Full text of DOI:10.1111/1467-9361.00097

Review of Development Economics, 4(3), 307–325, 2000
Effects of Primary Exports on Industrial Exports
and GDP: Empirical Evidence
Zhenhui Xu*
Abstract
This paper analyzes the effects of the growth of nonfuel primary exports on the growth of industrial exports
and GDP in 74 economies between 1965 and 1992. There is clear evidence of positive effects, both in the
short term and in the long term, of the growth of primary exports on the growth of industrial exports and
GDP in more than two-thirds of the economies. Therefore, governments in developing countries should not
discriminate against the export of primary products, as some earlier studies suggest. Instead, they should
adhere to policies that aim at export promotion.
1. Introduction
International trade theory and development theory both suggest that export growth
contributes positively to economic growth. Since exports are a component of GDP,
export growth contributes directly to GDP growth. In addition, export growth is widely
believed to promote GDP growth indirectly through linkage effects, foreign exchange
earnings, and its positive externalities on the domestic economy (Hirschman, 1958;
McKinnon, 1964; Chenery and Strout, 1966; Balassa, 1978; Krueger, 1980; Helpman and
Krugman, 1985; Grossman and Helpman, 1993). With the development of time-series
econometrics, empirical support for the export-led growth hypothesis is growing
(Onsfowora et al., 1996; Xu, 1996, 1998).
Although conventional wisdom maintains that export growth promotes economic
growth, a diversity of views exists concerning the effects of different components of
exports on economic growth. Exports may include (1) nonprocessed primary products,
(2) processed primary products, and/or (3) manufactured products. Some studies seem
to suggest that the positive effects of exports on economic growth can exist only for
exports of manufactured products and processed primary products, but not for non-
processed primary products and low-level processed primary products. The following
quote states this view unmistakably:
Indeed, despite exporting primary goods for over a hundred years, a sig-
nificant number of countries in Latin America, Africa and South Asia have
remained poor. An important distinction between a success story such as
South Korea and a failure such as Honduras is that the former has grown
through exporting manufactured goods to the developed countries, whereas
the latter has remained backward while exporting primary goods such as
bananas and coffee. (Eswaran and Kotwal, 1993, p. 163)
The effect of primary export growth (PEG hereafter) on economic growth is indeed
critical because of important theoretical and policy implications. Primary products
remain major exports for most developing countries today. Even in those developing
*
5
Xu: Georgia College & State University, Milledgeville, GA 31061, USA.Tel: (912) 445-2592; Fax: (912) 445-
249; E-mail: zxu@mail.gcsu.edu. I thank Berhanu Abegaz, Carl Moody, John Whitaker, and the anonymous
referee for helpful comments. All remaining errors are entirely my responsibility.
©
Blackwell Publishers Ltd 2000, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA

3
08 Zhenhui Xu
countries which have successfully increased their exports of manufactured goods in the
past several decades, primary exports still account for more than 50% of their total
merchandise exports (see Table 1). If the growth of primary exports has insignificant
or negative effects on economic growth for the majority of developing countries, then
it is necessary to re-examine the validity of the comparative advantage theory since it
would not hold for developing countries with a comparative advantage in primary
exports. It would also call for a different prescription for economic development in
those countries.
The process of industrialization will undoubtedly require a developing country to
increase its export of manufactured goods over time. The question then becomes how
the developing country gets to the point of exporting manufactured goods. Economists
and policymakers once believed that economic development could be reached more
rapidly by following the growth of primary exports. Perhaps the best-known economic
argument for PEG is the linkage effect (Hirschman, 1958). The growth of the primary
export sector stimulates the growth of other industries through its demand for domes-
tic inputs and services, raising domestic workers’ income and thus their consumption
demand. In addition, primary exports can create opportunities for formerly unem-
ployed and underemployed resources to produce greater output for export to foreign
markets; i.e., as a vent for surplus. The expansion of potential markets for primary
exports can lead to expanded supplies of foreign investment, domestic saving, labor,
and skilled manpower to complement the fixed factors of production. The argument
that exports have positive externalities on the domestic economy can apply to PEG as
well. PEG can stimulate manufacturing and hence lead to the exportation of manu-
factured goods and, ultimately, economic development.
On the other hand, there are also concerns about the adverse effects of PEG, espe-
cially its long-term effects, on economic growth. These concerns focus on the two pos-
sible channels through which PEG can affect economic growth negatively. The first
channel is the terms-of-trade effect on primary exports, known as the Prebisch–Singer
hypothesis. Many economists have tested the Prebisch–Singer hypothesis and found
1
conflicting results. In a recent study, Cuddington and Wei (1992) used improved data
and methods to measure the net barter terms of trade for 24 nonfuel primary com-
modities between 1900 and 1988. Other than the sharp fall of prices in 1920, they found
no evidence of any statistically significant downward trend.The second channel relates
to the degree of instability in primary product markets (MacBean, 1966). The insta-
bility in primary export markets can cause wide variations in export earnings and
domestic investment, thereby affecting GDP growth adversely.
Historically, almost all economies have attempted to follow the path of development
through PEG. Some have succeeded, while others have not. Among those who have
succeeded, the US, Canada, Australia, and Denmark all achieved industrialization by
following the vent-for-surplus path at least partly. Although others have not yet indus-
trialized, some have shown significant structural changes as a result of their primary
exports. In a recent study, the World Bank (1993) noted that many Asian economies
proceeded to export manufactured goods as a result of their primary exports. Their
export-promotion efforts resulted in improved price and incentive structures, financial
systems, and fiscal environment in these economies, and consequently attracted foreign
investment. Such structural changes are important signs of economic development
(Choi and Beladi, 1996).
The above discussion indicates that PEG alone is not sufficient for economic growth.
It is true that at the very least, the revenues from primary exports can relax the bind-
ing foreign exchange constraint to allow increases in imports of capital goods and
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 309
Table 1. Share of Nonfuel Primary Exports in Total Merchandise Exports
965 1970 1975 1980 1985 1990 1992 (A) (B)
36.2
1
(C)
Sierra Leone
Central African Rep.
Solomon Islands
Somalia
Zimbabwe
Belize
Niger
Tanzania
Liberia
36.2 39.8 36.0 63.8 63.8 63.8 1.76 46.1 0.98
55.5 76.2 73.9 65.2 56.5 56.5 1.26 61.9 1.00
96.5 92.7 98.7 97.0 96.5 96.2 1.05 95.4 1.00
93.8 96.8 94.4 97.1 97.1 99.4 1.05 96.5 0.99
64.3 64.3 64.3 67.9 67.3 67.3 1.05 65.5 0.99
82.0 80.1 82.3 75.1 84.7 81.8 1.03 80.8 1.00
96.3 91.3 96.7 98.5 98.5 98.5 1.03 96.5 1.00
80.6 82.3 81.0 80.0 81.4 81.4 1.02 81.1 0.94
97.3 98.1 95.4 98.2 98.2 98.7 1.02 97.7 1.00
99.3 95.7 98.7 98.2 98.2 98.2 1.00 97.2 0.98
80.4 84.7 76.4 80.5 80.5 80.5 1.00 80.6 1.00
50.2 50.8 18.2 50.2 50.2 50.2 1.00 45.7 0.97
93.0 91.8 91.5 91.5 91.5 91.7 1.00 91.7 0.98
97.7 97.2 97.7 97.7 97.7 97.7 0.99 97.8 1.00
44.7
91.8
92.4
64.3
82.0
95.7
79.4
96.4
98.0
80.4
50.2
91.5
98.3
Ghana
Dominican Rep.
South Africa
Zaïre
Burundi
Gambia
100.0 100.0 99.9 99.4 99.5 99.4 99.5 0.99 99.6 1.00
Malawi
Guyana
Paraguay
Zambia
Chad
97.2
96.4
91.6
99.8
93.1
99.8
94.5
98.8
97.0
99.6
95.6
95.1
98.3
95.5
97.0
95.9
94.7
99.8
13.1
86.0
80.7
94.7
72.6
25.7
84.2
80.3
94.2
81.7
97.9
21.2
94.2
96.7
93.7
41.0
45.1
96.3 94.7 93.3 95.0 96.2 96.2 0.99 96.1 1.00
96.9 97.4 86.4 94.9 94.9 90.7 0.98 93.8 1.00
91.1 88.3 88.2 94.5 90.1 84.3 0.98 89.9 1.00
99.8 99.0 97.9 97.9 97.9
—
0.98 98.7 0.99
95.2 84.0 90.2 90.3 90.3 90.3 0.97 91.0 0.96
99.6 96.1 98.2 96.6 96.6 96.6 0.97 97.6 0.98
83.8 82.6 83.7 91.1 91.3 91.7 0.97 88.0 0.99
99.1 95.4 95.4 95.4 95.4 91.9 0.97 95.8 1.00
89.3 88.4 98.7 92.9 92.9 91.8 0.96 93.7 1.00
97.3 95.9 94.5 94.5 94.5 96.0 0.95 95.2 0.99
88.3 85.1 87.0 95.3 89.8 86.3 0.94 90.9 0.99
94.4 94.2 63.5 87.0 89.1 89.1 0.94 88.9 0.97
96.5 94.6 91.6 90.1 91.4 93.2 0.93 93.2 1.00
95.6 93.6 89.1 87.9 87.9 87.8 0.92 91.0 1.00
97.1 98.2 98.2 96.9 88.7 87.8 0.91 95.4 1.00
95.5 89.1 89.0 91.9 86.8 84.4 0.91 91.3 0.99
88.9 87.3 87.7 85.8 84.8 79.7 0.90 87.4 0.96
97.1 94.0 92.4 89.1 88.5 94.2 0.89 94.9 0.97
14.1 12.8 13.3 11.0 11.2 11.5 0.86 12.6 0.92
72.0 75.1 74.8 77.8 73.4 68.1 0.85 75.2 0.99
80.7 80.7 45.7 64.7 68.2 72.2 0.84 73.5 1.00
93.4 82.5 88.3 80.0 78.8 78.8 0.83 85.4 0.94
74.4 60.2 58.2 58.2 58.2 55.2 0.80 63.0 1.00
23.2 21.0 22.3 22.0 20.5 19.5 0.80 22.6 0.97
79.9 73.7 65.1 74.6 64.6 72.4 0.77 74.1 0.99
79.6 69.6 61.8 64.5 60.9 58.8 0.76 68.5 1.00
88.5 85.0 78.7 74.4 71.1 70.9 0.76 81.9 0.99
70.9 71.8 61.8 71.8 60.7 59.0 0.74 68.9 0.99
97.8 95.4 62.5 64.1 71.6 72.4 0.73 81.6 0.91
21.1 16.9 17.5 16.9 15.3 15.8 0.72 18.1 0.95
89.4 92.7 87.9 66.8 66.8 66.8 0.71 80.1 0.98
78.2 78.0 66.1 80.0 65.2 65.2 0.67 74.7 0.90
85.7 75.1 73.4 71.3 63.0 65.0 0.67 76.1 0.97
73.4 70.0 47.3 16.0 27.5 41.0 0.67 45.8 0.97
45.1 44.1 35.4 41.5 29.4 42.4 0.65 41.4 0.98
Sudan
Nicaragua
Mauritania
Mali
Myanmar
Honduras
Togo
Iceland
Burkina Faso
Papua New Guinea
Chile
Madagascar
Ethiopia
United Kingdom
Guatemala
Dominica
Côte d’Ivoire
Haiti
France
Costa Rica
Uruguay
New Zealand
El Salvador
Peru
Belgium
Benin
Senegal
Argentina
Barbados
Jamaica
©
Blackwell Publishers Ltd 2000

3
10 Zhenhui Xu
Table 1. (Cont.)
1
965
1970
1975 1980 1985 1990 1992 (A) (B)
(C)
Fiji
98.0
31.1
56.4
21.5
50.1
91.6
94.1
58.2
86.7
36.9
99.1
31.8
32.2
65.2
42.9
96.6
56.5
96.9
63.0
93.6
96.4
39.5
32.9
65.0
97.7 97.7 98.2 91.3 63.1 62.0 0.64 91.3 1.00
26.1 27.2 28.5 20.5 18.6 16.9 0.60 24.9 0.96
43.5 40.2 40.2 36.2 31.8 30.3 0.56 40.7 0.97
United States
Denmark
Austria
India
17.3 12.6 15.6 12.4 11.2
9.7
0.52 14.7 0.98
47.2 54.1 40.7 35.8 24.8 26.3 0.50 39.4 0.97
85.3 70.3 59.6 48.8 44.7 40.6 0.49 65.0 0.98
89.9 86.6 71.6 55.7 44.2 41.7 0.47 71.0 0.98
42.7 36.9 37.2 25.9 26.3 25.2 0.45 36.0 0.90
64.4 44.9 37.0 38.6 38.6 41.3 0.45 51.2 0.93
33.8 27.3 22.7 18.3 16.3 14.2 0.44 25.2 0.97
98.6 88.6 66.0 58.3 41.7 27.5 0.42 73.8 0.96
24.2 20.2 16.6 14.8 12.8 11.8 0.40 19.2 0.98
29.5 23.4 17.8 16.7 12.6 10.0 0.39 20.7 1.00
57.8 52.3 41.4 29.0 25.2 26.9 0.39 43.1 0.99
32.5 22.4 25.9 18.7 15.6 14.5 0.36 24.8 0.98
91.7 82.5 71.9 59.8 34.8 32.3 0.36 71.2 0.99
41.0 26.6 24.5 20.3 19.3 19.2 0.34 29.5 0.95
98.2 88.6 72.6 47.4 32.6 31.5 0.34 74.2 1.00
41.5 43.9 44.4 35.8 19.8 19.8 0.31 38.2 0.98
90.8 80.6 62.5 42.5 29.0 24.6 0.31 64.3 0.99
90.4 74.1 71.7 34.1 29.8 27.0 0.31 64.3 0.98
Brazil
Morocco
Canada
Greece
Portugal
Sri Lanka
Sweden
Israel
Ireland
Finland
Thailand
Spain
Mauritius
Pakistan
Philippines
Turkey
Korea, Rep.
Malta
22.4 16.4
21.4 10.9
64.9 62.2 69.5 40.9
9.9
6.4
5.6
7.9
5.4
3.8
6.1
4.9 0.14 13.9 0.99
4.3 0.11 12.6 1.00
6.1 0.09 51.4 1.00
Nepal
Note: The values under (A) are calculated by dividing the share of nonfuel primary exports in total mer-
chandise exports in 1990 by the share in 1965.The values under (B) are the average shares of nonfuel primary
exports in total merchandise exports between 1965 and 1992. The values under (C) are the average shares
of nonfuel primary and industrial exports in total merchandise.The countries are sorted based on the values
in column (A).
intermediate goods (McKinnon, 1964; Chenery and Strout, 1966). It is also true that to
developing countries, the ability to import capital goods and intermediate goods is a
critical determinant of whether the transition from primary exports to manufacturing
exports can be successful. Yet such ability cannot be realized unless there are govern-
ment policies and economic environment compatible to PEG. PEG would help build
a base for industrial production and export diversification only when an economy can
effectively use its foreign exchange earnings from primary exports.
The aforementioned studies have raised two interesting questions concerning the
effects of PEG in promoting economic growth: (1) Does PEG have any positive effects
on the growth of industrial exports and GDP? (2) If so, are such effects sustainable in
the long-term? The answer to these questions requires empirical evidence both over
time and within a specific institutional arrangement. Despite a large literature in this
area, such empirical evidence is not currently available.
In this paper, I adopt a multivariate vector-autoregression (VAR) framework to
investigate the effects of PEG on the growth of industrial exports and GDP for a group
of 74 economies. The focus on a multivariate dynamic analysis provides us with the
2
following advantages over the existing studies in the literature. First, it allows for
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 311
different economic and institutional structures for each country. Second, it can deal
with the simultaneity problem among PEG, industrial exports, and GDP. Third, the
multivariate VAR approach in this paper is consistent with the previous literature’s
focus on causality as it incorporates the possible existence of short-term relations
between PEG and the other two variables. In addition, it permits us to investigate
the long-term cumulative effects of PEG on the growth of industrial exports and GDP
by allowing for interactions among variables, both the contemporaneous correlation
and the dynamic feedback. With the VAR framework, we can test the following
three hypotheses: (1) PEG does not affect the growth of industrial exports and
GDP in the short-run; (2) there is no long-term sustainable effect of PEG on the
growth of industrial exports and GDP; and (3) the effects of PEG on the growth of
industrial exports and GDP are universal to all countries whose main exports are
primary products.
The results of this paper show that none of the above three hypotheses can be
supported. There is clear evidence of positive effects, both in the short term and in
the long term, of PEG on the growth of industrial exports and GDP for more than
two-thirds of the 74 countries in the sample. Furthermore, such positive effects are
enhanced in the long term in many countries, especially in developing countries. This
evidence supports the hypothesis that PEG has sustainable positive effects on the
growth of industrial exports and GDP. It is also interesting to note that the long-term
effects of PEG on the growth of industrial exports and GDP are completely different
from the short-term effects for several countries. While some countries turned their
short-term negative effects into positive ones in the long term, others showed the oppo-
site. These results suggest that the effects of PEG on the growth of industrial exports
and GDP are not universal for all economies. Country-specific economic, political, and
policy differences must play an important role in determining how PEG affects indus-
trial exports and economic growth in each country.
2. The Sample and Data
The terms “primary goods” and “industrial goods” used in this study stand for nonfuel
primary products and manufactures, respectively, as defined by the World Bank. The
nonfuel primary products comprise commodities in SITC revision 1, Sections 0, 1, 2, 4,
and Division 68, whereas manufactures include commodities in SITC revision 1, Sec-
tions 5 through 9 excluding Division 68. For more details, see “Glossary of Terms” in
the World Bank’s World Tables (1992).
To analyze the dynamic effects of the growth of nonfuel primary exports on the
growth of industrial exports and GDP, a sample of 74 economies was selected. The
sample period is 1965–92 for most economies, but 1965–87 for Liberia, 1965–90 for
Somalia, 1965–91 for Gambia, Malta, and Zambia, and 1967–92 for Solomon Islands.
Table 1 shows the profile of the share of nonfuel primary exports in total merchandise
exports for each economy between 1965 and 1992, while the sample period for each
country is reported in Table 2.
The 74 countries in Table 1 are sorted by the values under column A, the ratios of
the share of nonfuel primary exports in total merchandise exports in 1990 to the share
in 1965. These values provide us with important information about the direction in
which each economy has been going. Based on these values, the 74 economies can be
divided into three groups: the economies whose share of nonfuel primary exports rose
in the last three decades, the economies whose share remained approximately
unchanged, and the economies whose share fell.
©
Blackwell Publishers Ltd 2000

3
12 Zhenhui Xu
Table 2. Specification of VAR Models
Sample period
Deterministic components
Lags
Schwarz’s BIC
Sierra Leone
Central African Rep.
Solomon Islands
Somalia
Zimbabwe
Belize
Niger
Tanzania
Liberia
(1965–1992)
(1965–1992)
(1967–1992)
(1965–1990)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1987)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1991)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1991)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
Constant
1
1
1
2
1
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
1
2
1
1
1
1
1
1
1
2
1
-10.932
-10.777
-7.050
Constant and trend
Constant and trend
Constant
Constant and trend
Constant
-8.404
-15.905
-9.349
-8.877
None
Constant
-12.821
-15.189
-11.327
-10.498
-13.157
-11.526
-11.154
-8.694
Constant and trend
Constant
Constant and trend
Constant and trend
None
Ghana
Dominican Rep.
South Africa
Zaïre
Burundi
Gambia
Malawi
Guyana
Paraguay
Zambia
Chad
Constant and trend
Constant
Constant and trend
Constant and trend
Constant and trend
Constant and trend
None
-12.108
-10.822
-12.173
-13.419
-9.151
Sudan
Constant
-9.776
Nicaragua
Mauritania
Mali
Myanmar
Honduras
Togo
Constant and trend
Constant and trend
Constant
Constant and trend
Constant
Constant and trend
Constant and trend
Constant
Constant
Constant and trend
Constant
Constant
Constant
None
None
Constant and trend
Constant and trend
Constant and trend
Constant
Constant
Constant and trend
Constant
Constant and trend
Constant and trend
Constant
-10.531
-10.494
-7.615
-10.188
-12.669
-10.216
-13.796
-12.174
-10.387
-11.827
-11.983
-10.370
-18.418
-14.284
-9.259
-12.228
-13.521
-20.194
-14.896
-13.281
-16.443
-12.575
-11.225
-18.322
-10.232
-10.303
-12.491
-10.603
-13.138
Iceland
Burkina Faso
Papua New Guinea
Chile
Madagascar
Ethiopia
United Kingdom
Guatemala
Dominica
Côte d’Ivoire
Haiti
France
Costa Rica
Uruguay
New Zealand
El Salvador
Peru
Belgium
Benin
Senegal
Constant
Argentina
Barbados
Jamaica
Constant and trend
Constant and trend
None
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 313
Table 2. (Cont.)
Sample period
Deterministic components
Lags
Schwarz’s BIC
Fiji
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1992)
(1965–1991)
(1965–1992)
Constant and trend
Constant
Constant
Constant and trend
Constant
Constant and trend
Constant and trend
Constant and trend
Constant and trend
Constant and trend
Constant
Constant and trend
Constant and trend
Constant and trend
Constant and trend
Constant
Constant and trend
Constant
Constant and trend
Constant
Constant and trend
Constant and trend
Constant and trend
Constant and trend
1
1
1
1
1
3
1
1
1
1
1
1
2
1
3
1
1
1
1
1
1
1
3
1
-10.503
-16.922
-19.783
-19.176
-14.784
-15.203
-15.605
-16.349
-14.290
-16.626
-12.893
-19.836
-16.126
-16.868
-16.866
-14.834
-16.859
-10.051
-13.855
-14.524
-13.620
-14.905
-13.009
-12.004
United States
Denmark
Austria
India
Brazil
Morocco
Canada
Greece
Portugal
Sri Lanka
Sweden
Israel
Ireland
Finland
Thailand
Spain
Mauritius
Pakistan
Philippines
Turkey
Korea, Rep.
Malta
Nepal
The primary criterion for selecting these economies is data availability. Yet there
are two additional considerations. First, for a few economies such as Hong Kong,
Singapore, Switzerland, Japan, Bahamas, and Bermuda, primary exports in total mer-
chandise exports are trivial based on the values under column B, the average shares
of nonfuel primary exports in total merchandise exports for the sample period. Hence,
to reflect the importance of primary exports, the economies whose average share of
nonfuel primary exports in total merchandise exports is less than 10% were excluded.
Second, the values under column C are the average shares of nonfuel primary and
industrial exports in total merchandise exports for the sample period. In order to focus
on the effects of nonfuel primary exports on industrial exports and GDP, the
economies whose average share of nonfuel primary and industrial exports in total mer-
chandise exports is less than 90% were also excluded. These economies are largely
major oil-producing countries such as Algeria, Kuwait, Libya, Saudi Arabia, and
Venezuela.
The 74 countries included in this study have a broad representation. Geographically,
they include 27 African countries, eight Asian countries, five Caribbean countries, 14
European countries, 13 Latin American countries, one Middle-Eastern country, two
North American countries, and four Pacific countries. Economically, they include
32 low income countries, 17 lower-middle income countries, 11 upper-middle income
countries, and 14 high income countries. In addition, based on trade policies, they
include countries like Korea and other industrialized countries that pursued strongly
outward-looking policies, countries such as Argentina, Dominican Rep., Sudan, Peru,
©
Blackwell Publishers Ltd 2000

3
14 Zhenhui Xu
and other developing countries that pursued strongly inward-looking policies, and
3
countries that are somewhere in between.
The data needed for this study are real GDP, real nonfuel primary exports, and
4
real industrial exports. They are collected on an annual basis from the World Bank’s
World Data CD-ROM (1994). “GDP at market prices (constant 1987 US dollars)” was
taken as real GDP. “Exports of nonfuel primary products (current US dollars)”,
“
(
Exports of manufactures (current US dollars)”, and “Merchandise export price index
1987 = 100)” were used to derive the time series for real nonfuel primary exports,
and real industrial exports. The data on merchandise exports, “Merchandise exports
current US dollars)” were also collected from the same source to calculate the values
reported in Table 1.
(
3. Methodology
VAR Specifications
The VAR approach is well suited for this study for the reasons discussed in the intro-
duction. In applying the VAR, the time series of real GDP, real nonfuel primary
exports, and real industrial exports are all transformed from levels to logs of levels.
Because most algorithms used for fitting ARIMA models will fail if the time series
are nonstationary, appropriate differencing is important in time-series analysis. In a
VAR model, however, the asymptotic distribution that characterizes the estimates will
be the same whether the model is estimated in levels or in differences (Park and
Phillips, 1988, 1989; Sims, Stock, and Watson, 1990). In fact, Fuller (1976) shows in
his Theorem 8.5.1 that differencing does not gain any asymptotic efficiency in an
autoregression, even if it is appropriate; differencing can actually throw information
away and produce almost no gain. Nevertheless, for small samples, the distributions
of the estimates may well be improved by estimating the VAR model in differences
(Hamilton, 1994, p. 553, p. 652). Hence, I use first-differences rather than levels in esti-
mations. In addition, using first-differences can make interpretation easier since the
first-difference of the log of levels translates levels into growth rates.
Before estimating a VAR model, two decisions must be made. We have to decide
on the deterministic components and the appropriate lags for the model. Since arbi-
trarily chosen specifications for a VAR model will most likely produce unreliable
results, a data-based model selection criterion was used to specify the VAR model for
each economy in the sample.Among various model selection criteria, the one proposed
by Schwarz (1978), known as Schwarz’s Bayesian Information Criterion (BIC), has
been shown to outperform other alternatives (Mills and Prasad, 1992). The specifica-
tions of the VAR model for each economy in this paper are thus based on Schwarz’s
BIC.The results, along with the sample period for each economy, are reported in Table
2
. From Table 2, Schwarz’s BIC selected the specification of first-order VAR model,
VAR(1), without constant and trend for six countries, with constant but no trend for
4 countries, and with constant and trend for 34 countries. The BIC also chose a
2
VAR(2) model with constant but no trend for two countries, with constant and trend
for five countries, and a VAR(3) model with constant and trend for three countries.
Impulse Response Analysis
The inference of the effects of PEG on the growth of industrial exports and GDP
is based on the results of the impulse response analysis, using the VAR specifications
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 315
in Table 2. The impulse response function is designed to identify the dynamic
effects of an exogenous and temporary shock in one variable, say PEG, on other
variables, say the growth of industrial exports and GDP. I will focus the discussion
on the short-term and the long-term cumulative elasticities of the growth of indus-
trial exports and GDP with respect to PEG. The short-term elasticity is obtained
from the first step of the impulse response analysis. It measures the immediate impact
of PEG on the growth of industrial exports and GDP when an exogenous shock in
PEG occurs, and it is consistent with the causality tests in the previous literature.
Since it does not account for any dynamic feedback from the other two variables,
the short-term elasticity reflects the contemporaneous effects and is static in nature.
The long-term elasticity, on the other hand, is obtained by allowing all variables to
respond over time to the shock in PEG in the first step. It measures the long-term
cumulative effects of PEG on the growth of industrial exports and GDP at the last
step; i.e., the effects on the growth of industrial exports and GDP per unit of
cumulative long-term changes in the growth of primary exports. Because of the
dynamic interactions among the variables, the short-term effects of PEG on the growth
of industrial exports and GDP can be enhanced or dampened in the long term. To
obtain the long-term elasticity, the maximum number of steps for the impulse response
analysis is set at 20. In most cases, the values of the impulse response function
converge in less than 10 steps and, in all the cases, the values converge within the
20-step period.
It is well known that the results of impulse response analysis depend on the specific
ordering of the variables under investigation. Since the estimated matrix of variance
and covariance is not diagonal, orthogonalization is necessary before a meaningful
impulse response analysis can be conducted. Yet the orthogonalization strategy is not
unique. I confine the discussion of orthogonalization strategy to the use of triangular
matrices under the Choleski decomposition method. This method uses all the infor-
mation in the matrix of contemporaneous correlation among the estimated residuals.
In this study, there are three possible scenarios depending on whether PEG is ranked
first, second, or third. These three scenarios completely determine the range of results
in the impulse response analysis.
Among the three possible scenarios, one seems to be the most plausible a priori.
The analysis in the central scenario will focus on the case in which PEG is ranked
first. Here, shocks from PEG affect the growth of industrial exports and GDP con-
temporaneously, but shocks from GDP growth and industrial export growth have no
immediate effects on PEG. This choice is based on the view that exports of any
economy are driven primarily by foreign demand that is not contemporaneously
correlated with the performance of the domestic economy, and on the fact that primary
exports account for more than 50% of total merchandise exports for an overwhelm-
ing majority of the countries in our sample. Table 3 shows the short-term and the
long-term elasticities of PEG on the growth of industrial exports and GDP for the
central case.
To establish the robustness of the central case, however, the values of the impulse
response function under all of the alternative orthogonalization strategies were
also computed. Specifically, there is a total of six possible orthogonalizations for
each economy. Yet, the results of the impulse response analysis are the same under
some orthogonalization assumptions. Consequently, the values of the impulse response
function for only four different cases should be reported. The range of results
across all possible orthogonalizations for the long-term elasticities is reported in
Table 4.
©
Blackwell Publishers Ltd 2000

3
16 Zhenhui Xu
Table 3. Effects of PEG on the Growth of GDP and Industrial Exports: the Central Case
Elasticity of GDP w.r.t. PEG Elasticity of IEG w.r.t. PEG
Short-term
Long-term
Short-term
Long-term
Sierra Leone
Central African Rep.
Solomon Islands
Somalia
Zimbabwe
Belize
Niger
Tanzania
Liberia
-0.0036
0.0008
0.1297
0.1000
0.0863
0.0730
0.0108
0.0203
0.1736
0.0325
-0.0282
-0.0269
0.0409
0.1036
0.0708
0.0857
0.0420
0.0280
-0.0169
0.0252
0.1610
0.0231
0.0733
0.0351
0.1206
0.1209
0.1025
0.0951
-0.0012
0.0089
0.1034
0.0613
0.0406
0.1279
-0.0024
0.1291
0.0077
0.0434
0.1273
0.1342
-0.0019
0.0688
0.0221
-0.0318
0.1022
-0.0063
0.1186
-0.0107
0.0262
-0.0149
0.1091
0.1620
0.1852
0.0863
0.0622
0.0182
0.2625
0.0465
-0.1001
-0.0024
0.1317
0.1402
0.0312
0.0740
-0.1047
0.0194
-0.0815
0.0966
0.0951
-0.1191
0.0508
0.0438
0.1309
0.2517
0.1217
0.3587
0.0102
-0.1137
0.1324
0.1774
0.1258
0.1075
0.0036
0.0767
0.1682
0.0306
0.1552
0.2092
-0.2124
0.1001
0.0646
-0.1467
0.1442
0.0374
0.0919
-0.0260
0.4831
0.6729
0.9497
0.6409
0.9375
0.8494
0.3833
0.3719
2.0026
0.4040
0.2207
-0.2087
0.9180
1.5941
1.1778
0.2346
-1.2044
0.5766
-0.0714
0.8272
1.7404
0.6793
2.0783
-0.0254
0.8398
0.2605
0.3980
0.7368
1.1167
0.1706
0.9663
0.2273
1.0680
0.3092
0.6607
0.4602
0.9141
0.5307
0.0278
0.1787
0.2264
0.5101
-0.1343
0.5498
0.5466
1.0186
-0.6531
0.4438
0.7322
0.5689
0.8157
0.8327
0.9331
0.6806
0.4183
0.5052
3.4362
0.7943
0.4248
-0.3928
0.9910
1.4600
1.4291
0.4593
-1.4957
0.6408
-0.1346
0.8334
1.2349
0.5320
1.3141
0.3037
0.7631
0.3368
1.0354
0.8421
0.6719
1.3674
1.8238
0.5868
2.2284
0.2947
0.7695
0.3550
0.8280
0.3520
-0.1441
0.1341
0.4598
0.3510
-0.0663
0.7838
0.5233
0.8296
-0.6359
0.5942
Ghana
Dominican Rep.
South Africa
Zaïre
Burundi
Gambia
Malawi
Guyana
Paraguay
Zambia
Chad
Sudan
Nicaragua
Mauritania
Mali
Myanmar
Honduras
Togo
Iceland
Burkina Faso
Papua New Guinea
Chile
Madagascar
Ethiopia
United Kingdom
Guatemala
Dominica
Côte d’Ivoire
Haiti
France
Costa Rica
Uruguay
New Zealand
El Salvador
Peru
Belgium
Benin
Senegal
Argentina
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 317
Table 3. (Cont.)
Elasticity of GDP w.r.t. PEG
Elasticity of IEG w.r.t. PEG
Short-term
Long-term
Short-term
Long-term
Barbados
Jamaica
Fiji
United States
Denmark
Austria
India
0.0337
0.0885
-0.0367
0.0643
-0.2251
0.1323
-0.0192
-0.2481
0.0950
0.0588
-0.0092
0.1851
0.0225
0.0838
-0.2346
0.0716
0.1808
-0.0091
0.0042
0.0521
0.0256
0.1038
-0.0140
0.0660
0.0410
-0.0047
0.3223
0.0497
-0.0664
0.0128
-0.4696
0.2395
0.0161
-0.5639
0.0888
0.1097
-0.0585
0.1682
0.0397
0.2152
-0.3967
0.0844
-1.3368
0.0336
-0.0007
0.1059
0.0240
0.4060
-0.0340
0.1508
0.2009
-0.0119
0.2018
0.1794
0.7096
0.2219
0.3397
0.4252
0.3639
1.3238
-0.2093
0.6277
0.5060
0.6746
-0.4040
0.3237
0.6518
0.3106
0.4791
0.3384
0.1672
-0.6837
0.2045
0.3433
0.8370
0.3080
0.1798
0.3045
0.0876
-0.1239
1.0521
0.3504
-0.3106
0.6195
0.6786
0.8732
-1.2134
0.6200
0.4514
0.9638
-0.3438
0.5841
0.9367
0.1374
-1.0266
0.3465
0.3106
-0.6080
0.3099
0.6183
1.4290
0.3847
0.3263
0.3851
Brazil
Morocco
Canada
Greece
Portugal
Sri Lanka
Sweden
Israel
Ireland
Finland
Thailand
Spain
Mauritius
Pakistan
Philippines
Turkey
Korea, Rep.
Malta
Nepal
4. PEG and the Growth of GDP and Industrial Exports
This section presents the empirical results of the effects of the growth of nonfuel
primary exports on the growth of industrial exports and GDP, based on the impulse
response analysis. GDP, PEG, and IEG in Tables 3 and 4 stand for the growth rates of
real GDP, real primary exports, and real industrial exports, respectively.
PEG and Long-Term GDP Growth
From Table 3, 55 of the 74 countries in the sample show positive cumulative long-term
effects of PEG on GDP growth, while the remaining 19 countries show negative elas-
ticities. Although Spain and South Africa show negative elasticities, the values are
almost zero. The range for all elasticities is between -1.34 and 0.41. Five countries—
Uruguay, Brazil, Israel, Denmark, and Finland—show elasticities smaller than -0.20;
nine other countries—Philippines, Barbados, Liberia, Honduras, Costa Rica, Iceland,
Austria, Sweden, and Malta—show elasticities greater than 0.20. Hence, the long-term
effects of PEG on GDP growth are concentrated in a relatively narrow range across
countries.
©
Blackwell Publishers Ltd 2000

3
18 Zhenhui Xu
Table 4. Effects of PEG on the Growth of GDP and Industrial Exports: Sensitivity Analysis
Elasticity of GDP w.r.t. PEG Elasticity of IEG w.r.t. PEG
Lower
Upper
Sign
Lower
Upper
Sign
Sierra Leone
Central African Rep.
Solomon Islands
Somalia
Zimbabwe
Belize
Niger
Tanzania
Liberia
[0.0050,
[-0.0292,
[0.0146,
0.0310,
-0.0149,
0.1710,
0.1620,
0.1853,
0.0960,
0.0825,
0.0183,
0.2625,
0.0466,
-0.0446,
0.0470,
0.2499,
0.1402,
0.0313,
0.0901,
-0.0881,
0.0194,
-0.0631,
0.0967,
0.0951,
-0.0551,
0.1389,
0.0442,
0.1325,
0.2517,
0.1218,
0.3588,
0.0112,
-0.1052,
0.1706,
0.1774,
0.1364,
0.1075,
0.0627,
0.0768,
0.2730,
0.0307,
0.1553,
0.2213,
-0.2071,
0.1968,
0.0647,
-0.1026,
0.1443,
0.0673,
0.0920,
-0.0125,
4]
0]
4]
4]
3]
2]
4]
2]
3]
1]
0]
3]
4]
1]
1]
3]
0]
1]
0]
4]
1]
0]
3]
4]
4]
4]
4]
4]
3]
0]
2]
4]
4]
1]
3]
2]
4]
1]
2]
2]
0]
4]
2]
0]
3]
4]
2]
0]
[0.2332,
[-0.3424,
[0.2085,
[0.2784,
[0.2980,
[-0.0687,
[0.1629,
[0.2009,
[-1.1321,
[0.1618,
[0.0980,
[-0.3929,
[0.4692,
[-2.4950,
[-0.7597,
[0.0820,
[-2.2552,
[-0.2803,
[-0.1639,
[0.2043,
[-0.3552,
[-0.3403,
[-0.0860,
[0.3037,
[0.2890,
[-0.7088,
[0.7979,
[0.3744,
[-0.3105,
[1.0419,
[0.5075,
[0.3548,
[1.5298,
[-0.9193,
[-0.5561,
[-0.3271,
[-0.0467,
[-0.6034,
[-3.0042,
[-0.6011,
[-0.2253,
[-0.2157,
[-0.2826,
[-0.3218,
[-0.2683,
[0.2180,
[-0.8174,
[0.1699,
0.7322,
0.5689,
1.0440,
1.1273,
0.9332,
0.7764,
0.4184,
0.5053,
3.4362,
0.7944,
0.4249,
-0.1198,
1.0455,
1.4600,
1.4291,
0.4833,
0.0692,
0.6408,
-0.0719,
0.8334,
1.2349,
0.5321,
1.4929,
0.6268,
0.7631,
0.3369,
1.0355,
0.8421,
0.6720,
1.4837,
2.0822,
0.6823,
2.2284,
0.2948,
0.8031,
0.3551,
0.8444,
0.3521,
-0.1441,
0.1342,
0.4638,
0.4355,
0.0282,
0.7838,
0.5556,
0.8644,
0.0875,
0.5942,
4]
2]
4]
4]
4]
3]
4]
4]
2]
4]
4]
0]
4]
2]
2]
4]
1]
2]
0]
4]
2]
2]
3]
4]
4]
1]
4]
4]
2]
4]
4]
4]
4]
1]
2]
2]
3]
2]
0]
2]
2]
2]
1]
2]
2]
4]
1]
4]
[0.0163,
[-0.3086,
[-0.0436,
[0.0544,
[-0.0075,
[-0.0147,
[-0.0566,
[-0.1066,
[-0.0024,
[0.0516,
[-1.0718,
[-0.0418,
[-0.0103,
[-0.2683,
[-0.1142,
[-0.0824,
[0.0599,
[-0.3718,
[-0.1631,
[-0.0027,
[0.0057,
Ghana
Dominican Rep.
South Africa
Zaïre
Burundi
Gambia
Malawi
Guyana
Paraguay
Zambia
Chad
Sudan
Nicaragua
Mauritania
Mali
Myanmar
Honduras
Togo
[0.0320,
[0.0778,
[0.0487,
[0.2724,
[-0.0145,
[-0.1455,
[-0.0745,
[0.0681,
Iceland
Burkina Faso
Papua New Guinea
Chile
Madagascar
Ethiopia
United Kingdom
Guatemala
Dominica
Côte d’Ivoire
Haiti
[0.0978,
[-0.3273,
[-0.0449,
[-0.0758,
[0.1558,
[-0.1442,
[-0.4564,
[-0.0573,
[-0.3281,
[0.0311,
[-0.0916,
[-0.2179,
[-0.0566,
[0.0315,
[-0.0591,
[-0.0335,
France
Costa Rica
Uruguay
New Zealand
El Salvador
Peru
Belgium
Benin
Senegal
Argentina
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 319
Table 4. (Cont.)
Elasticity of GDP w.r.t. PEG
Elasticity of IEG w.r.t. PEG
Lower
Upper
Sign
Lower
Upper
Sign
Barbados
Jamaica
Fiji
United States
Denmark
Austria
India
[0.1997,
[-0.2564,
[-0.0981,
[-0.0562,
[-0.4697,
[0.0275,
[-0.0018,
[-0.6223,
[-0.0267,
[0.0203,
0.3223,
0.0498,
-0.0420,
0.0209,
-0.1587,
0.2396,
0.0397,
-0.2706,
0.0889,
0.1642,
-0.0129,
0.1682,
0.0397,
0.2152,
-0.1368,
0.0844,
2.0461,
0.0513,
-0.0008,
0.1686,
0.0241,
0.4060,
-0.0145,
0.1509,
0.2009,
-0.0083,
4]
2]
0]
2]
0]
4]
3]
0]
2]
4]
0]
2]
4]
4]
0]
2]
3]
3]
0]
4]
2]
4]
0]
4]
3]
0]
[-0.3066,
[-0.6156,
[0.4250,
[0.1385,
[-0.8825,
[0.0798,
[0.1944,
[-0.7949,
[-1.2135,
[-0.4836,
[-0.2892,
[0.2670,
[-0.3554,
[-0.0130,
[0.4219,
[-0.2495,
[-1.0267,
[-0.1372,
[0.1431,
[-0.8402,
[0.0694,
[0.1160,
[0.2520,
[0.0204,
[-0.3033,
[0.1133,
0.4936,
-0.1239,
1.0754,
0.3505,
-0.1769,
0.6195,
0.7088,
0.8733,
-0.8852,
0.6883,
0.4862,
0.9639,
0.0985,
0.5841,
1.0307,
0.1418,
0.7887,
0.3465,
0.3107,
0.0805,
0.3099,
0.6184,
1.4290,
0.3848,
0.3455,
0.3852,
3]
0]
4]
4]
0]
4]
4]
2]
0]
2]
2]
4]
2]
3]
4]
2]
3]
2]
4]
1]
4]
4]
4]
4]
2]
4]
Brazil
Morocco
Canada
Greece
Portugal
Sri Lanka
Sweden
Israel
Ireland
Finland
Thailand
Spain
Mauritius
Pakistan
Philippines
Turkey
Korea, Rep.
Malta
[-0.0585,
[-0.0419,
[0.0073,
[0.0428,
[-0.4074,
[-0.0681,
[-1.3369,
[-0.0077,
[-0.0354,
[0.0347,
[-0.0095,
[0.1774,
[-0.1069,
[0.0459,
[-0.0956,
[-0.0219,
Nepal
Note: The values under “sign” indicate the number of positive values out of four possible outcomes.
With two-thirds of the countries showing positive long-term elasticities of GDP
growth with respect to PEG, this empirical result provides evidence in support of the
hypothesis that PEG promotes GDP growth and that such positive effects are sus-
tainable in the long term. At the minimum, primary exports are a component of GDP.
In addition, they provide the exporting countries with a source of foreign exchange
revenues. These revenues can be used for imports of equipment, technology, and
other capital goods which can promote long-term GDP growth. As discussed in the
introduction, the channels through which PEG can promote GDP growth include
(1) stimulating domestic production, (2) relaxing foreign exchange constraints to allow
imports of manufactured goods, and (3) improving productivity from using these
imported capital goods. The first channel is the direct effect of PEG on GDP growth,
while the second and third channels are the indirect effect.
Country-specific factors can help explain the negative long-term elasticities found
in the sample. For example, primary exports accounted for more than 80% of total
merchandise exports on average for Guyana, Dominican Rep., Zambia, Nicaragua,
Peru, Fiji, and Papua New Guinea. Although these countries all show negative long-
©
Blackwell Publishers Ltd 2000

3
20 Zhenhui Xu
5
term elasticities, the causes can be very different. While Fiji is constrained by geo-
graphic remoteness, small domestic markets, and vulnerability to adverse weather,
Papua New Guinea is constrained by difficulties in establishing transportation and
communication between its population centers and the world markets. These con-
straints have prevented the two resource-rich Pacific countries from effectively pro-
moting their exports. For Zambia, the negative long-term elasticity can result from its
excessive reliance on a single export product (copper) and on public sectors in the pro-
duction of goods and services and the falling world prices of copper since 1975. On the
other hand, Guyana, Dominican Rep., Nicaragua, and Peru are well-known examples
of political and economic turmoil or misguided government policies.
PEG and Long-Term Industrial Export Growth
The effects of PEG on the growth of industrial exports are more positive. From Table 3,
62 of the 74 countries show positive long-term elasticities of industrial export growth
with respect to PEG; only 12 countries show negative long-term elasticities.With more
than four-fifths of the countries in the sample showing positive long-term cumulative
effects of PEG on the growth of industrial exports,there is strong support for the hypoth-
esis that PEG has sustainable long-term effects on the growth of industrial exports.
6
This result is consistent with the historical experience in economic development.
Economic development is a process of transformation of an economy from mainly
rural and agricultural to predominantly urban, industrial, and service in orientation.
Almost all industrialized nations grew out of agricultural economies. Hence, ignoring
the agricultural sector, as many developing countries did in the 1950s under the de-
velopment strategy of import substitution, can cost an economy dearly. On the one
hand, agricultural growth is complementary to industrial growth and is an integrated
part of an industrialization strategy (Mellor, 1966). On the other hand, the pursuit of
an agriculture- and employment-based development strategy requires an open trading
regime (Mellor, 1986, p. 80).
The range of long-term elasticities for industrial growth is, however, wider than that
for GDP growth. It is between -1.50 (Guyana) and 3.44 (Liberia). Furthermore, the
values of the elasticities are tilted toward the upper end. Of the 62 countries with
positive long-term elasticities, the values of the elasticities are greater than one for
11 countries, between 0.50 and 0.99 for 30 countries, between 0.25 and 0.50 for 18
countries, between 0.10 and 0.25 for two countries, and less than 0.10 for one country.
Of the 12 countries with negative long-term elasticities, the values of the elasticities
are lower than -1.00 for three countries (Finland, Morocco, Guyana), between -1.00
and -0.50 for two countries (Senegal, Mauritius), between -0.50 and -0.25 for three
countries (Denmark, Sri Lanka, South Africa), between -0.25 and -0.10 for three coun-
tries (Jamaica, Zambia, France), and greater than -0.10 for one country (El Salvador).
Of the 19 countries that show negative effects of PEG on GDP growth, five—Guyana,
Zambia, South Africa, Finland, and Denmark—also display negative effects on the
growth of industrial exports. This indicates that PEG negatively affects GDP growth
not only directly, but also indirectly by crowding out industrial exports for these five
countries.
Long-Term versus Short-Term Effects
The short-term elasticities are reported in Table 3 next to the long-term elasticities to
facilitate comparison. The short-term elasticities show that PEG has immediate posi-
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 321
tive effects on GDP growth for 55 countries. The 19 countries for which PEG has
immediate negative effects on GDP growth were Brazil, Israel, Denmark, Fiji, Domi-
nican Rep., South Africa, India, Zambia, Turkey, Argentina, Greece, Thailand, Benin,
Nepal, Peru, Sierra Leone, Guatemala, Uruguay, and Burkina Faso. Of these, Brazil,
Israel, and Denmark are the only ones with a value lower than -0.20. The negative
effect for the other 16 countries is very small, ranging from -0.037 to -0.001. With the
exception of South Africa, India, Thailand, Benin, Sierra Leone, Guatemala, and
Burkina Faso, the negative short-term effects in these countries are reinforced into
larger long-term negative effects.
The short-term elasticities reported in Table 3 also show that PEG has immedi-
ate positive effects on the growth of industrial exports for 65 countries. The remain-
ing nine—Guyana, Mauritius, Senegal, Sri Lanka, Morocco, Mali, South Africa, El
Salvador, and Zambia—show negative short-term effects. With the exception of Mali
and Zambia, the negative effect for the other seven countries is large, ranging from -
0
.134 to -1.205. While the negative short-term effects are reinforced in the long term
for Guyana, Morocco, South Africa, and Zambia, the other five countries show reduced
negative effects in the long term. These results provide additional support for the
hypotheses that PEG promotes the growth of industrial exports and GDP and that
such positive effects are sustainable in the long term.
One interesting aspect of the results is that for some economies, the long-term effect
is completely different from the short-term effect; they work in opposite directions.
This raises an interesting policy issue; i.e., the choice of appropriate policies for the
government of a developing country to adopt for promoting its long-term economic
growth. For Central African Rep., Spain, Papua New Guinea, Nicaragua, Guyana, and
Finland, the effect of PEG on GDP growth is positive in the short-term but negative
in the long-term. For India, Thailand, Benin, Sierra Leone, Guatemala, and Burkina
Faso, however, the opposite is true. For the effects of PEG on the growth of industrial
exports, El Salvador is the only country which turned short-term negative effects into
long-term positive effects. Finland, Denmark, France, and Jamaica, which display
negative long-term effects, show positive effects in the short term.
Sensitivity Analysis
The above discussion is based on the orthogonalization strategy that, I have argued,
is the most plausible and meaningful based on a priori theoretical grounds.To examine
the robustness of the results, I also performed sensitivity analysis by considering all
possible triangular orthogonalizations using the Choleski decomposition. Table 4
shows the lower and upper bounds for the long-term elasticities of GDP and IEG with
respect to PEG under all possible orthogonalizations, as well as the number of posi-
tive outcomes out of four possible outcomes.
The values of all GDP elasticities, with the exception of Finland, fall in a relatively
narrow range, from -1.34 to 0.41 across countries, as they are in the central case. In 38
cases, the values in the central case coincide with, or are very close to, the upper or
lower bounds of the possible values.This suggests that the actual distribution of results
across countries may be even narrower than the range identified in the central case.
Of the 74 countries in the sample, 41 show absolute robustness and another 10 show
strong robustness in terms of the qualitative nature of the results. In other words, the
sign of the elasticities of GDP growth in the central case is maintained across all of
the orthogonalization outcomes (four in four) for 41 countries, and across most of the
orthogonalization outcomes (three in four) for 10 countries. The 41 countries include
©
Blackwell Publishers Ltd 2000

3
22 Zhenhui Xu
2
1
4 of the 55 countries that display positive elasticities in the central case and 17 of the
9 countries that display negative elasticity in the central case. The ten countries all
show positive elasticities in the central case. For another 14 countries, the sign of the
elasticities in the central case is maintained across half of the orthogonalization out-
comes (two in four). For the rest of the countries, the sign of the elasticities can be
maintained only in the central case. They include seven countries—Sudan, Gambia,
Burundi, Haiti, Ghana, Paraguay, and United Kingdom—that display positive long-
term elasticities of GDP growth, and two countries—South Africa, Finland—that show
negative long-term elasticities.
The range of results for the effects of PEG on industrial export growth is rather
wide, from -3.00 to 3.44. Yet with the exception of France, Burundi, and Guyana, the
values of the elasticities of industrial export growth fall in the range in the central case.
There are 15 countries with upper bound elasticities greater than 1.0, and six countries
with lower bound elasticities lower than -1.0. The values in the central case coincide
or are very close to the upper or lower bounds of the possible values in 44 cases.
In this case, 39 show absolute robustness; the sign of the elasticities of industrial
export growth in the central case is maintained across all of the orthogonalization
outcomes (four in four). These 39 countries include 33 of the 62 countries that display
positive elasticities in the central case and 6 of the 12 countries that display negative
elasticity in the central case. Another nine countries show strong robustness across all
of the orthogonalization outcomes (three in four). They include five countries that
display positive elasticities in the central case and four countries that display negative
elasticity in the central case. For the remaining 26 countries, Finland is the only country
for which the sign of the elasticities can only be maintained in the central case.
To summarize, the sensitivity analysis indicates that the central results about the
effects of PEG on the growth of industrial exports and GDP are robust for 51 and 48
countries in our sample of 74 countries, respectively. Therefore, around two-thirds of
the 74 countries show that the long-term elasticities obtained under the central case
are the most plausible results.
5. Summary and Concluding Remarks
In this paper, I adopted a multivariate VAR approach to investigate the effects of
nonfuel primary exports on industrial exports and GDP in 74 countries. This method-
ology permits the identification of the long-term effects of PEG on the growth of indus-
trial exports and GDP by taking into account the dynamic feedback between primary
exports and the other two variables, as well as the short-term effects.
There is strong evidence that PEG has positive effects on the growth of indus-
trial exports and GDP for most countries, and that such positive effects are sustain-
able in the long term. In 74 countries, the positive effects of PEG on GDP growth are
detected for 55 countries in the short term and for 55 countries in the long term. The
positive effects of PEG on the growth of industrial exports are detected for 65 countries
in the short term and for 62 countries in the long term. Some countries are even able to
turn the short-term negative effects into long-term positive effects.These results suggest
that country-specific institutional, political-economy, and policy differences play an
important role in driving the different results of PEG on industrial exports and economic
growth.
Since primary exports still account for a lion’s share of total exports for the
majority of developing countries, the empirical results found in this study have
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 323
important theoretical and policy implications. Specifically, they provide empirical
support for the comparative advantage theory of international trade. For most develop-
ing countries, primary exports contribute positively to their economic growth, at
least to the growth of their industrial exports. Therefore, the governments in devel-
oping countries should not discriminate against primary exports, as some earlier
studies suggested. Instead, they should adhere to policies that aim at export
promotion.
Appendix
This appendix briefly discusses the VAR model used in the study. For detailed discus-
sion of the VAR approach, see Sims (1980), Hamilton (1994), or other time-series
econometric books.
The VAR model provides a means for forecasting systems of interrelated time-series
variables and for analyzing the dynamic impact of random disturbances on the vari-
ables. It makes the least theoretical demands on the structure of a model. It is well
suited to forecasting variables where each variable helps predict other variables. In a
VAR model, every equation has the same right-hand variables, which include only
lagged values of all of the endogenous variables. As a result, the problem of simul-
taneity does not arise in the VAR model, and the best technique for estimations is the
ordinary least squares.
In this study, primary export growth (PEG), industrial export growth (IEG),
and GDP growth (GDP) are jointly determined by a three-variable VAR model. If the
time series of PEG, IEG, and GDP are nonstationary, then the appropriate model may
be a vector-error-correction model, which is a VAR model with the error-correction
terms. But since PEG, IEG, and GDP are all measured by the rate of growth, they
are stationary, suggesting that the VAR model is appropriate. The estimated VAR
model is
GDP = Sa PEG + Sb IEG + Sd GDP + c + t + e ,
(1a)
(1b)
(1c)
t
1i
t-i
1i
t-i
1i
t-i
1
1
1,t
IEG
t
= Sa2iPEGt-i + Sb2iIEGt-i + Sd2iGDPt-i + c
2
+ t
2
+ e2,t
,
PEG
t
= Sa3iPEGt-i + Sb3iIEGt-i + Sd3iGDPt-i + c
3
+ t
3
+ e3,t,
where PEG, IEG, and GDP are the endogenous variables, the constant, c, and the
time trend, t, are the exogenous variables, a, b, and d are the parameters, and e is
the innovations that may be contemporaneously correlated with each other but are
uncorrelated with their own lagged values and with all of the right-hand side variables.
Schwarz’s (1978) BIC was used to select the deterministic components and the lag
structures for each country. The results are reported in Table 2.
A change in the innovations in the above VAR model, say e3,t in (1c), will immedi-
ately change the value of current PEG. It will also alter all future values of PEG, IEG,
and GDP since lagged PEG appears in all three equations. Such effects of a one-
standard-deviation shock to one of the innovations on current and future values of the
endogenous variables are analyzed by the impulse response function. If the innova-
tions, e1,t, e2,t, and e3,t, are uncorrelated, then interpretation of the impulse response will
be straightforward. e1,t is the innovation for GDP, e2,t for IEG, and e3,t for PEG. Since
the innovations are usually correlated, however, decomposition is needed. The
Choleski decomposition is employed in the paper.
©
Blackwell Publishers Ltd 2000

3
24 Zhenhui Xu
References
Balassa, B., “Export and Economic Growth: Further Evidence,” Journal of Development
Economics 5 (1978):181–9.
———, “Outward Orientation,” in H. Chenery and T. N. Srinivasan (eds.), Handbook of
Development Economics, vol. 2, Amsterdam: North-Holland (1989):1653–59.
Chenery, H. and A. Strout, “Foreign Assistance and Economic Development,” American
Economic Review 56 (1966):679–733.
Choi, E. K. and H. Beladi, “Why East Asian Countries Grow Faster Than Others: a Recipe for
Economic Growth,” Journal of International Trade and Economic Development 5 (1996):
2
07–37.
Cuddington, J. T. and H. Wei, “An Empirical Analysis of Real Commodity Price Trends: Aggre-
gation, Model Selection and Implication,” Estudios Economicos 7 (1992):159–79.
Eswaran, M. and A. Kotwal,“Export Led Development: Primary vs. Industrial Exports,” Journal
of Development Economics 41 (1993):163–72.
Fuller, W. A., Introduction to Statistical Time Series, New York: Wiley (1976).
Grossman, G. M. and E. Helpman, Innovation and Growth in the Global Economy, Cambridge,
MA: MIT Press (1993).
Hamilton, J. D., Time Series Analysis, New Jersey: Princeton University Press (1994).
Helpman, E. and P. Krugman, Market Structure and Foreign Trade, Cambridge, MA: MIT Press
(1985).
Hirschman, A. O., The Strategy of Economic Development, New Haven, Conn: Yale University
Press (1958).
Krueger, A., “Trade Policy as an Input to Development,” American Economic Review 70
(1980):288–92.
Lewis, S. R., “Primary Exporting Countries,” in H. Chenery and T. N. Srinivasan (eds.), Hand-
book of Development Economics, vol. 2, Amsterdam: North-Holland (1989).
MacBean, A. I., Export Instability and Economic Development, Cambridge, MA: Harvard
University Press (1966).
McKinnon, R., “Foreign Exchange Constraint in Economic Development and Efficient Aid
Allocation,” Economic Journal 74 (1964):388–409.
Mellor, J. W., The Economics of Agricultural Development, Ithaca, NY: Cornell University Press
(1966).
———, “Agriculture on the Road to Industrialization,” in J. Lewis and V. Kallab (eds.), Devel-
opment Strategies Reconsidered, Washington, DC: Overseas Development Council (1986).
Mills, J. A. and K. Prasad, “A Comparison of Model Selection Criteria,” Econometric Reviews
1 (1992):201–33.
1
Onafowora, O. A. et al., “Trade Policy, Export Performance and Economic Growth: Evidence
from Sub-Saharan Africa,” Journal of International Trade and Economic Development 5
(
1996):341–59.
Park, J. Y. and P. C. B. Phillips, “Statistical Inference in Regressions with Integrated Processes:
,” Econometric Theory 4 (1988):528–33.
——, “Statistical Inference in Regressions with Integrated Processes: 2,” Econometric Theory
(1989):95–131.
1
—
5
Schwarz, G., “Estimating the Dimension of a Model,” Annals of Statistics 6 (1978):461–4.
Sims, C. A., “Macroeconomics and Reality,” Econometrica 48 (1980):1–48.
Sims, C. A., J. H. Stock, and M. W. Watson, “Inference in Linear Time Series Models with Some
Unit Roots,” Econometrica 58 (1990):113–44.
World Bank, World Development Report, New York: Oxford University Press (1987).
—
—
——, World Tables, Baltimore: Johns Hopkins University Press (1992).
——, The East Asian Miracle: Economic Growth and Public Policy, New York: Oxford
University Press (1993).
Xu, Z., “On the Causality Between Export Growth and GDP Growth: An Empirical Reinves-
tigation,” Review of International Economics 4 (1996):172–84.
©
Blackwell Publishers Ltd 2000

PRIMARY EXPORTS, INDUSTRIAL EXPORTS, AND GDP 325
———, “Export and Income Growth in Japan and Taiwan,” Review of International Economics
6
(1998):220–33.
Notes
1. For a summary of the test results, see Balassa (1989).
2. See Sims’ (1980) influential work for further discussion of the VAR approach in analyzing
the dynamics of economic systems.
. The classification for geographical location and income is based on the World Bank’s World
3
Data (1994) from which the data for each country are collected, while the classification for trade
policies can be found in the World Bank’s World Development Report 1987.
4. An alternative measure of economic growth can be GDP minus exports, which is an appro-
priate measure if the positive externality of exports is the focus of this paper. Yet to verify the
proposition and hypotheses discussed in the introduction, GDP appears to be an appropriate
variable.
5
. For a discussion of the general economic condition for these economies, see TIDE Country
Reports from the World Bank’s World Data CD-ROM (1994).
. For a survey of the experience of primary exporting countries, see Lewis (1989).
6
©
Blackwell Publishers Ltd 2000