COMMODITY CONCENTRATION AND EXPORT INSTABILITY: A MISSING LINK OR HUNTING A SNARK?
| Published date | 01 May 1981 |
| Author | C. Thanassoulas,C. W. Lawson |
| DOI | http://doi.org/10.1111/j.1468-0084.1981.mp43002006.x |
| Date | 01 May 1981 |
COMMODITY CONCENTRATION AND EXPORT
INSTABILITY: A MISSING LINK OR HUNTING
A SNARK?
C. W. Lawson and C. Thanassoulas
Despite considerable efforts and substantial ingenuity, economists have failed to
establish any reliable statistical link between the commodity concentration and
export instability of less developed countries (LDCs). Several explanations have been
advanced to account for the elusive quality of so plausible a relation. If exports are
highly concentrated in stable commodities, or conversely, diversified into products
whose proceeds are highly correlated, then the expected positive association between
commodity concentration and export instability will fail to materialize. In addition
MacBean and Nguyen (1980) have argued that for any one country, not only can a
given level of concentration be consistent with a range of instability indices, but also
the link is more tenuous as concentration rises. So a cross-section analysis of the
relatively highly concentrated LDCs may well fail to establish the expected connection.
An alternative explanation has been proposed by Tuong and Yeats (1976). If the
Gini-coefficient, which is the usual measure of concentration, is 'unstable, and
sensitive to the level of trade data used in its construction, then this may be an impor-
tant reason for the different conclusions of the diversification studies'.' As theories of
export instability give no precise indication of the appropriate SITC level at which to
measure commodity concentration, and empirical studies have employed different
levels,2 this is a plausible explanation. Using 1974 data from fifty developed and
developing countries, Tuong and Yeats demonstrate that there is a certain degree of
variation in the ranking of countries, depending upon the SITC level used for the
Gini-coefficient. They argue that such concentration indices, calculated at a single
data level, can hide substantial inter-country divergences in trade structure: diver-
gences which emerge only in comparisons of indices calculated at different levels.
To reduce this problem they propose an alternative Full Information Index (F) which
combines data from several levels.3
'Tuong and Yeats (1976), P. 299.
2lbid., p. 299.
The Gini-coefficient is defined as
where there are N export goods, X is the annual value of exports of good i, and X the total value
of exports. The Full Information index for three levels of data is
19 99
F=100I XXifXiJj
t ¿0 j0 k1
where x, x, and X/k are the shares of the one, two and three digit SITC shipments in total trade.
201
To continue reading
Request your trial