Economic Integration of Yunnan with the Greater Mekong Subregion

YUNNAN AND THE GREATER MEKONG SUBREGION
[Asian Economic Journal 1998, Vol. 12 No. 3]
Asian Economic Journal 2006, Vol. 20 No. 3, 303–317
303
Economic Integration of Yunnan with the
Greater Mekong Subregion*
Sandra Poncet
Received 16 June 2004; accepted 5 May 2006
This paper examines the process of economic integration between the Chinese
province of Yunnan and its riparian areas of the Mekong region. The gravity
model of trade is used to investigate the evolution of Yunnan’s international
trade integration between 1988 and 1999. Although Greater Mekong Subregion
cooperation efforts have had a positive effect on trade, trade has progressively
decreased from an above-standard level to a normal level, according to the gravity
model of trade. During this process, Yunnan’s trade has increased with other
countries such as Singapore, Indonesia and Malaysia. This evolution is in line
with Yunnan’s development and indicates a progressive re-orientation of its trade
toward more developed partners. The results suggest that the Mekong cooperation
project has to broaden its perspective, taking into consideration Yunnan’s expand-
ing trade relations with countries outside the Greater Mekong Subregion.
Keywords: economic integration, gravity model, Greater Mekong Subregion,
Yunnan.
JEL classification codes: F14, F15.
I.
Introduction
Yunnan is China’s seventh biggest province. This southwestern region’s neighbors
to the east are Guizhou and Guangxi, with Sichuan and Tibet to the north. It
also shares international borders spanning 8800 km with Myanmar to the west
and Laos and Vietnam to the south-east. A landlocked province, it is often con-
sidered to be inaccessible and backward, especially given that its terrain is
94 percent mountainous. Yunnan’s apparent lack of economic potential is rein-
forced by China’s strategy of international opening that had long been limited to
the coastal region. Although noted by D’Hooghe (1994) as having a long tradi-
tion of foreign relations and trade with its neighboring countries, until the 1980s
Yunnan remained a backwater province in an isolated Communist China. Border
trade between Yunnan and Southeast Asia revived with China’s 1978 open door
policy and especially after the improvement of diplomatic relations in 1984,
following decades of unsettled relations.
* Centre d’Economic de la Sorbonne, Université Paris 1 Panthéon-Sorbonne and CEPII, Bureau
405, 106 boulevard de l’hospital, Paris 75667 Cédex 13, France. Email: sandra.poncet@univ-paris1.fr.
I would like to thank two anonymous referees and the participants of the Third Euroseas Conference
for helpful comments on earlier versions. I am solely responsible for any errors that remain.
doi: 10.1111/j.1467-8381.2006.00237.x

---

ASIAN ECONOMIC JOURNAL
304
Since the mid-1990s, Yunnan has emerged as an international gateway to the
dynamic economies of Southeast Asia. The Yunnan Government has recently
highlighted export-oriented economic strategy as a priority. The province has
developed border trade with neighboring countries; namely, Myanmar, Laos and
Vietnam. It has also promoted relations with other countries of the region, such
as Thailand. Yunnan has played an active role in the Golden Quadrangle (com-
prising Yunnan, Myanmar, Laos and Thailand), now engulfed by the Greater
Mekong Subregion (GMS). The GMS comprises Yunnan and the 5 riparian
countries of the Mekong River: Myanmar, Laos, Thailand, Vietnam and Cam-
bodia. The GMS Economic Cooperation Program was initiated in 1992, initiated
by the Asian Development Bank, and aims at promoting economic cooperation
in terms of trade, foreign direct investment and tourism, linking countries that
already share common borders, natural resources, and a long history.
The present paper examines Yunnan’s integration with the GMS over the
period 1988–1999. The gravity model of trade reveals a high degree of trade
integration between Yunnan and its neighbors (Laos, Myanmar and Vietnam).
Moreover, the evolution of trade flows seems to suggest that GMS cooperation
will encourage broader international integration. Results seem to follow this
trend, emphasizing the rapid increase of imports with outside countries such as
Japan, the USA and even Singapore.
This paper proceeds as follows. Section II reviews the trade history and trade
pattern of the region. The gravity model is presented in section III, followed by
a description of the data in section IV. Section V conveys the results of an
empirical analysis of Yunnan’s international trade. The final section provides
concluding remarks.
II.
Yunnan’s Trade History and Trade Pattern
Despite deep cultural and historical similarities, economic cooperation and trade
between Yunnan and its riparian countries had been obliterated by difficult
diplomatic relations. Since the 1990s, the Yunnan Government has given priority
to the promotion of trade with its 5 riparian countries of the Mekong Region.
The Mekong River, also the ‘mother of waters’ or the Lancang, flows 4200 km
from the mountains of Central China, through Yunnan, Laos, Thailand, Cambodia
and Vietnam, before merging into the South China Sea (see Figure 1).
As Than (1997) points out, ‘the history of the Mekong is the history of the
relationships among these countries’. Before border lines were imposed by French
and British colonizers, the Mekong served as the boundary line between the
people of the federated Tai Kingdoms.1 The Tai confederation was split into
1.
Four Tai Kingdoms, each with its own supreme overlord, operated together in a ‘Tacit Alliance’
(Berman, 1998): Kengtung Kingdom of today’s Burmese Shan State, Lan Zhang Kingdom in current
north-western Laos, Lan Na Kingdom in today’s northern Thailand and Sipsong Panna in today’s
Yunnan.

---

YUNNAN AND THE GREATER MEKONG SUBREGION
305
Figure 1
Greater Mekong Subregion
5 different nations: China, Thailand, Burma, Laos and Vietnam. Culturally and
linguistically, Yunnan (and especially the Xishuangbanna Autonomous Prefec-
ture) is closely related with its neighbors through Tai and minority groups strad-
dling national boundaries and, increasingly, moving between national territories
(Hinton, 1998). It is only since the mid-1980s and the normalization of Chinese
relations with its Southeast Asian neighbors that Yunnan has started working to
establish subregional economic cooperation.
Yunnan’s regional engagement genuinely started in 1991 when China agreed
to join and establish a network called the Mekong Development Research
Network, initiated by the Canadian International Development Research Centre.
This network brought together Yunnan and its 5 riparian countries of the Mekong
River. Economic cooperation was promoted further through the GMS Economic
Cooperation Program. Tables 1 and 2 outline the evolution of Yunnan’s engage-
ment in international markets. Annual rates of export and import growth rose
above the Chinese average after 1992. The province’s imports grew 2.4-fold
over the 1992–1997 period compared with a 1.4-fold increase in the rest of
China’s imports over the same period. Export evolution in Yunnan converges to
that of other provinces after 1992 (with a 31 percent average annual growth rate
between 1993 and 1997), whereas before it was well below them.

---

ASIAN ECONOMIC JOURNAL
306
Table 1
Yearly evolution of trade in Yunnan and the rest of China
1988 –1992
1993 –1997
1998 –2002
(%)
(%)
(%)
Yunnan’s exports
9.1
31.0
5.5
Rest of China’s exports
19.8
24.8
19.4
Yunnan’s imports
25.0
35.3
2.3
Rest of China’s imports
11.4
9.2
27.8
Note: Average annual growth rates over the sub-periods.
Sources: Chinese State Statistical Bureau (various years) and Yunnan Province Statistical Bureau
(various years).
Table 2
Shares of exports and imports in GDP
Yunnan (%)
National average (%)
Exports
1988
4.2
11.8
1992
5.8
17.8
1995
9.2
21.3
1999
4.6
19.3
2002
5.3
26.0
Imports
1988
1.3
13.8
1992
2.8
16.7
1995
5.5
18.9
1999
2.8
16.8
2002
3.0
23.6
Sources: Chinese State Statistical Bureau (various years) and
Yunnan Province Statistical Bureau (various years).
When openness is measured by the share of exports and imports in GDP,
Yunnan remains quite closed compared to the national average. The export to
GDP ratio increased from 5.8 percent (approximately one-third of the national
average) in 1992 to 9.2 percent (approximately half the national average) in
1995 before falling back to 4.6 percent in 1999, which might be largely attrib-
uted to the Asian crisis. It reached 5.3 percent in 2002, only one-fifth of the
national average (Table 2).
Despite its lower openness ratio, international trade is an especially strategic
engine for growth in Yunnan because it is deeply landlocked. Gallup et al.
(1999) highlight three reasons why landlocked countries might be at a disadvant-
age: cross-border migration of labor is more difficult than internal migration;
infrastructure development is more challenging to coordinate across borders
than domestically; and, coastal economies might have military or economic
incentives to impose costs on landlocked countries. Yunnan’s prospects and

---

YUNNAN AND THE GREATER MEKONG SUBREGION
307
Table 3
Relative Importance of Greater Mekong Subregion countries in
Yunnan’s Exports, 2002 (% of total)
Yunnan
Rest of China
Ratio (Yunnan/
rest of China)
Vietnam
9.4
0.6
15.1
Thailand
2.6
0.9
2.9
Laos
0.7
0.0
54.5
Cambodia
0.1
0.1
1.1
Myanmar
20.7
0.1
156.6
Japan
9.1
14.9
0.6
Hong Kong
17.0
18.0
0.9
USA
5.0
21.6
0.2
Sources: Chinese State Statistical Bureau (various years); Yunnan Province Statistical Bureau (vari-
ous years).
opportunities could, however, be on its borders. The long borderline that it
shares with its neighbors could help the province overcome its geographic
difficulties through the shared historical and cultural ties mentioned above.
Yunnan’s entry in the GMS agreement is consistent with the argument made
by MacKellar et al. (2000) that regional trade arrangements significantly mitig-
ate the negative effects of being landlocked on trade. The privileged ties that
link Yunnan and its neighbors appear regularly in the international trade profile
of the province. In Table 3, data from 2002 comparing Yunnan with the rest of
China highlights that the riparian countries of Yunnan are overrepresented in
trade relations.
Trade between Yunnan and Myanmar, Laos and Vietnam is significantly greater
than trade between those countries and other Chinese provinces (ratios reported
in the last column of Table 3 equal 157, 55 and 15 times, respectively). Myanmar
and Laos export primary goods, such as agricultural products (e.g. rice, corn
and seafood), timber and minerals (e.g. jade and rubies), and import mainly
low-priced consumer goods and machinery, such as electrical appliances and
equipment from Yunnan. D’Hooghe (1994) stresses that these countries are the
main export outlets for Yunnan’s products, which are not yet suitable for the
world market because of packaging and design.
The magnitude of these statistics justifies the need to quantify the impact of
Yunnan’s economic integration through the GMS initiative, which can be
estimated by a gravity equation. The next section presents the gravity model of
trade that is used to investigate the evolution of Yunnan’s trade integration with
the GMS.
III.
The Methodology
The gravity model is the most commonly used analytical framework to examine
bilateral trade. It is largely inspired by the ‘Law of Universal Gravitation’

---

ASIAN ECONOMIC JOURNAL
308
proposed by Newton in 1687. It holds that the attractive force between two
objects i and j is a positive function of their respective masses (Y and Y ) and a
i
j
negative function of the distance (D ) between them. Therefore, the attraction
ij
force, T , is given by:
ij
Y Y
T = G i j

,
(1)
ij
2

Dij
where G is the gravitational constant depending on the units of measurement.
Tinbergen (1962) proposes that roughly the same functional form could be
applied to international trade flows. The trade flow from origin i to destination j,
Y Y
α β
T , is expressed as T = G i j

, with Y and Y representing the relevant economic
ij
ij
i
j
Dθ

ij
sizes of the two locations and D the distance between them. Taken from Equa-
ij
tion (1), the expected unit production elasticities of trade are α = β = 1 and the
anticipated distance elasticity θ is 2.2 Despite the absence of theoretical founda-
tions until the late 1970s, the gravity model turned out to be a successful and
robust approach to explaining international trade. Some of the important steps
towards its formalization include Anderson (1979) showing that the gravity
model can be derived from expenditure share equations (assuming commodities
are distinguished by place of production), followed by Helpman’s (1984) and
Bergstrand’s (1985) demonstration that the gravity model can be derived from
models of trade in differentiated products. Deardorff (1998) later establishes that
the gravity model is consistent with the Heckscher–Ohlin model that includes
transport costs. As Helliwell (1998) notes, the gravity model has gone from
being a theoretical orphan to being the favored child of all main theories of
international trade.
Therefore, the gravity equation of trade explains aggregate bilateral flows, T ,
ij
between 2 countries, i and j, by including both attractive and repulsive elements.
The attraction aspect corresponds to the importer’s and the exporter’s economic
sizes (both proxied by GDP). The repulsion component is the geographical
distance: a proxy for trade costs, including transport and transaction costs.
A linear relationship between trade flows, economic sizes and distance can be
obtained by taking the natural logarithm of the gravity expression. The equation
to be estimated, in its most basic form, is:
ln T = c + α ln (Y Y ) − θ ln D + ε ,
(2)
ij
i
j
ij
ij
where c is a constant term, T denotes trade flows between country i and country
ij
j, Y and Y are, respectively, the GDP of the importing and the exporting coun-
i
j
tries, D is the distance between the capital cities of the 2 trading partners, and ε
ij
ij
is the disturbance term. As explained by Head (2000), the inclusion of the error
term, ε , delivers an equation that can be estimated by OLS.
ij
2.
The value of 2 is derived from the Newtonian formula.

---

YUNNAN AND THE GREATER MEKONG SUBREGION
309
The successful empirical performance of the gravity model is well known and
is commonly invoked to assess trade patterns both between countries and within
preferential trade areas. In particular, it makes the model a useful tool to analyze
the evolution of regional trade integration. The intensity of non-standard trade
relations is measured by dummy variables for specific partners. A positive and
statistically significant coefficient for a dummy variable implies that trade flows
exceed the normal level; that is, the level predicted by the countries’ economic
sizes and the distance between them. Therefore, it describes preferential trade
relations. By contrast, a negative and statistically significant coefficient implies
that the trade flows fall short of the predicted level.
Using this method to illustrate Yunnan’s international trade, the effect of
GMS cooperation on trade flows is assessed by introducing a dummy equal to
one if the partner is located in the GMS and zero otherwise. The positive and
significant coefficient of the dummy variable underlines above-standard trade
volume (therefore, greater economic integration) between Yunnan and the other
riparian countries of the Mekong. This attests to the effectiveness of preferential
trade in boosting regional trade. Furthermore, countries of the GMS are differen-
tiated to identify the privileged trade partners of Yunnan. Yunnan’s trade rela-
tions with others countries are also investigated, including ASEAN members not
located in the GMS, Japan and the USA.
The time dimension of the dataset makes it possible to assess the evolution of
Yunnan’s economic integration over the years. The inclusion of country dum-
mies by year underlines the re-orientation process of Yunnan’s trade. Imports
and exports are analyzed separately.
IV.
Description of the Data
The data presents bilateral trade flows between Yunnan and approximately 80 of
its international trade partners between 1988 and 1999. Table 4 shows the list of
countries included in the regression. The countries for which Yunnan either does
not declare trade flows or with which Yunnan does not trade are both indicated
by a missing value. The endogeneity of which country pairs have positive trade
has the potential to generate selection bias. The Heckman two-stage procedure
will address this matter.
The dataset covers 230 non-zero observations. All data on trade flows and
GDP for Yunnan are extracted from the Chinese Ministry of Foreign Economic
Relations and Trade (various years) and the Yunnan Province Statistical Bureau
(various years). It is necessary to acknowledge that these statistics understate the
value of total transactions. Indeed, they include neither the small-scale free trade
between border residents nor informal border trade (smuggling). Despite the
process of trade liberalization, smuggling (mainly of drugs, arms, automobiles,
motorcycles, electrical and electronic equipment) is recognized by local govern-
ments as a means to circumvent prohibition, high duties and red tape (Hendrischke,
2000; Hinton, 1998). The potential understating of economic transitions’ value

---

ASIAN ECONOMIC JOURNAL
310
Table 4
List of countries covered in the trade flow dataset
Afghanistan
Mexico
Algeria
Morocco
Argentina
Myanmar
Australia
Nepal
Austria
Netherlands
Bangladesh
New Zealand
Belgium
Niger
Brazil
Nigeria
Brunei
North Korea
Bulgaria
Norway
Canada
Oman
Chile
Pakistan
Cuba
Panama
Czech Republic
Papua New Guinea
Denmark
Paraguay
Dominica Republic
Peru
Egypt
Philippines
El Salvador
Poland
Ethiopia
Portugal
Fiji
Romania
Finland
Russia
France
Saudi Arabia
Germany
Singapore
Hong Kong
South Africa
Hungary
South Korea
India
Spain
Indonesia
Sri Lanka
Iran
Sweden
Ireland
Switzerland
Israel
Syria
Italy
Taiwan
Japan
Thailand
Jordan
Togo
Kazakhstan
Turkey
Kenya
Ukraine
Kuwait
United Arab Emirates
Laos
UK
Lebanon
USA
Macao
Uruguay
Malaysia
Venezuela
Malta
Vietnam
Mauritius
Yugoslavia
might lead to underestimating the impact of Yunnan’s economic integration
with the GMS. However, smuggling practices as a proportion of total trade are
expected to decline in the face of growing trade liberalization and deepening
regional integration. On the one hand, findings of the positive and statistically
significant effect of the GMS cooperation efforts on transaction volume would

---

YUNNAN AND THE GREATER MEKONG SUBREGION
311
likely be more robust if smuggling practices had also been included. On the
other hand, a reduction in regional trade flows from the above-standard level to
the normal level would not be moderated.
No statistics for trade between Cambodia and Yunnan are available, prevent-
ing the assessment of their level of economic integration. Data on GDP and
population of most countries comes from the World Bank’s World Development
Indicators (various years). Missing data for some Asian countries (Taiwan,
Myanmar) is taken from the Asian Development Bank’s Key Indicators of
Developing Asian and Pacific Countries (various years). The distance between
Yunnan and each trading partner is measured using the ‘greater circle’ distance
formula between Kunming (the capital of Yunnan) and the capital city of each
country. This formula approximates the shape of the earth as a sphere and
calculates the minimum distance along its surface. As emphasized in various
studies (e.g. Head and Mayer, 2000), ‘greater circle distance’ is an effective
measurement tool in an international context. One should bear in mind that the
distance term might not reflect real transaction costs, as some other studies have
suggested. Most studies have difficulties estimating real transport costs between
cities; using the distance variable can help avoid these obstacles. It should be
emphasized that the introduction of a term expressing economic size of part-
ner countries allows the existing heterogeneity between trade partners to be
taken into account (e.g. how Laos is a relatively small economy compared to
Vietnam). Moreover, results are robust when using Ho Chi Minh instead of
Hanoi as the city of reference when computing the distance between Yunnan
and Vietnam.3
V.
The Findings
Equation (2) is the simplest form of the gravity equation. However, there is
reason to believe that GDP per capita has a positive effect on trade flows of
a given size. As argued by Frankel et al. (1995), as countries become more
developed they tend to specialize more and trade more. Consequently, in using
the gravity model to describe trade flows between Yunnan and its trade partners,
joint per capita income of partners is included in addition to Y , Y , D and
i
j
ij
various dummy variables. Therefore, the following equation is estimated:
ln T = c + α ln (Y Y ) + β ln [(Y /pop )(Y /pop )] − θ ln D + ε ,
(3)
kj
k
j
k
k
j
j
kj
kj
where subscripts k and j denote Yunnan and its trading partner j, respectively.
Equation (3) is estimated separately for imports and exports on pooled
cross-section and time-series panel data. Yearly fixed effects are introduced
to appropriately take into account the panel nature of the dataset. As a pre-
liminary measure, estimations are verified to be robust given a potential selec-
tion bias based on the Heckman selection model (full maximum-likelihood)
3.
Results not reported in the present paper are available from the author upon request.

---

ASIAN ECONOMIC JOURNAL
312
procedure to account for the existence of missing trade flows.4 The results are
presented in Table 5. Heteroskedasticity is accounted for by applying White’s
(1980) method. Columns 1–4 of the table present the impact of integration on
Yunnan’s exports and columns 5–8 show the impact on its imports. Results in
columns 2 and 6 rely on the Heckman selection model. The lack of significance
of the inverse of the Mills ratio suggests that our results do not suffer from a
selection bias.
The gravity model gives a reasonably good explanation of Yunnan’s trade
patterns, evidenced by the relatively high values of R2 (approximately 70 percent
for exports and 30 percent for imports). The two basic explanatory variables
(GDP and distance) have the expected signs and are statistically significant at
the 1-percent level. The coefficient of joint GDP is reasonably close to 1 while
the coefficient of distance is negative and close to its predicted value of −2. The
coefficient on joint GDP per capita, however, is not statistically significant. One
possible explanation is that Yunnan’s trade is largely of an inter-industry rather
than an intra-industry type.5
Baseline estimations appear in columns 1 and 5 (exports and imports, respec-
tively). In columns 3 and 7, a dummy variable that takes the value of one when
the partner is Myanmar, Laos, Vietnam or Thailand is introduced to measure the
impact of GMS cooperation. It turns out that the coefficient of this dummy
variable is positive for both export and import flows. In the latter case it is
significant at the 10-percent level, underlining the above-standard import trade
flows between Yunnan and these 4 countries. This result illustrates one of the
positive effects of the regional collaboration. Yunnan’s imports from the 4
neighbors are 2.45 times greater than the normal level [exp(0.90) = 2.45] pre-
dicted by the gravity model.
To obtain more detailed results, the aggregate Mekong region is decomposed
into four terms for the individual countries (columns 4 and 8). Examining the
estimated coefficients of the dummy variables for the 4 riparian countries of the
Mekong River, it is found that Yunnan’s trade with Myanmar (both for exports
and imports) and Laos (for imports) is significantly greater than the normal
level, whereas that with Thailand is significantly lower. The insignificant coeffi-
cient of the Vietnam dummy indicates that Yunnan’s trade with Vietnam is in
line with the relative size of and distance between the 2 partners.
These regression results indicate that on average, over the period 1988–1999,
Yunnan had close and privileged trade relations with at least 2 of its neighboring
countries (Myanmar and Laos), confirming the successful influence of this case
of regional cooperation. Trade integration in terms of Yunnan’s imports from
Laos and Myanmar appears slightly greater than its exports to these countries.
4.
The unbalanced sample can be subject to a non-ignorable selection rule; that is, the probability
of a partner country being included in the sample is not independent of model error and, in particu-
lar, to the yearly specific effects.
5.
This explanation is consistent with that of Kawai and Urata (1998).

---