.07
5.78***
ln(DREXijt)
7.19***
(0.06)
(-0.41)
(2.64)
(-0.017
(2.46)
(2.30)
-4.79***
-4.59***
-4.67***
-4.79***
-4.70***
ln(DISTij)
-4.52***
(-11.65)
(-11.20)
(-2.93)
(-11.67)
(-2.95)
(-10.89)
-0.06
0.24
1.56***
-0.07
1.55***
0.37***
BORDij
(-0.20)
(0.81)
(2.23)
(-0.21)
(2.21)
(2.11)
-1.44***
-1.59***
-1.33***
-1.61***
-1.45***
LANGij
-1.33***
(-5.65)
(-2.44)
(-5.25)
(-2.46)
(-5.24)
(-5.25)
2.26***
0.51
2.23***
0.51
2.14***
2.22***
CURRijt
(5.21)
(0.54)
(5.04)
(0.54)
(8.99)
(5.03)
0.50***
0.81*
0.34*
0.81*
0.57***
0.34*
TRAGijt
(2.73)
(1.76)
(1.78)
(1.80)
(3.86)
(1.79)
-0.83***
-0.68***
-0.90***
-0.69***
-0.81***
-0.90***
CONFijt
(-7.67)
(-3.06)
(-8.24)
(-3.08)
(-10.66)
(-8.23)
-0.82***
1.81
-1.32***
1.78
-0.811***
-1.37***
MEMBijt
(-3.86)
(1.37)
(-4.92)
(1.34)
(-3.98)
(-4.45)
-0.94
-3.409***
-8.99***
-3.43***
-1.06***
-10.10***
LLOCKij
(-1.64)
(-3.16)
(-4.52)
(-3.23)
(-2.64)
(-2.63)
0.27
-4.06***
-27.46***
-4.04***
0.14
-31.40***
ILAND ij
(0.85)
(-2.53)
(-4.11)
(-2.53)
(0.50)
(-2.33)
2.06***
1.88
1.84***
1.86
2.02***
1.84***
PORTij
(4.16)
(1.46)
(3.57)
(1.46)
(8.59)
(3.56)
19.08***
0.01
63.22***
-23.89
-20.57***
36.39
Constant
(3.40)
(0.22)
(3.13)
(-0.63)
(-12.90)
(0.44)
420
420
420
420
420
420
Observations
0.81
0.54
0.81
0.54
0.80
0.81
2SLS R2
Notes: The regressand, ln(Xijt) is the log real exports. Numbers in parenthesis are t-statistics. *, ** and *** indicate
significance level at 10 percent, 5 percent and 1 percent, respectively. White Heteroskedasticity-Consistent
Standard Errors & Covariance used. Instrument variables used are (i) Voice and Accountability, (ii) Rule of Law,
and (iii) Control of Corruption.
The results obtained by using z variables do not deviate much from the earlier results. The
impact of TRAG is still positive and statistically significant, while the impact of CONF is
negative and statistically significant. Nevertheless, in specifications 5 and 6, with the first
differencing TRAG has a significant impact on exports yielding an increase in exports by about
125 percent each, respectively. Hausman Test is applied to compare the OLS and 2SLS
estimates and to determine whether the differences are statistically significant. This procedure
tests the null hypothesis (H0) that the error term ε ijt of the OLS and the error term of the 2SLS
(say, v2 ) are not correlated. The test fails to reject the H0 concluding the exogeneity of z
variables because ε ijt and v2 are not correlated. In addition, the testing of overidentifying
― 16 ―
restrictions used in the 2SLS suggests that the model is just identified. Thus, the impact of
TRAG in the last two specifications is consistent and reliable. Similar results from the first
differencing method further justify the robustness of the estimates.
5.2.6 Fixed and Random Effects Models
F Test was conducted to test whether or not the fixed effects coefficients are equal by
comparing the sum of squared residuals (SSR) from the fixed effects model and the random
effects model. The computed p-value of zero soundly rejects the H0 of equal intercepts. In
addition, p-values of the Hausman Test are essentially zero for almost all periods, and so the H0
of the random effects model in favor of the fixed effects model is rejected. As in the pooled
OLS model, one can also observe that almost all the coefficients of GDP and POPN for both
the fixed effects model and random effects model are statistically significant and positive, while
the coefficient for the DIST is statistically significant with a conventional negative coefficient.
The coefficient for DREX is largely insignificant in the case of both fixed and random effects.
5.3 Effect of SAPTA
Researchers more often than not have raised debates that SAPTA has not been the main vehicle
for enhancing intra-SAARC trade. This contentious argument necessitates further empirical
testing. As we have seen in the earlier tests that the impact of TRAG is mostly seen in the subperiod 4 and sub-period 5, i.e., the period after SAPTA came into operation. This gives a high
possibility for us to conclude that the major impact on exports could have arisen because of the
SAPTA, and at the same time, undermine the role of TRAG as such. In order to unravel this
paradox, three more tests were carried out using SAPTA as one of the dummy regressors (see
Table 5).
Interestingly, the coefficient for SAPTA, despite being positive does not show significant
impact at conventional significance levels even in the post-SAARC periods, and as we move
backwards in time, SAPTA is rather negatively associated with exports especially for those
countries that do not have bilateral trade agreements. On the other hand, it is fascinating to
observe that TRAG has a significant impact during the later two periods from 1985-2005 and
from 1980-2005. Although the impact of TRAG also diminishes as we move backwards towards
the earlier period, the impact is not statistically significant. This could simply mean that the
impact of TRAG increased over time during the later sub-periods not specifically because of the
inception of the SAPTA. Stated otherwise, SAPTA has not been the main vehicle for increasing
the impact of TRAG on exports in the later periods. Therefore, the increased intra-regional
exports in the post-SAARC periods could have apparently stemmed from the delayed impact of
― 17 ―
the existing bilateral trade agreements amongst SAARC countries, besides the catalytic effect
of SAPTA to some extent. This observation is further elaborated in the subsequent section.
Table 5. Effects of SAPTA and TRAG
Impact of SAPTA
Impact of TRAG
Countries with
Trade
Agreements
(1985-2005)
Countries
without Trade
Agreements
(1985-2005)
Countries with
and without
Agreements
(1985-2005)
Not significant,
positive
coefficient.
Significant at
1% significance
−0.84
Not Significant,
negative
coefficient.
(1980-2005)
−0.79
(1980-2005)
Not Significant,
but negative
coefficient.
(1971-2005)
Significant at
1% significance
(1971-2005)
(1980-2005)
Not significant,
positive
coefficient.
(1971-2005)
Significant at
1% significance
−0.35
level, e
=
-56.83%.
(1980-2005)
Significant at
1% significance
level, e
=
-54.62%.
(1971-2005)
Significant at
1% significance
−0.95
Countries
with Trade
Agreements
(1985-2005)
Countries
without Trade
Agreements
(1985-2005)
Countries with
and without
Agreements
(1985-2005)
Significant at
1% significance
0.56
(1980-2005)
level, e =
75.07%.
(1980-2005)
Significant at
1% significance
0.41
level, e =
50.68%.
(1971-2005)
(1971-2005)
Significant at
1% significance
−0.25
−0.56
level, e
=
level, e
=
level, e
=
level, e
=
-29.53%.
-61.33%.
-22.12%.
-42.88%.
Notes: Intercepts and coefficients for all standard covariates not reported for brevity and ease of presentation.
TRAG/SAPTA variable for specific periods are omitted to avoid collinearity.
5.4 Lagged Effects of TRAG
Table 6 shows the comparative lagged effects of TRAG. Adjusting for unobserved
heterogeneity using country and time dummy effects provide impressive results.
Table 6. Comparative Lagged Effects of TRAG
Specification
(1)
(2)
(3)
With no Country
With Country
With Time
and Time Dummies
Dummies
Dummies
(1985-2005)
(1985-2005)
(1985-2005)
%
No. of
%
No. of
%
No. of
Year
Increase
Times
Increase
Times
Increase
Times
Increase
Increase
Increase
1986
4.73
1.05
6.19
1.06
6.79
1.07
1987
6.07
1.06
9.46
1.09
10.00
1.10
1990
15.75
1.16
22.86
1.23
27.27
1.27
1995
40.99
1.41
55.39
1.55
65.54
1.66
2000
36.67
1.37
57.66
1.58
52.45
1.52
2004
81.85
1.82
118.72
2.19
121.37
2.21
Note: The percentage increase is measured from the base year 1985.
(4)
With Country and
Time Dummies
(1985-2005)
%
No. of
Increase
Times
Increase
6.21
1.80
9.51
1.10
23.68
1.24
57.10
1.57
58.59
1.59
127.84
2.28
For instance, in specification (1), the impact of TRAG after 15 years (i.e., in 2000) from
the inception of SAARC in 1985 is about 37 percent increase, an increase by 1.37 times. During
the same year, the impact of TRAG is nearly 58 percent, 52 percent and 59 percent in
― 18 ―
specification (2), (3) and (4), respectively. However, in 2004 (after 19 years), there is a
dramatic increase in the TRAG effects from about 82 percent in specification (1) to almost 119
percent in specification (2), 121 percent in specification (3) and 128 percent in specification (4).
That is, the impact of TRAG is more than twofold using country and time dummy effects. These
results are evocative and consistent with the findings of Baier and Bergstrand (2005), whose
estimates suggest that an FTA on average doubles two member countries’ bilateral trade after
10 years.10
It may be noted that the overall percentage increase of exports after 19 years from the base
year 1985 to 2004 is 885 percent (see Table 7). If we consider the cumulative average impact of
TRAG during this time period in specifications (1), (2), (3) and (4) in 2004 as 113 percent, then
the independent role of the TRAG amongst different variables in enhancing intra-regional
exports works out to be approximately 13 percent. Understandably, it took nearly two decades
for TRAG to show reasonably clear impacts, which illustrates the sluggish nature of SAARC’s
progress in trade integration.
Table 7. Intra-SAARC Exports (1985-2004)
Year
US$ Million
Elapsed no. of Years
0
600.83
1985
1
553.61
1986
2
614.76
1987
3
786.67
1988
4
862.25
1989
5
862.96
1990
6
1013.15
1991
7
1238.75
1992
8
1191.47
1993
9
1433.54
1994
10
2023.65
1995
11
2144.45
1996
12
2173.94
1997
13
2466.26
1998
14
2180.00
1999
15
2593.37
2000
16
2826.68
2001
17
2997.97
2002
18
4773.32
2003
19
5919.36
2004
Note: The percentage increase is measured from the base year 1985.
Source: Author’s calculation using data from the UNCTAD Handbook of Statistics 2005.
6.
% Increase
0.0
-7.86
2.32
30.93
43.51
43.63
68.63
106.17
98.31
138.59
236.81
256.92
261.82
310.48
262.83
331.63
370.47
398.97
694.46
885.20
Summary of Findings
Part I of this study has estimated a generalized form of Gravity Model to determine the impact
of trade agreements on exports in the SAARC region. The model performed well empirically
yielding reasonably precise and good estimates, which are largely consistent with results of the
earlier studies employing a gravity model and pooled trade data.
― 19 ―
The fundamental question that arose from the results was whether trade agreements have
had a significant positive impact on the volume of intra-regional exports of SAARC countries.
The answer is yes, but one should interpret with caution. This is because the empirical tests
have found scarce evidence of the impact of trade agreements on exports in pre-SAARC I (subperiod 1 from 1971 to 1979), pre-SAARC II (sub-period 2 from 1980 to 1984), and SAARC I
(sub-period 3 from 1985 to 1995). However, a significant positive impact of trade agreements
on exports is observed in SAARC II or post-SAPTA period (sub-period 4 from 1996 to 2005)
and for SAARC I+II (sub-period 5 from 1985 to 2005), even amidst sustained significant
negative impact of conflict in all sub-periods. The phenomenon is observable irrespective of the
estimation techniques applied. Thus, the results soundly support our hypothesis indicating
positive benefits of having trade agreements amongst SAARC countries.
The next important question was to ascertain whether the signing of the SAPTA has
stimulated intra-SAARC trade in the region. Empirical tests find little evidence of the impact of
SAPTA signaling very modest role played by SAPTA in inducing trade creation within the
region. This is because the test results show that the positive impact appears to have emanated
not exclusively from the SAPTA per se, but as a coalesced effect arising from the delayed
impact of the existing trade agreements amongst SAARC countries.
Based on the results obtained from this model, the next pertinent question that follows is:
What would be the economic impact or welfare implications for SAARC countries consequent
to ratification of SAFTA, and also by extending the notion of FTAs for inter-regional trade with
other interested observer countries? However, the Gravity Model cannot efficaciously estimate
the welfare effects of FTAs, or more expressly, the trade creation and trade diversion effects. In
order to tackle the latter set of three research questions posed in this study, the most appropriate
tool is found to be the global AGE/CGE model, such as the GTAP Model. In view of this, we
employ the GTAP Model in Part II to address the welfare implications of SAFTA as well as
trade integration with five observer countries to explore the relevance and economic potential
for pan-Asian-European economic cooperation. This is also done with a view to throwing light
on any possibilities for SAARC nations in having FTAs with the observers and thereby
extending and/or expounding new perspectives for future memberships.
― 20 ―
PART II
ASSESSING THE ECONOMIC IMPACTS AND WELFARE IMPLICATIONS OF
SAFTA AND SAFTA+5: THE SOUTH ASIAN EXPERIENCE
1.
Introduction
Free trade and regional economic integration is increasingly becoming a pervasive trend in the
current era of globalization and multilateral trading system. Emulating the regional trade blocs
in Europe and the Americas, the movement towards bilateralism and regionalism has been
gaining momentum in South Asia and East Asia during the last decade with the unfolding of
numerous trade agreements signed one after another based upon reciprocity (ADB 2006;
Harrigan et al. 2006). This is perhaps the reason why Asami (2005: 7) reckons that regional
integration is “inevitable as globalization becomes the order of the day.” In fact, the headlong
rush for bilateralism, regionalism, and free trade during the last decade has ushered a new era in
the global trading system. Regional trading agreements (RTAs), for all intents and purposes,
have become one of the major international developments in recent times, which commonly
take the form of bilateral trade agreements (BTAs), preferential trading arrangements (PTAs),
free trade agreements (FTAs), customs unions, common market, economic union, or such
agreements leading to one or the other. BTAs, PTAs and FTAs in particular are assuming a
prominent role in economic integration in the developing region of South Asia.
South Asian Association for Regional Cooperation (SAARC) was established in 1985
when the seven nations of South Asia comprising Bangladesh, Bhutan, India, Maldives, Nepal,
Pakistan and Sri Lanka teamed up for a common purpose of reducing poverty, strengthening
regional cooperation and accelerating economic growth in the region. A first step towards
fulfilling this aspiration transpired when South Asian Preferential Trading Arrangement
(SAPTA) became operational in 1995. Subsequently, on July 1, 2006, South Asian Free Trade
Area (SAFTA) became operational creating a framework for establishment of a free trade area
covering almost 1.5 billion people. Under the Agreement, member states of the SAARC bloc
have concurred to bring their tariffs down to 0-5 percent by 2016. Furthermore, five observer
countries, particularly China, Japan, South Korea, the United States, and the European Union
(EU) have lately been showing keen interest in associating with this region.11 Japan and the EU
have already started negotiations for bilateral FTA with India, while China is in the queue.
These developing issues provide incentive for us to ask some fundamental questions such as:
(1) What are the economic effects of SAFTA on trade flows as a result of the reduction in
― 21 ―
tariffs given the present economic structures of SAARC countries and varied levels of
development? (2) What will be the welfare implications of FTA amongst the SAFTA members
and the five observers that integrate the North and the South? and (3) Which of the contracting
parties are likely to have potential welfare benefits and most feasible FTAs?
The motivation for this study emerges from the fact that free trade and regional integration
is one of the most important building blocks for economic growth in the region. The main
objective of this paper is therefore to evaluate the economic impacts and welfare implications of
SAFTA amongst the member states as well as FTAs of SAARC nations with three East Asian
giants, i.e., China, Japan and South Korea, and the two big players of the West, the United
States and the EU (hereinafter referred to as “+5”). The first hypothesis we test is whether the
compensation by means of preferential tariff concessions from the winners to losers will ensure
all countries to gain from FTAs. In other words, selective combinations of tariff rates are
expected to result in welfare gains of both the contracting parties. The second hypothesis to be
tested is whether or not SAFTA and FTAs with the aforementioned +5 countries will be welfare
improving to SAARC member countries because the positive effects from trade creation are
expected to be larger than the negative effects from trade diversion. In order to do so, we
employ the Global Trade Analysis Project (GTAP) Model – a multiregion, multisector applied
general equilibrium (AGE) 12 model based on perfect competition and constant returns to scale.
The remainder of the paper is organized as follows: Section 2 provides the theoretical
considerations on free trade encompassing different forms of trade integration tools, and the
momentum for broader economic agenda in South Asia. The model calibration, aggregation
strategy and the data are discussed in Section 3. Section 4 presents the simulation scenarios and
welfare experimentation design, while Section 5 evaluates the results from several bilateral and
plurilateral FTA options for SAARC countries with +5 countries. Section 6 details the summary
of important findings.
2.
Some Issues on Free Trade and Welfare
2.1 Bilateralism or Plurilateralism?
Bilateralism comprises the political, economic and cultural relations between two states, while
regionalism constitutes more than two states that express a particular identity and shape
collective action within a geographical region. Plurilateralism, on the other hand is in “between
bilateralism and multilateralism, and indicates a policy of three or more countries concluding a
regional economic agreement, and promoting trade liberalization” (Oyane 2001: 9). Plurilateral
agreements are the contractual agreements that are made in between the states and/or blocs of
diverse geographic regions. Plurilateralism provides the possibility of enabling relatively simple
― 22 ―
negotiations between multiple countries with common interests, and expanding in a domino
effect the resultant liberalization (U.S. Council of Economic Advisers 1995). Amongst the
many trade agreements in the world, plurilateral agreements are one of the most important
developments witnessing some of the historical moments in international trade. Without
restricting to any particular region of the world, plurilateral agreements have made their mark
all over the world. Two of the major agreements comprise Middle East Free Trade Area (USMEFTA) and Euro-Mediterranean free trade area (EU-MEFTA). Multilateralism, on the other
hand, is a term in international relations that refers to a large number of countries working in
concert on a given issue. Good examples are the United Nations (UN) and the World Trade
Organization (WTO).
The fad for free trade and economic integration is in effect questioning the virtues of
bilateral versus multilateral trading system. Proponents have their own set of arguments for
favoring their respective positions. Raihan and Razzaque (2007: 17) argue that bilateralism is
trade-creating because countries can “lock-in” reforms via bilateral FTAs or RTAs,13 which is
often politically not executable under multilateralism. Moreover, while multilateral trade talks
are much more complex, trade liberalization can take place more easily through bilateral talks,
as bilateral agreements have greater flexibility and ease that is lacking in most compromisedependent multilateral systems. On the flip side, Raihan and Razzaque (2007) as well point out
that bilateral FTAs undermine the spirit of multilateralism. They affirm that there is a
possibility of discrimination against the excluded countries, and too much involvement in
bilateral negotiations may distract attention from multilateral liberalization, and as such, the
world might be divided into a few protectionist blocs, further strengthening the opposition to
multilateral liberalization. Khor (2006) argues that multilateralism tends to have a systematic
bias toward rich countries and multinational corporations, harming smaller countries which
have less negotiating power. Furthermore, the “spaghetti bowl” phenomenon, as propounded by
Bhagwati (2005: 28), can emerge because of the traversing of simultaneous bilateral trade
negotiations.
Nonetheless, Burfisher and Zahniser (2003) maintain that a country need not necessarily
follow a stringent single policy towards liberalization in a fundamentally globalized world.
Dual trade reforms involving bilateral and plurilateral trading arrangements form the best
possible options for taking full advanatage of liberalized economies. Multilateralism is clearly
beneficial in that it engages virtually every country in the world in a mutual process of trade
reform. In contrast, while the bilateral and plurilateral are exclusive and discriminatory, they are
― 23 ―
capable of much deeper trade reforms since their adherents are fewer, more like-minded and
committed, and often linked geographically and historically.
2.2 Free Trade and National Welfare
Hudgins (1996: 231) contends that all forms of trade liberalization are “valid means” to opening
world markets. Each of the channels has a specific role for free trade and they should not be
discarded without a good reason. Low (2004: 2) asserts that “free trade remains the first best
trade policy.” Indeed, everyone stands to gain from free trade, either through the mechanics
involving economies of scale, or the offering of more opportunities for learning and innovation
(Caves et al. 2007; Krugman and Obstfeld 2003). Even critics concede, in general, that freer
trade through bilateral or regional trade liberalization improves the welfare of countries by
promoting wealth creation. However, some conditions, opening markets with only selected
trading partners could become a conduit for trade diversion (Weintraub 1996).
Brown et al. (2003) point to the fact that separate bilateral FTAs have positive, but
generally small welfare effects on the partner countries, and potentially disruptive sectoral
employment shifts in some countries. They argue that regional agreements such as APEC,
ASEAN+3, and a Western Hemisphere FTA would increase global and member country
welfare, but much less so than the multilateral trading organizations, such as the WTO. While
they also detect evidence of trade diversion and detrimental welfare effects on some nonmember countries in the case of PTAs, the welfare gains from multilateral trade liberalization
are found to be considerably greater and uniformly positive for all countries.
A global scale multilateral trade framework may have an advantage in terms of resource
allocation, economic welfare and economic prosperity in theory; however, the next best
framework has always been the bilateral, or plurilateral agreements, which enables lowering of
trade barriers amongst members without having to lower barriers for the non-members.
Bilateral FTAs and plurilateral RTAs also prevail over multilateral framework like the WTO in
terms of dealing with difficult trade problems, as the WTO normally must cater to the lowest
common policy denominator (Hudgins 1996). All in all, there is a general consensus amongst
trade analysts on the existence of a similar relationship between these arrangements. In so far as
bilateral, plurilateral and multilateral trade liberalizations are concerned, they are all a
complementary means to opening world markets, which in turn contributes to ultimately
achieving the goal of greater national welfare and economic liberty (Doshi 2008).
2.3 SAFTA and Broader Economic Agenda
The worldwide proliferation of PTAs and successful implementation of India’s bilateral FTA
with Nepal, Bhutan and Sri Lanka was somehow the precursor that led to the signing of the
― 24 ―
SAFTA on January 6, 2004 (see Baysan et al. 2006; Mohanty 2003), which eventually became
operational since July 1, 2006. The SAFTA framework covers tariff reductions, rules of origin
(ROO), safeguards, institutional structures, and dispute settlement. It also calls for the adoption
of various trade facilitation measures, such as harmonization of standards and mutual
recognition of test results, harmonization of customs procedures, and cooperation in improving
transport infrastructure. These measures are expected to help significantly reduce the cost of
international trade, especially regional trade. The SAFTA trade liberalization process will take
10 years to complete. However, this extended timeline of the SAFTA Agreement is viewed by
some analysts to weaken SAFTA’s impact if other trading arrangements supersede it (Batra
2005).
The tariff reduction by non-LDCs (India, Pakistan, and Sri Lanka) to LDCs (Bangladesh,
Bhutan, Nepal, and Maldives) would be completed in two phases: In Phase I (1/1/2006 –
31/12/2007), the existing tariff rates above 20 percent are to be reduced to 20 percent within
two years, and tariff below 20 percent is to be reduced on a margin of preference basis of 10
percent on actual tariff rates for each of the two years. Phase II (1/1/2008 – 31/12/2012)
requires tariffs to be reduced to 0-5 percent within 5 years (Sri Lanka is given six years). The
tariff reduction by LDCs would also be completed in two phases as well. In Phase I (1/1/2006 –
31/12/2007), the existing tariff rates above 30 percent will be reduced to 30 percent within two
years and tariff below 30 percent to be reduced on margin of preference basis of 5 percent for
each of the two years. In Phase II (1/1/2008 – 31/12/2015), tariffs will be reduced to 0-5 percent
in equal installments, but not less than 10 percent annually (see SAFTA 2004).
Very few studies have hitherto attempted to examine the welfare effects of SAFTA. These
studies have demonstrated mixed results. For instance, Baysan et al. (2006) surmise the
economic case for SAFTA to be rather weak on account of small economic size of SAARC
countries vis-à-vis rest of the world, prevalence of high levels of tariff and para-tariff
protections, sectoral exclusions and stringent ROO. In contrast, Rodríguez-Delgado (2007)
shows that SAFTA can provide the highest increase for SAARC countries in terms of trade
flows they could expect from any RTAs. Others like Raihan and Razzaque (2007) conclude that
a full implementation of SAFTA will lead to welfare gains for all South Asian countries, with
the exception of Bangladesh. Such conflicting arguments call for the need to re-examine the
economic impacts of the SAFTA.
Intent on their quest for a greater liberalization, proponents of regionalism are proposing a
broader Asian Economic Community (AEC), encompassing ASEAN+3, ASEAN+4, ASEAN+6,
and/or East Asian Summit (EAS) countries (see Kumar 2005; Mohanty and Pohit 2007). This
― 25 ―
provides further impetus for an exploration of an additional agenda, i.e., the impact of a
significantly broader integration of South Asia with +5 countries – linking South Asia to the Far
East and the West. SAFTA+5 is, therefore, expected to have an integration potential much more
across-the-board with membership open to some of the most influential economies in the world,
and more so by way of economic mass and geographical coverage.
3.
Methodology and Data
3.1 The GTAP Model and the AGE Framework
Several researchers in the area of international trade and development are ever more becoming
ardent users of the GTAP Model today because the database accompanying the GTAP Model is
well suited to analyze the consequences of a free trade area (see Gehlhar 1997; Young and Huff
1997). In fact, a multiregion AGE approach has a number of advantages over partial
equilibrium in that the model not only allows for endogenous movements of regional prices and
quantities in response of technological change, but also provides a consistent framework that
“avoid pitfalls of under- or overcounting welfare effects in a multimarket setting” amongst
others (Frisvold 1997: 324). According to Raihan (2008: 16), the GTAP Model “is the best
possible way for the ex ante analysis of economic and trade consequences of comprehensive
multilateral or bilateral trade agreements.”
The GTAP Model employed in this paper covers 57 industrial sectors including agriculture,
manufactures and services in 87 countries/regions. It handles the bilateral trade via Armington
assumption (see GTAP 2005). The basic innovations of this model include the treatment of
private household preferences using the non-homothetic Constant Difference of Elasticities
(CDE) functional form, explicit international trade and transport margins, and a global banking
sector that links global savings and investment. It also allows users a wide range of closure
options, including a selection of partial equilibrium closures that facilitate comparison of results
to studies based on partial equilibrium assumptions (see Hertel 1997; Hertel and Tsigas 1997).
The model integrates and incorporates a macro framework of the multiregion open
economy model using a wide set of variables, parameters, and equations (see Swaminathan
1997). In contrast to the closed economy, the multiregion model includes separate conditional
demand equations for domestic and imported intermediate inputs. The savings and investment
are computed on a global basis, so that all savers in the model face a common price for this
savings commodity. This implies that if all markets in the multiregional model are in
equilibrium, all firms earn zero profits, and all households are on their budget constraint, then
global investment must equal global savings to satisfy the Walras’ Law (Brockmeier 2001).
― 26 ―
As in the Michigan Model applied by Brown et al. (2003) that incorporates some aspects
of the New Trade Theory including increasing returns to scale, monopolistic competition, and
product variety, the GTAP Model operates in much the similar way and the database is
formulated and solved using the General Equilibrium Modeling Package (GEMPACK)
software as initially illustrated in Harrison and Pearson (1996). Besides, capital and labor are
assumed to be mobile across economic sectors with the assumption of full employment. The
labor component is divided into skilled and unskilled labor, which is combined in a Constant
Elasticity of Substitution (CES) function to form a composite labor input, and sectoral output is
a CES function of capital and composite labor. The GTAP world model allows for greater
regional and sectoral disaggregation and more detailed treatment of taxes and subsidies.
3.2 Model Calibration and Aggregation Strategy
To estimate and simulate the effects of FTA, we develop two scenarios: a base scenario with
unaltered trade policies, and a free trade scenario amongst SAARC countries (SAFTA effects)
as well as FTAs with the observer countries (SAFTA+5 effects). SAFTA, in this context, stands
for those SAARC countries who are signatories to SAFTA Agreement. As aforementioned, the
descriptor code “+5” is assigned for China, Japan, South Korea, the United States and the EU.
The model evaluates the effects of both bilateral and plurilateral FTAs so as to precisely
compare and contrast the extent of these effects quantitatively. The model takes into account
the cross-sectional data from a single base period, and imposes a detailed theoretical structure
on the interactions amongst different data elements. The model is exploited by changing the
shocks and observing how the remaining variables adjust. This is a comparative-static model
that can be effectivley used to analyze the reactions of the economy at a point in time. The
results exhibit the difference between two alternative future states, with and without the policy
shock.
From 87 regions and 57 sectors, the GTAP dataset for this model is aggregated down to 10
regions and 20 sectors, respectively (see Table 8). Individual country/region is separated to the
extent possible so as to distinguish the welfare and trade effects of policy changes by
country/region and sectors based on similarities in factor shares and characteristics. The
regional analysis largely focuses on the SAARC countries. The five primary factors include
land, unskilled labor, skilled labor, capital, and natural resources. The aggregations are set up
with a view to test five major effects under a number of different scenarios/experiments: (i)
Effects of bilateral FTAs amongst SAARC countries, (ii) Effects of plurilateral FTA amongst
SAARC countries, (iii) Effects of bilateral FTAs of SAARC members as a single entity with +5,
― 27 ―
(iv) Effects of bilateral FTAs of individual SAARC countries with +5, and (v) Effects of
plurilateral FTAs of SAARC members as a single entity with +5.
Table 8. Regional and Sectoral Aggregation
Sector Code
Description
Food and agriculture products
1 Crops
Farm animals and products
2 Livestock
Dairy and meat products
3 Dairy
Forestry and logging
4 Forestry
Fishing and related activities
5 Fishing
Mining and extraction
6 Mining
Beverages and tobacco products
7 Beverages
Manufactures and recycling
8 Manufactures
Textiles and clothing
9 Textiles
Leather tanning and products
10 Leather
Chemical and mineral products
11 Chemical
Automobiles and spares
12 Automobile
Metals and metal products
13 Metals
Office equipment and apparatus
14 Electronics
Machinery and equipment
15 Machinery
Basic utilities
16 Utility
Retail and wholesale trade
17 Trade
Transport and communication
18 Transport
Construction works
19 Construction
Other services
20 Services
Note: Rest of South Asia (RSA) includes Afghanistan, Bhutan, Maldives, Nepal, and Pakistan.
Source: GTAP6 database.
1
2
3
4
5
6
7
8
9
10
Region Code
BGD
IND
LKA
RSA
CHN
JPN
KOR
USA
EU
ROW
Description
Bangladesh
India
Sri Lanka
Rest of South Asia
China
Japan
South Korea
United Sates of America
European Union 27
Rest of the World
3.3 Import Tariff and Export Subsidy
The analyses in this paper lay special focus on the reduction of import tariffs and export
subsidies mainly because these are the two most important protection measures available in
quantifiable terms that influence trade in South Asia to a large extent.14
Table 9. Import Tariffs by Source (mean % ad valorem rate)
S/N
Code
Country/Region
1
BGD
Bangladesh
2
IND
India
3
LKA
Sri Lanka
4
RSA
Rest of South Asia
5
CHN
China
6
JPN
Japan
7
KOR
South Korea
8
USA
United Sates of America
9
EU
European Union 27
10
ROW
Rest of the World
Source: Author’s calculation based on GTAP6 database.
Import Tariffs
19.4
31.9
14.5
20.9
16.5
8.7
14.7
2.8
4.4
9.2
Table 9 provides an overview of the average import tariffs levied by SAARC, +5 and rest
of the world (ROW) by source. The average import tax imposed amongst the SAARC countries
is 21.7 percent. Japan, the United States, the EU and ROW impose well below 10 percent. As
regards the export subsidies, the United States and the EU provide fairly significant amount of
― 28 ―
subsidies to the agriculture sector, viz., crops, dairy and livestock. As such, subsidies have also
a bearing in our simulations.
3.4 The Data
The GTAP6 Aggregate Package (GTAPAgg6) is the main source of the data for simulations.
The full GTAPAgg6 covers 87 countries or regions, 57 commodities or sectors, and five
primary sectors. The database corresponds to the world economy based on 2001 benchmark.
The GTAPAgg6 helps prepare an aggregation scheme and then uses the scheme to prepare an
aggregated database for the GTAP economic model. The RunGTAP software program, version
3.40 is used to run the general equilibrium simulations, which is designed to work with version
6.2 of the GTAP Model and the GTAP6 database. The RunGTAP is a visual interface to the
GTAP Model. It supports various versions of GTAP which are distinguished chiefly by level of
aggregation. It incorporates a detailed treatment of international trade margins and other
enhancements (see GTAP 2005).
4.
Simulation Scenarios and Experimental Design
Experiments are based on 10 regions and 20 sectors model using a full multiregion general
equilibrium closure. Simulations are designed in such a way so as to capture the effects of
SAFTA as well as FTAs with +5 both in terms of intra-regional and inter-regional dimensions.
Policy experiments are carried out exhaustively and encompass several major integration
options. The experiments are novel in two respects. First, for each set of simulation scenarios,
experiments are performed in three stages by applying fixed, equal, and varying tariff
combinations. Second, each group of experiments is meticulously arranged with the aim of
creating FTA negotiation scenarios as close to reality as possible. The tariff rates are ultimately
lowered down to 0-5 percent in equal annual installments in keeping with the objective of the
SAFTA tariff reduction schedule, setting the highest tariffs at 30 percent for LDCs and 20
percent for non-LDCs.
For example, under the plurilateral FTA, the first stage experiments start from 10 percent,
5 percent, and then 0 percent for fixed tariffs on all traded commodities by all contracting
parties. The second stage experiments comprise tariff combinations of 10 percent, 5 percent, 0
percent as well, but they are levied equally by both the contracting parties for a maximum of
three most protected sectors. The third stage experiments consider removal of protections in
varied combinations (e.g., 30 percent - 20 percent, 10 percent - 5 percent, and 5 percent - 0
percent), selecting up to three sectors with the highest tariff rates. Where the tariff rates of
LDCs are already below 30 percent, it starts from 20 percent; and where the nonLDCs/developed countries’ tariff rates are already below 20 percent, the next level starts from
― 29 ―
10 percent or below, depending upon each case. The benefit of such a manipulation is that the
technique partly reflects an actual negotiation process, and in addition, we can find out whether
the preferential treatment, i.e., compensation by way of tariff concessions offered by nonLDCs/developed countries to LDCs/developing countries can indeed be welfare enhancing to
the contracting parties. In other words, we test whether the lowering or the complete removal of
tariffs from the most protected sectors will be welfare improving to the parties involved. Thus,
tariff combinations are designed on a case-by-case basis depending upon the results of previous
experiments. The final objective is to find the best possible tariff combinations for most viable
FTAs.
5.
Simulation Results
5.1 Effects of SAFTA
In this section, we discuss the results that estimate the welfare effects of the SAFTA as given by
the Equivalent Variation (EV).15 The EV is the amount or percentage of additional income that
consumers require to achieve the post-simulation level of utility given pre-simulation price
level. A positive value indicates welfare improvement and a negative value denotes welfare
deterioration.16 The results of the simulations for bilateral as well as plurilateral FTAs amongst
SAARC countries are as follows:
5.1.1 Welfare Gains and Losses
Figure 1 depicts the case of the most feasible bilateral FTA effects. Clearly, the welfare of
Figure 1. SAFTA Welfare Gains and Losses
(Bilateral: amongst SAARC)
SAARC countries increases by
nearly US$339 million. It appears
that both +5 and ROW incurs a net
339.15
Region
SAARC
welfare loss of about US$140 and
US$54 million, respectively. This
+5
-140.07
means
that
combinations
feasible
amongst
tariff
SAARC
-54.42
ROW
-200
the
countries would manifestly help
-100
0
100
200
300
400
Million US$
increase trade flows within the bloc.
The welfare loss for +5 and ROW
can be explained by the fact that SAARC’s imports from +5 and ROW are diverted, as SAARC
members increase trade within the bloc.
Applying a fixed tariff in the case of plurilateral FTA amongst SAARC countries show
that Bangladesh is the biggest loser, whereas India and RSA are the largest gainers. All other
countries including ROW face a welfare loss. As regards the bilateral FTA amongst SAARC
― 30 ―
countries, India gains considerably through BDG-IND FTA. Further reduction of tariffs
increases welfare for India and vice-versa for others. BDG-LKA FTA brings gains only to Sri
Lanka, while BDG-RSA FTA brings exclusive gains to RSA. IND-LKA FTA generates welfare
gains to both the countries. In terms of IND-RSA FTA, both India and RSA gain significantly
at tariff rates of 10 percent and 5 percent, but India faces a welfare loss at 0 percent. Except for
China, all others lose. Finally, the LKA-RSA FTA shows that both Sri Lanka and RSA gain as
tariff is further reduced.
In the second stage simulations with equal tariff combinations, Bangladesh is again the
biggest loser, but the losses are not so significant. India, despite being the winner, does not gain
as much as in the first stage. RSA reigns as the biggest winner though. Amongst +5, most of
them lose with the exception of China. As regards the bilateral FTAs, welfare improves for
India, Sri Lanka and RSA, while it declines for Bangladesh. Concurrently, the welfare of China,
Japan, the United States and the EU also improves.
In order to find the best possible combinations for Bangladesh to be better off, a third set
of experiments with varying tariff combinations was performed. This time, the LDCs
(Bangladesh and RSA) impose a fixed tariff rate of 30 percent to non-LDCs (India and Sri
Lanka), while the LDCs impose tariff at a descending rates of 20 percent, 5 percent and 0
percent. Interestingly, the experiment with 30 percent - 20 percent tariff combination results in
gains for all SAARC countries except for Sri Lanka; while all non-members become worse off
except for China. When the tax rate is lowered down to 10 percent - 5 percent combination,
Bangladesh becomes worse off than Sri Lanka. All non-members as well become worse off.
Lowering it further to 5 percent - 0 percent is not an optimal combination as only India and
RSA gains.
From the above experiments, it is deduced that with the right combination of tariffs, FTA
improves welfare of both the contracting parties. A good example is the case of LKA-RSA FTA.
Overall, this provides a good support for our first hypothesis.
5.1.2 Trade Creation and Trade Diversion Effects
Next, we discuss the changes in export sales in 20 sectors of Bangladesh, India, Sri Lanka, RSA,
China, Japan, South Korea, the United States, the EU, and ROW under the plurilateral
SAFTA.17 First, we take a look at India’s import changes in the crops sector. India increases its
import of crops by US$182.69 million from Bangladesh, US$114.98 million from Sri Lanka,
and US$209.06 from RSA. The total increase in imports of crops accounts for US$506.73
million. On the contrary, India decreases its import of crops by US$2.51 million from China,
US$11 million from Japan, US$2.79 million from South Korea, US$16.04 million from USA,
― 31 ―
US$19.91 million from the EU, and US$43.20 million from ROW, which sums up to US$95.45
million. The difference between the increase in trade volume of crops (U$506.73) and a
decrease in trade volume of crops (US$95.45) is the trade creation effect as a result of the
SAFTA, which is equal to US$ 411.28 million.
Similarly, RSA also increases its total trade volume with Bangladesh accounting for
US$413.13 million, with India accounting for US$2,005.39 million, and with Sri Lanka
accounting for US$40.08 million. At the same time, RSA decreases its trade with other RSA
members accounting for US$23.47 million, with China : US$82.90 million, Japan : US$58.80
million, South Korea : US$49.09 million, the United States : US$482.46 million, the EU :
US$485.35 million, and ROW : US$547.45 million. The net trade creation effect is US$729.08
million (2,458.60–1,729.52). This indicates that there is a significant trade creation effect
particularly amongst the SAARC members under the SAFTA scenario.
As for non-members, China decreases its trade with SAARC members and increases its
trade with the outside world. For instance, China decreases its trade with Bangladesh by
US$219.42, with India by US$27.50, with Sri Lanka by US$10.26, and with RSA by US$15.88.
Therefore, the total decrease in trade volume of China’s trade with the SAARC members adds
up to US$273.06 million. This decrease represents the trade that is diverted away from the
SAARC region as a result of the SAFTA, and so it is termed as the trade diversion effect.
5.1.3 Changes in Industry Output, Private Household Demand, Aggregate Exports and
Aggregate Imports
The launching of SAFTA has major impacts on industry’s output, household demand and
exports and imports of seven SAARC countries. The industry output of SAARC countries shifts
significantly under the SAFTA scenario. Bangladesh’s agriculture and service industries shrink,
while the manufacturing sector expands. Both India and Sri Lanka’s manufacturing and service
sector expands, but their agriculture sector declines. The case of RSA is just the reverse: the
agriculture sector expands while the manufacturing and service sectors decline. As expected,
little impact is observed as far as the non-members are concerned. Considering the resource
endowments of each of the countries, the changes in the pattern of production are not surprising.
Bangladesh, India, and Sri Lanka are continuously moving away from the traditional agriculture
to more broad-based growth in the manufacturing sector. For example, the SAFTA scenario
expands Bangladesh’s textile sector by US$498 million. Likewise, India emerges as the major
supplier of chemical (US$346.7 million), automobile (425.6 million) and machinery (381.5
million). Agriculture still stands as a dominant sector for the RSA.
― 32 ―
To the extent that private household demand is concerned, the demand for both agriculture
and manufactured products in all SAARC countries increases. There is a rise in demand for
services especially in India, Sri Lanka and RSA with the exception of Bangladesh, but there is a
decline in demand in the case of all non-members. Aggregate exports and imports in agriculture
and manufacturing sectors increase in all SAARC countries, while the reverse is true for the
non-members.
5.1.4 Changes in Terms of Trade, GDP Indices, and Allocative Efficiency
Table 10 illustrates the changes in terms of trade, GDP indices, and allocative efficiency. The
results show that SAFTA has positive effects on the terms of trade and GDP price indices of
India, Sri Lanka and RSA, while it has negative effects on Bangladesh and the non-members.
The results with regard to Bangladesh is consistent with the study by Raihan and Razzaque
(2007), who find that Bangladesh incurs a net welfare loss because the positive trade creation
effect is not large enough to offset the negative trade diversion effect. India receives the largest
gains in terms of GDP as well as allocative efficiency followed by RSA. This supports the
argument that an FTA is beneficial to member countries, but detrimental to non-member
countries. Non-members are at a disadvantage as a result of the trade diversion effect.
Table 10. Changes in Terms of Trade, GDP Indices, and Allocative Efficiency
under SAFTA Scenario
Change in GDP
Allocative
Change in Terms
Change GDP
Quantity Index
Efficiency
Countries
of Trade (%)
Price Index (%)
(US$ million)
(Regional EV)
BDG
-1.22
-0.86
-112.69
-112.47
IND
0.28
0.34
166.41
166.41
LKA
0.92
1.12
21.23
21.24
RSA
2.47
2.99
72.66
72.69
CHN
-0.01
-0.02
6.38
6.37
JPN
-0.01
-0.03
-6.25
-6.29
KOR
-0.02
-0.04
-10.78
-10.80
USA
-0.01
-0.02
-14.00
-13.53
EU
0.00
-0.02
-14.00
-14.24
ROW
-0.01
-0.02
-59.50
-59.27
Note: The change in terms of trade (2nd column) and GDP price index (3rd column) are compared to the base
scenario fixed at 1 vis-à-vis the value of the post simulation under the FTA scenario.
Table 11 shows the changes in trade balance and allocative efficiency in three major
sectors of agriculture, manufacturing and services. There are major fluctuations in trade
balances with respect to India and RSA as a result of major shuffling in their industrial output
patterns. However, the allocative efficiencies of both the economies increase in all three sectors.
As for the non-members, there is not much impact except for a growth in the trade balances of
the EU and ROW in the services sector.
― 33 ―
Table 11. Changes in Trade Balance and Allocative Efficiency Effect
under SAFTA Scenario
Sector
Agriculture
Manufacturing
Services
Sector
Agriculture
Manufacturing
Services
BDG
-91.18
-62.05
5.84
BDG
-7.02
-105.72
0.26
Change in Trade Balance (US$ million)
IND LKA
RSA CHN JPN
KOR
USA
EU
ROW
10.46
-36.92
34.08
5.09
162.77
118.40
36.16
-96.04
283.98
-15.32
-87.65
285.56
Allocative Efficiency Effect: Commodity Summary
IND LKA
RSA CHN JPN
KOR
USA
EU
ROW
0.12
-13.66
0.02
-0.56
5.41
-14.86
-8.77
-48.49
-1.65
-483.19
420.69
-254.15
101.63
57.72
7.07
-57.20
14.34
-54.76
7.75
10.72
2.77
434.07
-442.14
-295.83
42.06
27.85
2.80
15.91
-25.96
24.09
1.11
4.92
0.36
26.34
40.67
84.02
-0.16
-0.26
-5.79
-4.68
-6.15
0.04
5.1.5 Viable FTAs amongst SAARC Countries
Table 12 displays the most viable FTAs amongst SAARC members.
Table 12. Viable FTAs under SAFTA Scenario*
Bilateral FTA
Tariffs % combination
Contracting Countries
Fixed
Equal
Varying
IND-LKA
0-0
1
IND-RSA
10-10
2
IND-RSA
5-5
3
10-10
IND-RSA
4
5-5
IND-RSA
5
0-0
IND-RSA
6
10-10
LKA-RSA
7
30-20
BDG-IND
8
20-15
BDG-RSA
9
10-15
IND-LKA
10
20-30
IND-RSA
11
5-10
IND-RSA
12
0-5
IND-RSA
13
10-5
LKA-RSA
14
5-0
LKA-RSA
15
Notes: *Viable FTA refers to FTA scenario that provides welfare gains to both/all the contracting parties at the
tariff level as stipulated under the SAFTA Agreement. Fixed: tariffs are fixed for all traded sectors; Equal: tariffs
are fixed equally for three highly protected sectors; and Varying: tariffs vary for three protected sectors based on
SAFTA tariff reduction schedule and on individual country’s development and trade characteristics.
S/N
While IND-LKA FTA is the most viable within the framework of fixed as well varying
tariff rates, IND-RSA FTA is the most flexible of all, since this FTA would be possible in the
case of fixed, equal and varying tariff combinations. There is a good prospect for BDG-IND
FTA, but only at the varying tariff rates of 30 percent - 20 percent. This may be explained by
the fact that Bangladesh’s export base is still narrow vis-à-vis India, and more so being strongly
dominated by India’s diversified trade pattern in the region. This also implies that further
reduction of tariffs from this level would likely undercut the protected industries in Bangladesh,
such as textiles and leather by Indian producers with similar line of products. Sri Lanka and
RSA could have a successful FTA at varying rates within the range of 0-10 percent. Overall,
our hypothesis is further underpinned by the fact that there are maximum possible combinations
― 34 ―
available for successful FTAs at varying tariff combinations.18
5.2 Effects of SAFTA+5
5.2.1 Welfare Gains and Losses
The results for SAFTA+5 scenario in the case of bilateral and plurilateral FTAs are as follows:
Figure 2. SAFTA+5 Welfare Gains and Losses
(Bilateral: SAARC as Single Entity)
Figure 2 depicts the results of the
bilateral SAFTA+5 effects when
SAARC behaves as a single entity.
Region
SAARC gains to the sum of
696.75
SAARC
US$696.75 million; however, the
178.75
+5
gains to +5 are not so significant
because of the trade diversion from
-1,518.41
ROW
+5 by SAARC members into the
-2000
-1500
-1000
-500
0
500
1000
Million US$
intra-regional bloc.
Figure 3 shows the results of
Figure 3. SAFTA+5 Welfare Gains and Losses
(Bilateral: SAARC as Individual Countries)
when
bilateral
SAFTA+5
SAARC
countries
effects
have
individual FTA with +5. In this
BDG
IND
Country/Region
the
case, the total gains for SAARC
LKA
RSA
countries reduce to US$404.73
CHN
JPN
million, while the welfare of +5
KOR
USA
improves quite substantially with
EU
ROW
-800
-600
-400
-200
0
200
400
the exception of South Korea and
600
Million US$
ROW
EU
Welfare -752.50 319.58
USA
KOR
JPN
CHN
RSA
LKA
IND
BDG
-85.01
-87.39
510.35
306.24
241.20
50.17
94.81
18.55
the United States. This sends a clear
signal for why +5 in general will
benefit by integrating with the South Asian countries. ROW is yet again a loser for the same
reason as stated earlier.
Under the plurilateral FTA for SAARC taken as single entity, the results show that except
for South Korea and ROW, all others lose. The tariff combination at fixed 10 percent for all
traded commodities is certainly not a feasible proposition, and therefore, it will not be
acceptable to the losers. Subsequent experiments show significant gains to +5, but SAARC
loses. Under the bilateral FTA between SAARC and CHN, SAARC loses. Similarly, SAARCJPN FTA and SAARC-KOR FTA provide significant welfare gains to Japan and South Korea,
but the opposite is true for SAARC. In the case of SAARC-USA FTA both the parties gain.
However, in the case of SAARC-EU FTA, only the EU gains significantly, while SAARC loses.
― 35 ―
It may be noted that the difference in welfare losses to the United States and the EU as a result
of removal of its subsidies is rather insignificant. The results show that SAARC countries will
definitely reap benefits from the FTAs with +5 on fixed tariff combinations, but only on a caseby-case basis. Therefore, it is in the best interest of SAARC countries to enter into FTA on a
selective basis and liberalize only those sectors that ensure definite gains. The results also
suggest the existence of comparative advantage of SAARC nations over some of its partners. A
careful observation evinces that SAARC members face welfare losses through FTAs with
China and the EU particularly because the sectors they deal with are homogenous and highly
competitive, while the opposite is true for Japan, South Korea and the United States, wherein
the sectors are much differentiated.
It is fascinating to note in the last three experiments, in which the tariff protections are
levied at varying rates (30 percent - 20 percent, 10 percent - 5 percent, and 5 percent - 0
Table 13. Welfare Gains and Losses
percent), in addition to removal of agricultural
Region/Countries
subsidies by the United States and the EU, the gains
Welfare (US$ ml)
SAARC
CHN
JPN
KOR
USA
EU
ROW
1,203.20
7,571.00
1,754.00
6,200.00
6,130.70
95.80
-6,951.00
for SAARC countries increase considerably. In the
final experiment, where the tariff protections on
SAARC by +5 are only 5 percent and 0 percent, the
gains for SAARC accounts for nearly US$1.2
billion. At the same time China, South Korea, and the United States also gain significantly
accounting for about US$7.6 billion, US$6.2 billion, and US$6.1 billion, respectively (see
Table 13). All things considered, SAARC as a single entity should have no compunction to opt
for an FTA arrangement with +5. Experiments clearly indicate that the welfare of SAARC as
well of all the contracting members improve to a large extent. In other words, the highest gain
comes to the SAARC bloc if the
Figure 4. SAFTA+5 Welfare Gains and Losses
(Plurilateral: SAARC as Single Entity)
tariffs are lowered by SAARC and
+5 down to 5 percent and 0 percent,
respectively. This is an interesting
1,203.20
SAARC
Region
result because the plurilateral FTA
21,751.50
+5
with maximum liberalization is the
most rewarding of all FTAs to the
ROW
entire
-6,951.00
-10,000
-5,000
0
5,000
10,000
15,000
20,000
25,000
bloc,
inclusive
of
both
SAARC and +5 countries.
Figure 4 depicts the results of
Million US$
― 36 ―
the plurilateral SAFTA+5 effects as discussed above. While SAARC gains US$1.2 billion,
which is almost the double as compared with that of the bilateral effects, but +5 receives the
largest share of welfare gains under this scenario, amounting to 21.8 billion. This is not
surprising considering the fact that +5 is dominated by some of the world’s largest and
strongest economies. The results also suggest that those countries lacking comparative
advantage in terms of resource endowments, technology and the like, will be worse off in a free
trade. In order for all countries to enjoy the benefits of free trade in a plurilateral FTA scenario,
any country that loses will need to be compensated by winners. Therefore, the results once
again render support to our hypothesis that compensating the losers by winners by way of tariff
concessions results in welfare gains for all concerned.
5.2.2 Trade Creation and Trade Diversion Effects
As a case in point, China increases its import of crops by US$1.35 million from SAARC,
US$2,011.45 million from Japan, US$4,458.15 from South Korea, US$5.35 million from the
United States, and US$ 2,441.09 million from the EU under the plurilateral SAFTA+5 scenario.
The sum of these increases is US$8,917.39 million. On the other hand, China decreases its
import of crops from ROW by US$149.78 million. Thus, the trade creation is equivalent to
US$ 8,767.61 million (US$8,917.39–US$149.78).
Likewise, Japan increases its trade volume with SAARC by US$854.48 million, with
China by US$3,057.56 million, and with South Korea by US$1,141.52 million. At the same
time, Japan increases its imports from the United States, the EU and ROW accounting for
US$4,852.68 million, US$2,965.18 million and US$5,939.75 million, respectively. This
indicates that there is significant trade creation for Japan under the SAFTA+5 scenario. The
most interesting example of trade creation is the case of the SAARC bloc per se. The total trade
volume of SAARC under this scenario goes up by US$23.41 million from South Korea, and
US$6,384.41 million from the United States, while it decreases its trade from within SAARC
itself by US$157.41 million, from China by US$220.64 million, from Japan by US$309.54
million, from the EU by US$242.44 million, and from ROW by US$872.41 million. The net
trade creation is US$4,605.38 million (US$6,407.82–US$1,802.44). Taking another example,
the EU’s trade volume increases by US$ 20,255.24 million, while it decreases by US$12,715.13
million. The net trade creation of the EU alone under the SAFTA+5 scenario is US$7,540.11
million. Therefore, the overall trade creation effect is much higher if we take all the countries
into account, evidently supporting our second hypothesis. Needless to say, the SAFTA+5
scenario has a much smaller trade diversion effect.
― 37 ―
5.2.3 Changes in Industry Output, Private Household Demand, Aggregate Exports and
Aggregate Imports
The SAFTA+5 scenario has some major effects on industry’s output, household demand, and
exports and imports of all countries. The industry output of SAARC is quite the reverse of what
we saw in the SAFTA scenario. There is evidently a swapping in the comparative advantage
pattern based on trade complementarities and resource endowments: SAARC countries are
forced to pull back to the agriculture sector, while the manufacturing and service sectors are
dominated by Japan and South Korea that have a greater advantage over these sectors. SAARC
and China specialize in similar industries mostly comprising agro-based and manufacturing
sectors, clearly signaling their midway development phases. The United States and the EU are
still the major producers of agriculture products, attributable to their well developed system of
agricultural subsidies. SAARC, China and South Korea become the major exporters of textile
goods, while Japan and the EU specialize in machinery and manufactures, respectively.
Regarding the private household demand, there is a major increase in all the regions.
SAARC’s household demand for machinery goods increases sharply by US$1,147.79 million.
Interestingly, Japan’s agricultural imports expand by US$14,313 million. This shows that there
is a major shuffling of demand for products amongst the regions. This may also mean that, as
the regions specialize in specific products, the resource allocation efficiency improves for all
countries, raising not only the demand but also the overall production of those specialized
products. This trade pattern largely supports the Heckscher-Ohlin Theory
19
that the
international trade is largely driven by differences in country’s resources.
Aggregate exports for all countries increase significantly in agriculture and manufacturing
sectors, except for the United States in the manufacturing sector. SAARC, China, South Korea
and the United States experience a drop in exports of services. Aggregate imports also increase
in all the countries with the exception of Japan in the manufacturing and service sectors. As
expected, ROW’s imports decrease in all major sectors.
5.2.4 Changes in Terms of Trade, GDP Indices, and Allocative Efficiency
Table 14 shows the changes in terms of trade, GDP indices, and allocative efficiency. The
results establish that SAFTA+5 scenario has positive impacts on the terms of trade of all
countries, except for Japan and ROW. There is a mixed effect on change in GDP price indices.
However, the GDP quantity indices as well as the allocative efficiencies of SAARC, China,
Japan, and South Korea increase significantly, while there is a decrease in the case of the
United States, the EU and ROW.
― 38 ―
Table 14. Changes in Terms of Trade, GDP Indices, and Allocative Efficiency
under SAFTA+5 Scenario
Change in GDP
Allocative
Change in Terms
Change GDP
Quantity Index
Efficiency
Countries
of Trade (%)
Price Index (%)
(US$ million)
(Regional EV)
SAARC
0.15
-0.19
793.94
794.10
CHN
0.44
0.30
6,516.75
6,519.40
JPN
-0.48
-1.00
4,046.25
4,044.53
KOR
0.55
-1.61
5,516.31
5,497.42
USA
0.57
0.32
-621.00
-621.55
-85.43
EU
0.02
-0.13
-85.50
-238.87
ROW
-0.30
-0.58
-239.00
Note: The change in terms of trade (2nd column) and GDP price index (3rd column) are compared to the base
scenario fixed at 1 vis-à-vis the value of the post simulation under the FTA scenario.
Table 15 shows the changes in trade balance and allocative efficiency in the three major
sectors of agriculture, manufacturing and services. The trade balance of SAARC is negative for
the manufacturing sector and service sectors. However, there is a large fluctuation in the trade
balances of the two largest economies, the United States and Japan. They exhibit contrasting
changes especially with regard to agriculture and manufacturing sectors. The allocative
efficiencies of SAARC, China, Japan and South Korea turn positive, but in the case of the US
and the EU, it turns negative especially in the agriculture and service sectors.
Table 15. Changes in Trade Balance and Allocative Efficiency Effect
under SAFTA+5 Scenario
Sector
Agriculture
Manufacturing
Services
Sector
Agriculture
Manufacturing
Services
Change in Trade Balance (US$ million)
SAARC
CHN
JPN
KOR
63.50
-1097.94
-83.63
-1475.74
2279.62
-1001.16
-18802.07
17706.20
1979.93
-4803.47
3557.11
-970.82
USA
EU
ROW
35686.28
-34852.72
-4526.61
-731.94
1437.92
838.06
-12568.75
9414.65
7951.61
Allocative Efficiency Effect: Commodity Summary
SAARC
CHN
JPN
KOR
USA
79.17
685.39
21.31
1864.72
4584.64
70.04
2903.42
1115.26
-48.37
4564.39
758.44
129.30
-672.93
290.52
-200.79
EU
ROW
-430.81
566.07
-164.57
-304.98
-255.57
83.25
In Table 16, we report the most viable FTAs that SAARC as a single entity could have
with +5 on a plurilateral basis. SAFTA+5 FTA would be most viable within equal tariffs
ranging from 0-5 percent, as well as varying tariff rates between 0 percent and 10 percent.
However, this FTA is not feasible at fixed tariff rates.
― 39 ―
Table 16. Viable FTAs under SAFTA+5 Scenario (Plurilateral: SAARC as Single Entity)
Plurilateral
Contracting Countries
1
SAFTA and +5
2
SAFTA and +5
3
SAFTA and +5
4
SAFTA and +5
Note: As in Table 12.
S/N
Fixed
Tariffs % combination
Equal
5-5
0-0
Varying
10-5
5-0
Table 17 shows the most feasible tariff structure for bilateral SAFTA+5 FTA. With regard
to the SAFTA-USA FTA, fixed tariff rate of 0 percent - 0 percent combination is the most
feasible. SAFTA-CHN FTA and SAFTA-EU FTA would be feasible under the varying tariff
combinations, but lowering anything below 30 percent - 20 percent and 10 percent - 5 percent
in the case of SAFTA-CHN FTA and SAFTA-EU FTA respectively, is not feasible.
Table 17. Viable FTAs under SAFTA+5 Scenario (Bilateral: SAARC as Single Entity)
Bilateral FTA
Contracting Countries
1
SAFTA-USA
2
SAFTA-CHN
3
SAFTA-EU
Note: As in Table 12.
S/N
Fixed
0-0
Tariffs % combination
Equal
Varying
30-20
10-5
With regard to the bilateral FTA of SAARC as individual countries (see Table 18),
Bangladesh has viable FTAs at fixed and varying rates with China, Japan and South Korea but
at higher tariffs in general. Sri Lanka has the maximum flexibility to the extent of being able to
remove its tariffs completely. Sri Lanka’s FTA is feasible with China, Japan, South Korea, and
the EU at fixed, equal and varying tariffs from a minimum of zero to a maximum of 15 percent.
This is not a big surprise as Sri Lanka’s economy is the most liberalized of all amongst the
SAARC members. RSA also has a good possibility of having viable FTAs, particularly with
China and Japan at fixed, equal and varying tariff rates. As for India, it is quite evident that
IND-EU FTA is the most feasible of all. IND-CHN FTA and IND-JPN FTA are viable, but at a
much higher and varying tariff rates. This implies that the IND-CHN FTA and IND-JPN FTA
are feasible on the condition that China and Japan grants preferential tariff concessions to India.
― 40 ―
Table 18. Viable FTAs under SAFTA+5 Scenario
(Bilateral: SAARC as Individual Countries)
Bilateral FTA
Contracting Countries
BDG-CHN
1
LKA-CHN
2
LKA-CHN
3
LKA-JPN
4
LKA-KOR
5
LKA-EU
6
RSA-CHN
7
RSA-CHN
8
RSA-JPN
9
RSA-JPN
10
RSA-JPN
11
IND-EU
12
IND-EU
13
IND-EU
14
LKA-CHN
15
LKA-CHN
16
LKA-CHN
17
LKA-JPN
18
LKA-JPN
19
LKA-KOR
20
LKA-KOR
21
RSA-JPN
22
RSA-JPN
23
BDG-CHN
24
BDG-JPN
25
BDG-KOR
26
IND-CHN
27
IND-JPN
28
IND-EU
29
IND-EU
30
LKA-CHN
31
LKA-CHN
32
LKA-CHN
33
LKA-KOR
34
LKA-KOR
35
RSA-JPN
36
RSA-JPN
37
Notes: As in Table 12. RS=Removal of subsidies.
S/N
6.
Fixed
10-10
5-5
0-0
0-0
0-0
5-5 + RS
5-5
0-0
10-10
5-5
0-0
Tariffs % combination
Equal
Varying
10-10 + RS
5-5 + RS
0-0 + RS
10-10
5-5
0-0
5-5
0-0
5-5
0-0
5-5
0-0
30-20
20-5
20-5
20-10
20-10
20-5 + RS
5-0 + RS
15-10
10-5
5-0
10-5
5-0
30-5
10-0
Summary of Findings
Part II of this study has evaluated the major effects and welfare implications of SAFTA and
SAFTA+5. Exhaustive experiments were performed using the equivalent variation component
to gauge the welfare gains and losses of different countries/regions. Additional analyses were
carried out to investigate the trade creation and trade diversion effects, changes in industry
output, private household demand, aggregate exports and imports, changes in terms of trade,
GDP indices, and allocative efficiencies. Finally, some viable FTAs were identified amongst
SAARC and +5 countries.
― 41 ―
The findings of this study revealed that plurilateral FTAs with deeper liberalization is a
win-win situation for all members concerned generating largest welfare gains as opposed to
bilateral FTAs. The magnitude of gains varies from one FTA to another. The maximum
possible FTAs emerge from varying combinations of preferential tariffs as compensation by the
non-LDCs/developed countries to the LDCs. Our first hypothesis was strongly buttressed by the
fact that selective combinations of tariff rates result in welfare gains of both the contracting
parties generating a number of possible combinations for feasible FTAs amongst SAARC and
+5 countries. There was also ample evidence suggesting that SAFTA and SAFTA+5 are indeed
welfare enhancing, and result in not trade creation than trade diversion, corroborating our
second hypothesis. All the same, trade liberalization under the SAFTA scenario created lesser
gains vis-à-vis SAFTA+5 scenario. Trade liberalization also caused major fluctuations and
large adjustments in the industry output and sectoral production in all countries; however, the
household demand, aggregate exports and imports, terms of trade, GDP, and allocative
efficiencies for SAARC as well as +5 increased considerably, particularly in the case of
SAFTA+5 scenario.
― 42 ―
PART III
CONCLUSIONS AND POLICY IMPLICATIONS
The findings in Part I of this study definitely support the case for FTAs and further trade
integration amongst SAARC member nations – clearly signalized by the positive impact of
trade agreements seen in the post-SAARC periods. With the growing interest of observers
around the world, SAARC will certainly find new opportunities, but one can only become more
optimistic as SAFTA assuages SAPTA, and matures through further dismantling of both tariff
and non-tariff barriers. The weakness of SAPTA can be compensated by shaping and
sharpening the influence of SAFTA. Although there are undoubtedly good prospects to boost
future exports, this would, however, entail concerted efforts of the member nations to mitigate
conflicts, evolve new comparative advantages and complementarities, aggregate with other
regional blocs, and eliminate the existing impediments to intra-regional trade with the right
perspective and affirmative political will.
In addition, SAARC should take a more holistic and forward-looking approach by
including deeper forms of integration in other trade facilitation measures, such as services,
energy, institutional and infrastructure development, monetary and investment cooperation. In
sum, we can recapitulate that the future of SAARC countries depends, inter alia, not only on the
level of economic integration, but it is also largely dictated by the political soundness in the
region. Without easing political tensions, conflicts, and mistrust amongst the member nations, it
is quite unlikely to hope for any substantive trade integration in the region. Thus, the growth of
regional economic cooperation in South Asia calls for committed efforts and strong political
will from all leaders to bring about peace, harmony, and social security in the region.
In addition, comprehensive analyses in Part II of this study also point to some important
policy implications. First, plurilateral FTA amongst SAARC countries will not be a feasible
proposition, while the same for +5 countries will be the most rewarding of all FTAs. Secondly,
SAARC as a single entity will invariably benefit by having FTAs with South Korea, the United
States and the EU. Thirdly, the results suggest that possibilities exist for several viable FTAs.
Some noteworthy ones are SAFTA - +5, SAFTA-USA, SAFTA-CHN, SAFTA-EU, IND-RSA,
IND-LKA, IND-JPN, IND-EU, LKA-CHN, LKA-JPN, LKA-KOR, LKA-EU, RSA-CHN, and
RSA-JPN. This implies that preferential tariff compensation improves the chances of widening
the possibilities for more FTAs. However, while there will be a marked improvement in the
welfare of the member states, the welfare of ROW could be considerably reduced due to
― 43 ―
substantial trade diversion. Slashing import tariffs for all traded goods by a fixed proportion
will not be in the best interest of all members. As a starter, the best set of tariff combination is
to compensate the losers by way of tariff concessions by the winners. This means that nonLDCs should allow LDCs with a grace period to enable them to liberalize selected sectors over
time. Nevertheless, the long-run implication is to aim for deeper liberalization under the
SAFTA+5 scenario. Last but not least, the findings of this research sends clear signal that both
SAARC as well as the five observers can reap positive benefits by way of greater free trade
pact. In so doing, there is not only the manifestation of trade and economic benefits by taking
advantage of each country/region’s own comparative advantages, but such integration can
become a channel for SAARC countries for greater exposure to the multilateral trading system,
as they can intensify and step up international understanding by associating and learning from
the experiences of veteran leaders in trade, particularly Japan, the United States and the EU.
Closer regional economic cooperation may also mean deeper social, cultural and political ties in
the long run, and therefore, it can become a conduit for initiating a forward-looking approach
for a better world.20
While this research is by no means the end of algorithm on the subject, few limitations
may be set forth as follows. Notwithstanding the endeavor of this research, the Gravity Model
in this paper was kept simple using widely accepted techniques and conventional empirical
methodology. Future work could aim at customizing models with more contemporary
techniques. In the GTAP Model, the importance of other trade barriers, such as para-tariffs and
non-tariff barriers though well recognized could not be considered, as these components do not
lend themselves readily to quantification within the purview of the GTAP analysis. Moreover,
the GTAP6 data pertains to 2001 benchmark, hence future work could employ more recent data.
Lastly, the use of dynamic analysis might be another alternative to deduce more conclusive
findings.
― 44 ―
Endnotes
1 Despite theoretical skepticism of RTAs and the like, “most empirical studies find that trade creation dominates
trade diversion” (Rodríguez-Delgado 2007: 14).
2
Bangladesh declared independence from Pakistan on March 26, 1971 after the Bangladesh Liberation War.
3
See: http://www.southasianmedia.net/Magazine/Journal/safta_critique.htm (accessed August 20, 2007).
4
Afghanistan is excluded in this paper because it was formally admitted as the eighth member of the SAARC
bloc in the 14th Summit held in New Delhi, India on April 3-4, 2007.
5
Eichengreen and Irwin (1996) and Baier and Bergstrand (2005) also scaled the data by GDP deflators to
generate real exports and real GDPs.
6
Since [exp(-0.86)-1*100] = 57.68 percent.
7
For brevity, all tables of results are not shown.
8
See Section 5.4 for more discussion on lagged effects of TRAG.
9
First differencing of the panel data yields some potential advantages over fixed effects. See Wooldridge (2003:
467-468) for further details.
10 Using panel data, Rose (2004) estimated an FTA impact of 156 percent, while Tomz (2004) estimated at 114
percent.
11 During the 15th SAARC Summit held in Colombo, Sri Lanka from August 2-3, 2008, there were nine observer
countries. Other observers besides the five considered in this study are Iran, Myanmar, Mauritius and Australia.
This study takes into account only five observers because Iran, Myanmar, Mauritius are not treated
individually in the GTAP6 database; while Australia is separated, it is but a recent phenomenon. However, the
remaining four observers should certainly be a fruitful area for future research.
12 AGE models are a class of economic model that use actual economic data to estimate how an economy might
react to changes in policy, technology, or other external factors. AGE models are also referred to as
computable general equilibrium (CGE) models. They descend from the input-output models pioneered by
Leontief (1986), but assign a more important role to prices.
13 Bilateral FTAs are often referred to as RTAs as they fall within the domain of regional blocs.
14 Other trade barriers such as para-tariffs and non-tariff barriers (e.g., administrative delays, customs clearance,
restrictions on health safety, environmental, and religious reasons) are not taken into account as they are not
quantifiable within the purview of the GTAP analysis. The same is left for future research.
15 Note that all tables of results are not reported for the sake of brevity and space considerations. They are
available upon request from the author.
16 In economics, equivalent variation (EV) means “how much money would have to be taken away from the
consumer before the price change to leave him as well off as he would be after the price change” (Varian
2003: 255). The value of the equivalent variation is given in terms of the expenditure function: EV = e(p0,u1)
− e(p0,u0) = e(p0,u1) − w = e(p0,u1) − e(p1,u1), where w is the wealth level, p0 and p1 are the old and new
prices respectively, and u0 and u1 are the old and new utility levels, respectively.
17 Table of results are not reported for brevity.
18 There are many more combinations where one of the contracting parties gains while the other loses. They are
not considered because our interest lies in finding the best possible tariff combinations that would most likely
be acceptable or feasible to both/all contracting parties.
― 45 ―
19 The theory emphasizes the interplay between the proportions in which different factors of production are
available in different countries and the proportions in which they are used in producing different goods (see
Krugman and Obstfeld 2003: 67-86 for further details).
20 A good precedent along the line is that Japan is already providing US$200,000 annually to SAARC through
the Japan-SAARC Special Fund. Japan’s contribution and participation as an observer in itself is a testimony
to the deepening of relationship and commitment to further cooperation between Japan and the SAARC
countries (see http://www.nerve.in/news:25350010397, accessed February 22, 2008).
― 46 ―
References
ADB (2006), Asian Development Outlook 2006: Routes for Asia’s Trade, Manila: Asian Development Bank.
Aristotelous, K. (2001), “Exchange-rate volatility, exchange-rate regime, and trade volume: Evidence from the
UK-US export function (1889-1999),” Economic Letters, 72: 87-94.
Asami, T. (2005), “What is the Practical Approach to Promote an ‘East Asian Community’?” Newsletter 2005.8.1
(No.3, 2005), Tokyo: Institute for International Monetary Affairs.
Baak, S.J. (2004), “Exchange Rate Volatility and Trade amongst the Asia Pacific Countries,” Journal of
International Economic Studies, 8: 93-116.
Baier, S.L. and J.H. Bergstrand (2004), “Economic Determinants of Free Trade Agreements,” Journal of
International Economics, 64: 29-63.
Baier, S.L. and J.H. Bergstrand (2005), “Do Free Trade Agreements Actually Increase Members’ International
Trade?,” Working Paper No. 2005-3, Federal Reserve Bank of Atlanta.
Bandara, J.S. and M. McGillivray (1998), “Trade Policy Reforms in South Asia,” The World Economy, 21: 881896.
Bandara, S.J. and W. Yu (2003), “How desirable is the South Asian Free Trade Area? A Quantitative Economic
Assessment,” The World Economy, 26: 1293-1323.
Batra, A. (2005), “Country Perspectives: India,” in SAFTA-Opportunities and Challenges, 57-94. Washington
D.C.: United States Agency for International Development (USAID).
Baysan, T., A. Panagariya, and N. Pitigala (2006), “Preferential Trading in South Asia,” World Bank Policy
Research, Working Paper No. 3813, Washington D.C.: The World Bank.
Bhagwati, J. (2005). “Reshaping the WTO,” Far Eastern Economic Review, 162(22): 28.
Bhargava, K.K. (1998), “EU - SAARC: Comparisons and Prospects of Cooperation,” Discussion Paper C 15,
Center for European Integration Studies.
Billen, D., M.M. Garcia and N. Khasanova (2005), “Is the Effect of Exchange Rate Volatility on Trade More
Pronounced in Latin America than in Asia?,” Working Paper No. 434, The Kiel Institute for the World
Economy, Düsternbrooker Weg 120, D-24105 Kiel.
Brockmeier, M. (2001), “A Graphical Exposition of the GTAP Model,” GTAP Technical Paper No. 8. Available
at: https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=311 (accessed March 31,
2008).
Brown, D.K., A.V. Deardorff, and R.M. Stern (2003), “Multilateral, Regional and Bilateral Trade-Policy Options
for the United States and Japan,” World Economy, 26: 803-828.
Burfisher, M.E., and S. Zahniser (2003), “Multilateralism and Regionalism: Dual Strategies for Trade Reform.”
Available at: http://www.ers.usda.gov/AmberWaves/September03/Features/Multilateralism.htm (accessed
November 20, 2007).
Caves, R.E., J.A. Frankel, and R.W. Jones (2007), World Trade and Payments, 10th Edition, Boston: Pearson
International Edition.
Chowdhury, K. (2004), “Convergence of Per Capita GDP across SAARC Countries,” Working Paper No. 04-07,
Economics Working Paper Series, University of Wollongong.
― 47 ―
CUTS (2005), “Economic Cooperation in South Asia: Current State and Prospects,” Briefing Paper No. 2/2005,
CUTS Centre for International Trade, Economics and Environment (CUTS-CITEE).
Dash, K.C. (1996), “The Political Economy of Regional Cooperation in South Asia,” Pacific Affairs, 69: 2.
Dell’Ariccia, G. (1999), “Exchange Rate Fluctuations and Trade Flows: Evidence from the European Union,” IMF
Staff Papers, 46: 315-331.
Doshi, H. (2008), “Multilateral vs. Bilateral trade agreements,” UPI Asia Online, January 29, 2008. Available at:
http://www.upiasiaonline.com/Economics/2008/01/28/multilateral_vsbilateral_trade_agreements/7323/
(accessed March 8, 2008).
Eichengreen, B. and D. Irwin (1996), “The role of history in bilateral trade flows,” NBER Working Paper No.
5565.
Frisvold, G.B. (1997), “Multimarket effects of agricultural research with technological spillovers,” in T.W. Hertel
ed., Global Trade Analysis: Modeling and Applications, 321-346. Cambridge: Cambridge University Press.
Gehlhar, M.J. (1997), “Historical analysis of growth and trade patterns in the Pacific Rim: An evaluation of the
GTAP framework,” in T.W. Hertel ed., Global Trade Analysis: Modeling and Applications, 349-363.
Cambridge: Cambridge University Press.
Ghani, E., M. Din (2006), “Regional Trade Integration in South Asia: Rationale, Impediments and the Way
Forward,” Artnet, Policy Brief No. 7.
GTAP (2005), RunGTAP: A CGE Model of World Trade (Software), Global Trade Analysis Project, Purdue
University, Indiana, USA. GTAP Home Page: https://www.gtap.agecon.purdue.edu/
Harris, M.N., L. Mátyás (1998), “The Econometrics of Gravity Models,” Working Paper No. 5/98, Melbourne
Institute of Applied Economic and Social Research, The University of Melbourne, Australia.
Hassan, M.K., R.A. Mehanna and S.A. Basher (2002), “Regional Cooperation in Trade, Finance and Investment
amongst SAARC Countries: The Bangladesh Perspective,” Thoughts on Economics, 12(2): 7-32.
Hirantha, S.W. (2004), “From SAPTA to SAFTA: Gravity Analysis of South Asian Free Trade,” Department of
Commerce, Faculty of Management Studies and Commerce, University of Sri Jayewardenepura, Sri Lanka.
Harrigan, F., W. James, M. Plummer, and F. Zhai (2006), “Bilateralism or Regionalism: Alternative Scenarios for
Asian Trade Liberalization.” Paper presented at the Long-term Scenarios of Asian Growth and Trade
Finalization Conference, Shaping the future: Prospects for Asia’s long-term development over the next two
decades, 11-12 December 2006, ADB Thailand Resident Mission Conference Room.
Harrison W.J., and K. Pearson (1996), “Computing Solutions for Large General Equilibrium Models using
GEMPACK,” Computational Economics, 9: 83-127.
Hertel, T.W. (1997), Global Trade Analysis: Modeling and Applications. Cambridge: Cambridge University Press.
Hertel, T.W., and M.E. Tsigas (1997), “Structure of GTAP,” in T.W. Hertel ed., Global Trade Analysis: Modeling
and Applications, 13-73. Cambridge: Cambridge University Press.
Hudgins, E.L. (1996), “Regional and multilateral trade agreements: Complementary means to open markets,” Cato
Journal, 15(2-3): 231-241.
Jaggers, K. and R.T. Gurr (1999), “Transitions to Democracy: Tracking Democracy’s third Wave with Polity III
Data,” Journal of Peace Research, 32: 469-482.
Jayaratne, C. (2004), “Promoting Regional Integration in South Asia: A Private Sector Perspective,” Sri Lanka
Country Note, Asia Region. Washington D.C.: The World Bank.
― 48 ―
Kalegama, S. (2004), “SAFTA: A Critique,” South Asian Journal, April-June, 2004. Available at:
http://www.southasianmedia.net/Magazine/journal/safta_critique.htm (accessed September 6, 2006)
Kaufmann, D., A. Kraay and M. Mastruzzi (2003), “Governance Matters III: Governance Indicators for 19962002,” World Bank Policy Research, Working Paper No. 3106.
Kemal, A.R. (2004), “SAFTA and Economic Cooperation, Regional Cooperation in South Asia,” South Asian
Journal. Available at: http://www.southasianmedia.net/Magazine/Journal/ (accessed August 5, 2007).
Khor, M. (2006), “The WTO’s Doha Negotiations and Impasse: a Development Prespective,” Third World
Network, 16.
Krugman, P.R., and M. Obstfeld (2003), International Economics: Theory and Policy, 6th Edition, Boston:
Addison Wesley.
Kumar, N. (2005), “Towards a Broader Asian Community: Agenda for the East Asia Summit,” RIS-DP # 100,
November 2005, Research and Information System for Developing Countries (RIS), New Delhi.
Leontief, W. (1986), Input-Output Economics, New York: Oxford University Press.
Low, L. (2004), “A Comparative Evaluation and Prognosis of Asia Pacific Bilateral and Regional Trade
Arrangements,” Asian-Pacific Economic Literature, 18(1): 1-11.
Mansfield, E., V.M. Helen and B.P. Rosendorff (2002), “Why Democracies Cooperate More: Electoral Control
and International Trade Agreements,” International Organization, 56: 477-514.
Mátyás, L. (1997), “Proper Econometric Specification of the Gravity Model,” The World Economy, 20: 363-368.
Mátyás, L. (1998), “The Gravity Model: Some Econometric Considerations, The World Economy, 21: 397-401.
Mátyás, L. and L. Kónya, M.N. Harris (2000), “Modelling Export Activity of Eleven APEC Countries,” Working
Paper No. 5/00, Melbourne Institute of Applied Economic and Social Research, The University of
Melbourne, Australia.
Mohanty, S.K. (2003), “Regional Trade Liberalisation under SAPTA and India’s Trade Linkages with South Asia:
An Empirical Assessment,” RIS-DP # 48/2003, Research and Information System for the Non-Aligned and
Other Developing Countries (RIS), New Delhi.
Mohanty, S.K., and S. Pohit (2007), “Welfare Gains from Regional Economic Integration in Asia: ASEAN+3 or
EAS,” RIS-DP # 126, Research and Information System for Developing Countries (RIS), New Delhi.
Mukherji, I.N. (2000), “Charting a Free Trade Area in South Asia: Instruments and Modalities.” Paper presented to
the Second Conference of the South Asia Network of Economic Research Institute (SANEI), Kathmandu,
28-29 August.
Mukherji, I.N. (2004), “Towards a Free Trade Area in South Asia: Charting A Feasible Course for Trade
Liberalisation with Reference to India’s Role,” RIS-DP # 86/2004, Research and Information System for
Developing Countries (RIS), New Delhi.
Oyane, S. (2001), “‘Plurilateralism’ of the United States and its APEC policies,” Working Paper Series 00/01 No.
5, Institute of Developing Economies (IDE), APEC Study Center.
Pattanaik, S.S. (2006), “Making Sense of Regional Cooperation: SAARC at Twenty. Institute for Defense Studies
and Analyses, Strategic Analysis, 30: 139-160.
Pitigala, N. (2005), “What Does Regional Trade in South Asia Reveal about Future Trade Integration? Some
Empirical Evidence,” World Bank Policy Research, WP No. 3497, Washington D.C.: The World Bank.
Raihan, S. (2008), “SAFTA and the Bangladesh Economy: Assessments of Potential Implications.” Discussion
Paper No. 3, January 2008, South Asian Network on Economic Modeling (SANEM), Dhaka.
― 49 ―
Raihan, S., and A. Razzaque (2007), “Regional Trading Arrangements (RTAs) in South Asia: Implications for the
Bangladesh Economy,” in Raihan, S., and Razzaque, A. (eds.), WTO and Regional Trading Arrangement:
Quantitative Assessment of Potential Implications for Bangladesh, Dhaka: Pathak Samabesh.
Rodríguez-Delgado, J.D. (2007), “SAFTA: Living in a World of Regional Trade Agreements,” IMF Working
Paper No. WP/07/23, Asia and Pacific Department, Washington D.C.: International Monetary Fund.
Rose, A.K. (2004), “Do We Really Know That the WTO Increases Trade?” The American Economic Review, 94:
98-114.
SAFTA (2004), South Asian Free Trade Area (SAFTA) Agreement. Available at: http://www.saarcsec.org/main.php?t=2.1.6 (accessed November 23, 2007).
Singh, A.S. (2005), “What makes regional integration work? Lessons for South Asia,” Briefing Paper No. RECSA
3/2005, CUTS International 2005, Economics and Environment (CUTS-CITEE), Bhaskar Marg, Jaipur.
Swaminathan, P. (1997), “Glossary of GTAP Notation,” in T.W. Hertel ed., Global Trade Analysis: Modeling and
Applications, 386-396. Cambridge: Cambridge University Press.
Tomz, M., J. Goldstein and D. Rivers. (2004), “Membership Has Its Privileges: The Impact of GATT on
International Trade,” Stanford University, Department of Political Science, Working Paper, June 2004.
UNCTAD (2005), UNCTAD Handbook of Statistics 2005. Geneva: United Nations Conference on Trade and
Development, The United Nations.
U.S. Council of Economic Advisers (1995), Economic Report of the President: Transmitted to the Congress
Together with the Annual Report of the Council of Economic Advisers, Washington D.C.: U.S. Government
Printing Office.
Varian, H. R. (2003), Intermediate Economics: A Modern Approach, 6th Edition, New York: W.W. Norton & Co.
Weintraub, S. (1996), “Regionalism and Multilateralism in International Trade.” Paper presented at the Mont
Pelerin Society’s regional meeting, Cancun, Mexico, 17 January, 1996.
White, H. (1980), “A Heteroskedasticity-Consistent Covariance Matrix and a Direct Test for Heteroskedasticity,”
Econometrica, 48: 817-838.
Wooldridge, J.M. (2003), Introductory Econometrics: A Modern Approach. 2nd Edition, South Western, Ohio:
Thompson Learning.
World Bank (2005), World Development Indicators 2005, CD-ROM. Washington D.C.: The World Bank.
Young, L.M, and K.M. Huff (1997), “Free trade in the Pacific Rim: On what basis?,” in T.W. Hertel ed., Global
Trade Analysis: Modeling and Applications, Cambridge: Cambridge University Press, 235-252.
― 50 ―
The Impact of Trade Agreements and Regional Economic Integration on Trade
Flows
2009年2月
第1版第1刷発行
非売品
編集・発行 ： 富士ゼロックス小林節太郎記念基金
〒107-0052 東京都港区赤坂9丁目7番3号
電話 03-6271-4368
Printed in Japan