Opening the package:
the Asian Drivers and
poor-country trade
Christopher Stevens and Jane Kennan
Institute of Development Studies
University of Sussex, Brighton, BN1 9RE, UK
April 2005
Contents
Summary
1
The problem
Product breadth
Non-marginal effects
Problems with the traditional methodology
1
2
2
2
The contribution of this paper
Objective
Methodology
3
3
4
Which products and countries?
Widest trade balance effects
Incidence of trade effect by type
6
7
9
Product profiles
10
References
14
iii
Summary
The paper describes and applies a new methodology to focus attention on the products and
countries most affected by the emergence of the Asian Drivers as substantial, fast-growing
traders. It is needed because the Asian Driver impact is widespread and non-marginal. The
first means that most countries will be affected on several fronts, both directly and via
substitute/complementary products, and that these effects may be reinforcing or offsetting.
The latter means that the Asian Driver impact will be felt on world market prices and not be
limited to countries that trade directly with either China or India. Neither of the most
conventional focusing mechanisms is appropriate to this combination.
Papers such as this cannot answer questions such as ‘how will Ghana’s economy be affected
by Asian Driver change?’ But they can identify the most important value chains and types of
effect that require detailed scrutiny to provide an answer. The selection of methodology for
the attention focusing is important since it will affect the sectors and countries that are
highlighted.
When applied in respect of China to most sub-Saharan African (SSA) and South American
states the methodology both confirms some expectations and throws up some surprises. In
addition to the well recognised impact of China on mineral exporters (largely positive) and
clothing exporters (negative), the methodology also highlights the potential importance of
Chinese demand for animal feed and leather, and its impact in the aluminium, white goods
and brown goods markets. It also draws attention, for example, to the increased need for
OECD states to amend the origin rules in their trade preference agreements – not a
conclusion normally apparent from broad-brush analyses of China’s impact on SSA. The
countries affected by China include Benin, Sudan and Tanzania as well as the more obvious
candidates. The trade balance effect on most SSA states is likely to be positive – despite the
well rehearsed problems of clothing exporters.
The problem
Never has there been a commodities boom like it. Not only oil but also copper and iron ore
are fetching record prices, while aluminium and zinc have in the past few weeks hit their
highest levels for many years … The gains have not been limited to metals and minerals:
coffee, at five year highs, and raw sugar are both at least 50 per cent up on last year. This is
the first time that there has been such a strong synchronised rise across so many different
commodities. (Financial Times, 11 April 2005)
It will come as no surprise that this FT article points to a consensus view on the underlying
cause of these price rises: China. There is a wide acknowledgement that the bursting of
China, and to a lesser extent India, on to the world trade stage is bound to have substantial
impacts on the rest of the world, and that RoW includes poor developing countries. The
problem is how to go beyond this general recognition to say something about the impact of
the Asian Drivers on, say, Ghana that is not banal but does not consume research resources
that are disproportionate to the conclusions generated.
This paper aims to contribute to the process of ‘making the next step’. The title is intended to
draw an analogy with the impenetrable cellophane wrapping on boxes of goodies, where it is
clear from the outside that something interesting lurks inside, but the task of getting into it is
harder than you expect. The difficulties arise in the case of the Asian Drivers because their
effect on the world economy will be widespread and non-marginal.
1
Product breadth
Because the Asian Drivers will impact on world trade in many products, most countries are
likely to be affected on several fronts both positively and negatively. Countries producing
goods highly demanded by the Asian Drivers (e.g. some minerals) may see export growth.
Those countries exporting products in competition with the output of Asian Drivers (such as
clothing) will see exports fall while importing countries will gain from lower prices. If the
importers also have domestic industries that are in competition on the local market with
Chinese exports there will be distributional effects (between producers and consumers) and
possible knock-on effects on the feasibility of the country’s industrial policy.
The population of trade flows to be analysed is made even larger because trade-offs between
imported and exported goods are often likely to be much more subtle than these clear
illustrative examples. Growing Chinese consumption of meat, for example, is sucking in
substantial imports of feed materials. This is good news for exporters of oil seeds, but if they
are also importers of meat the net impact on their terms of trade is uncertain. The very
breadth of Asian Driver activity means that there will be impacts on the price and availability
of transport (with the FT article quoting shipyards in South Korea, Taiwan, Japan and China
as being fully booked for the next three years), and in the supply of finance for investment in
sectors that are not booming.
Non-marginal effects
Because the change is non-marginal it will affect world markets generally. Because the
impact will not be limited to countries that trade directly with China and India the potential
population of poor states to analyse is a large one. Jenkins and Edwards show that few of the
18 developing countries they analyse have significant trade with China: just four directed
over 5 percent of exports to China in 2002, whilst only seven (all but two Asian) sourced over
5 percent of imports (Jenkins and Edwards 2004: Tables 1–3, 10, 12, 14).
The exports of the two SSA countries with over 1.5 percent of exports going to China were
mainly minerals with well developed world markets in which the precise direction of trade is
relatively unimportant compared with global trends. Nigeria was not among them because it
sells its oil to USA, which is where the appropriate refineries are situated. It is not insulated
from the Asian Drivers effect compared with Cameroon because it does not export oil direct
to China whilst the latter does.
Nor can the analysis be limited just to the products that are imported/exported on to the world
market by China and India. There are substitutes and complementary products that must be
taken into account (see Box 1). Questions such as those posed in Box 1 for aluminium arise
across the board and many require specialist industrial knowledge to provide an answer.
Problems with the traditional methodology
Faced with these characteristics, neither of the obvious methodologies for providing an initial
trade analysis is wholly appropriate. One approach is to undertake a broad-brush trade review
to identify overlaps between the trade of the Asian Drivers and that of particular developing
countries. This will usually be at the HS 2-digit level or the 3-digit SITC level used by
Jenkins and Edwards 2004. The problem with this approach is that these categories often do
not coincide with actual product markets. Sometimes they are too broad; there are 96 HS 2digit chapters and c. 275 SITC 3-digit ones. There can be intense competition in one subsection that is masked because of the ‘noise’ from the remainder of the category. Sometimes
2
markets involve products that are classified under
different headings. The example of aluminium
cited in Box 1 is a case in point.
The traditional way to overcome this problem
when dealing with small developing countries is to
go to the other end of the spectrum and undertake
the analysis at a high level of disaggregation.
There are 5,705 HS 6-digit categories, which allow
a much more precise link to be made between trade
flow data and actual product markets. Even here,
though, there are problems. The clothing firm
Shibani in Mauritius is in effect producing a
different type of product to that of CMT, just down
the road. The first produces luxury garments using
very high-tech equipment and expensive materials,
whilst the latter is more run of the mill. But their
outputs are classified under the same 6-digit
heading.
With the Asian Drivers, though, such problems
become completely insurmountable. Such is the
breadth of their exports and imports that it
becomes a mammoth task to try to cover even the
most important at a high level of disaggregation.
Box 1. Linked products
The aluminium industry provides a good
example of the difficulty identifying whether
products are substitutes, complementary or
unconnected. One of China’s fastest-growing
imports is of aluminium oxide (alumina) – HS
281820 – and one of its fastest-growing
exports is aluminium and products – ex HS
76. Ghana is also an exporter of the latter.
As the two product groups are in different HS
chapters the possible link between them will
be investigated only if it is specially flagged.
But what is the link?
Was the aluminium oxide an input into the
production of aluminium? Almost certainly
part of it was. China’s production of primary
aluminium is reported to have almost
doubled between 2001 and 2004, and to
have risen by 1.04 million tonnes in the last
year (International Aluminium Institute
website – http://www.world-aluminium.org/).
Alumina production is reported to have
increased between 2003 and 2004 by only
four-fifths as much; yet it takes about two
tonnes of alumina to produce one tonne of
aluminium.
If so, Ghana will be affected by the increased
costs of its raw material, competition from
China in third markets and, potentially,
increased demand for its product if
consumption in China starts to exceed
domestic production. The net impact on
Ghana will be determined by the relative
importance of all three effects.
Consequently, any attempt to answer the question
‘what impact will China have on Ghana?’ risks
falling into one of two pits. If it goes for a highly
that is not all. Some of the alumina may
aggregated analysis the results may well go little But
not have been used to produce aluminium; it
beyond putting a few numbers on the rather is also an input into the electronics industry
general conclusions that can be drawn without any for insulators and filters. Chinese
will increase world demand for
research at all. These are that the outlook for manufactures
alumina. This will have knock-on effects on
primary commodity exports is brighter than it has Ghana related to the ease with which supply
been in the past, whilst that for manufactured can be increased.
exports is much darker. If, by contrast, the analysis
plunges into the details of traded goods it will take a substantial input of resources to
complete the job. And then, having spent significant time and money the conclusions are
likely to be … rather similar to those from the aggregate analysis! This is because, in truth,
there is so much uncertainty about the way in which product markets will evolve that the
impact on small participants such as Ghana must remain a matter of speculation.
The contribution of this paper
Objective
There must be a better way. This paper attempts to take the next step in developing an
appropriate methodology. It contributes to the task of making a more informed and focused
initial analysis of the potential areas in which in the Asian Drivers may affect the trade of
poor developing countries. The ultimate goal is to be able to identify more clearly than we
can at the present time which sectors in which countries deserve the highest priority for the
3
more detailed (and therefore resource-intensive) research that will be needed to advise policy
makers on how they should respond to the changing world.
The key to the approach trialled in this paper is the selection of groups of HS6 sub-heads that
appear to be related, for which export or import growth in the Asian Drivers has been
particularly rapid, and which are important products for developing countries. By taking
account only of those HS6 sub-heads for which trade growth has been strong and which are
relevant to developing countries, the problem of too much ‘noise’ inherent in the aggregate
trade analysis is diminished. But by re-aggregating these items into specially created product
groups, the exercise avoids simply creating a morass of indigestible information.
This is a first attempt, done in haste, and clearly contains errors (particularly in relation to the
re-aggregation part); it needs to be refined in further exercises. Many lessons were learned in
this initial application of the methodology which can be put into practice next time around.
Despite the limitations, though, it is instructive to see how far the conclusions appear to differ
from those derived in other studies of the potential trade impact of the Asian Drivers on
developing countries. To this end, a comparison is made in the next section between these
findings and those reported by Jenkins and Edwards (2004).
Methodology
The methodology requires four sequential and labour-intensive steps to be undertaken
separately for imports and exports by China and India. These are listed below, taking the
example of China’s imports; analogous steps were taken for China’s exports and are also
required for India’s trade, but the authors ran out of time before any significant progress
could be made on India!
♦ Step 1. Data on all China’s HS6 imports from the world were extracted from
Comtrade, and those that were valued at $25 million or more and had grown
fastest were selected for further study. The threshold for determining ‘fastest
growth’ was 150 percent of China’s import growth; only HS6 items that had
grown more than 1.5 times as fast as the average were selected, and there were
397 of these, running to 12 pages! In most cases, the growth rate was calculated
over the 5-year period ending in 2003 (the latest for which data are available), but
additional products were added if they had not been exported for the full five
years but appeared to be relevant.1
♦ Step 2. The selected items were grouped into 37 broad product groups, using the
researchers’ judgement. The items in these groups were then aggregated, and eight
selected for further study because they represented a high proportion of world
imports of the same group.2 These were: feed, chemicals, ferrous metals, cobalt,
copper articles, lead ores and concentrates, aluminium oxide and sulphur.
1
Comtrade offers two sources for imports from the world, one of them derived from the importing country’s
data and the other from the data supplied on exports to China by all other reporters. The two do not coincide!
It was decided to use the data on imports reported by China as likely to be the more comprehensive. In the
event, however, this may have been a mistake since when looking at other countries’ trade the data from the
source derived from their trade partners’ export statistics appeared more plausible. We have ended up,
therefore, using one of the two sources for China’s imports and the other for developing country imports.
2
The initial product groups were: meat, fish, fruit and nuts, feed, food, tobacco, chemicals, distillation
gases/oils, plastics, rubber, leather and articles, wood and paper, textiles, footwear, glass, iron/steel and
articles, marble, cobalt, lead, nickel, zinc, aluminium oxide, coal/anthracite, sulphur, pharmaceuticals,
engines and vehicles, silver, platinum, copper articles, miscellaneous boilers/machinery*, electrical/electronic
equipment,* materials for electronics, precision instruments*, railway locomotive parts*, floating structures*,
photographic goods*, watches*. The groups marked with an asterisk were not analysed further because of
4
♦ Step 3 involved collecting data from Comtrade on the exports and imports of all
the items in the eight product groups by every ‘significant’ sub-Saharan and Latin
American state. The term ‘significant’ was applied judgementally, largely to avoid
wasting limited time on countries unlikely to yield useful results (such as those
currently undergoing or recovering from civil war). Thirty sub-Saharan African
and ten Latin American countries were analysed.3
This information was used to identify countries that have significant exports or
imports of any of the eight product groups. Again, the term ‘significant’ was
applied judgementally (and with a much lower absolute threshold for the inclusion
of SSA than Latin American countries). The objective was to concentrate the
limited research resources on the countries for which it was most likely there
would be some evidence to analyse.
As a result of this step it was deemed possible to exclude from further analysis
some of the product groups on the grounds that SSA and Latin American trade in
them was trivial. Two were excluded from the analysis of SSA/Latin American
exports (lead ores and sulphur) and three (lead ores and concentrates, cobalt and
aluminium oxide) from the analysis of their imports.
♦ Step 4 involved a detailed analysis of the remaining product groups to determine
trends in China and in the main SSA/Latin American importers. An important
function was to determine how representative the figures used in Step 3 for a
single year were of the broader picture. Interestingly, the conclusion in all cases
was that there were no spurious results from the one or two year snapshots in
Steps 1-3: the countries that appear to trade in these goods do indeed do so over a
five year period. This has implications for the future use of the methodology.
Checking out trade over multiple years is time consuming. If the results are
important for analysis then it must be done. But if it is required only to check that
one year results are not a statistical blip the implication is that the thresholds set
for this analysis are sufficiently high that they reduce the danger.
For China’s exports the steps were very similar. Seventeen product groups were initially
identified,4 whittled down to eight in Step 2.5 But the favoured approach of aggregating
SSA/Latin American imports and exports of HS 6-digit items was not followed for textiles
and clothing, for which instead total imports/exports in Chapters 50–60 (except lightly
processed wool and cotton fibres) and Chapters 61–63 were used. This was done for purely
pragmatic reasons: the Chinese exports identified are in numerous HS6 sub-heads, and there
was insufficient time to extract developing country data on all of them individually!
their highly disparate nature. A more careful and resourced application of this methodology would seek to
produce further product groups from the asterisked groups and a ‘rump’ of items that did not fit neatly into
any of the groups.
3
Botswana, Lesotho, Namibia and Swaziland were not analysed because of the difficulty of distinguishing
their trade from that of South Africa. In a further application of this methodology Botswana and Namibia
would almost certainly be included because of their relevant trade and an attempt made to obtain accurate
data.
4
Pharmaceuticals, cosmetics, chemicals, plastics, leather and articles, wood and paper, textiles, clothing,
footwear, glass, ferrous metals and products, aluminium, engines and pumps, white goods, brown goods,
cars and parts, furniture.
5
For initial test. The selected groups were leather and articles, textiles, clothing, footwear, ferrous metals and
products, aluminium, white goods and brown goods.
5
No exclusions were made at Step 3 on the grounds of trivial developing country imports or
exports. But for three groups – footwear, brown goods and white goods – Brazil was the only
SSA/Latin American net exporter. All of the categories were widely imported in both SSA
and Latin America.
In Step 4, brown goods were not analysed to determine SSA/Latin American import trends.
This was not because they were unimportant, but because they were ubiquitous. Every
country (apart from Brazil – the only net exporter) imported a lot. It was assumed, therefore,
that no further research is needed to determine that all consumers in the countries considered,
apart from Brazil, will tend to gain from lower world prices (subject, of course, to intraindustry trade patterns that could only be picked up by the more focused studies to which
research such as this is a pointer).
Imports can of course be seen from different perspectives. A fall in price is a clear balance of
trade gain, but such gains may be perceived as less important than the increased competitive
threat they pose to domestic industries. This is not an issue that can be dealt with by looking
just at trade statistics; it requires more detailed country and/or product studies. But the sorting
permitted by this methodology helps to identify the areas in which such focussing could be
most useful.
Which products and countries?
The following four tables and two figures summarise the broad picture of the products and
developing countries that are of particular interest. The first two sets (Table 1/Figure 1, and
Table 2/Figure 2) do the same thing respectively for China’s imports and its exports. The
tables summarise the product groups that were identified as being important (on the grounds
of substantial Chinese trade and rapid growth, and of trade relevance to SSA and/or Latin
America). In each case the table shows whether the SSA/Latin American states were analysed
as exporters, importers or both.
Figures 1 and 2 then plot the countries that Table 1. The product groups for China’s
were identified as significant traders by imports
Product groups
SSA/Latin America analysed as:
product group. They show the relative
Importers
Exporters
importance (in terms only of numbers of 1 Feed
countries potentially affected – not on any 2 Chemicals
other score) of each product group and 3 Cobalt
whether it’s importance arises because 4 Copper articles
SSA/Latin American states are primarily 5 Lead ores
6 Aluminium oxide
importers or exporters or both.
7
Sulphur
It is immediately clear that three product
groups require further disaggregation. Two of Table 2. The product groups for China’s
these are unsurprising: the length of the exports
country lists for textiles and clothing probably
Product groups
SSA/Latin America
analysed as:
arises partly from the pragmatic decision not
Importers
Exporters
to aggregate 6-digit items but to work at 2
1 Leather and articles
digit chapter level, excluding only cotton and
2 Textiles
wool fibres. Ferrous metals also need to be 3 Clothing
split. China is a substantial and fast-growing 4 Footwear
importer and exporter: further disaggregation 5 Aluminium
is needed to distinguish more clearly imports 6 White goods
7
Brown
goods
and exports.
6
8
Ferrous metals
8
Ferrous metals
Figure 1. Chinese imports
Product groups
Importers
Exporters
Significant players
1
2
3
Bolivia
Colombia
Uruguay
Argentina
Chile
Colombia
Nigeria
South Africa
Argentina
Brazil
Burkina Faso
Ecuador
Ethiopia
Nigeria
Sudan
Tanzania
Brazil
Niger
Venezuela
4
6
7
Brazil
DRC
South Africa
8
Brazil
South Africa
Chile
Peru
South Africa
Zambia
Brazil
Guinea
Venezuela
Angola
Chile
Ecuador
Ghana
Kenya
Nigeria
Peru
Sudan
Argentina
Brazil
Colombia
Mauritania
South Africa
Venezuela
Zimbabwe
Figure 2. Chinese exports
1
2
3
4
Mauritius
South Africa
Angola
Argentina
Benin
Bolivia
Chile
Colombia
Congo
Ecuador
Ghana
Kenya
Madagascar
Mali
Mauritania
Mauritius
Niger
Nigeria
Senegal
Sudan
Venezuela
Angola
Benin
Cameroon
Chile
Congo
Ghana
Guinea
Mozambique
Nigeria
Sudan
Tanzania
Togo
Uganda
Venezuela
Angola
Argentina
Ghana
Nigeria
South Africa
Sudan
Argentina
Brazil
Ethiopia
Nigeria
Uruguay
Brazil
Zambia
Brazil
Brazil
Colombia
Kenya
Madagascar
Malawi
Mauritius
Peru
South Africa
Importers
Exporters
Significant players
Product groups
5
Angola
Chile
Colombia
Nigeria
6
Angola
Argentina
Chile
Ecuador
Ghana
Nigeria
Peru
South Africa
Sudan
Venezuela
Argentina
Brazil
Brazil
Cameroon
Ghana
Mozambique
South Africa
Venezuela
7
All
Angola
Bolivia
Chile
Colombia
Ecuador
Kenya
Nigeria
Tanzania
Brazil
Argentina
Brazil
South Africa
Venezuela
Widest trade balance effects
Table 3 summarises by country the information in Figures 1 and 2. It splits the effects into
those likely to favour a country’s trade balance (increased demand for their exports or lower
prices for their imports) and those likely to disfavour it (increased export competition in third
markets, or increased world demand for goods that are imported). Note that this neatly
sidesteps the tricky problem, noted above, of whether more, cheaper imports is an economic
as well as a trade balance gain. This cannot sensibly be handled at this level of analysis.
Two panes from Figures 1 and 2 contribute to the ‘favour’ column and the other two to the
‘disfavour’ one, with each being split into three sub-columns. The first two numbers in each
column show the frequency with which the country is cited in each of the two relevant panes,
and the third number is the sum of these two.
7
Some findings are as one might expect. Brazil Table 3. Trade overlap by country
Country
No. of cases in which China may
faces the widest range of trade loss products:
product:
10 in total. South Africa is well represented in
Trade balance
Trade balance
a
b
gain
loss
both columns but stands to gain in more
Africa
product groups than it loses. But there are some Sub-Saharan
Angola
0
6
6
1
0
1
results that are not intuitively obvious as well. Benin
0
2
2
0
0
0
1
0
1
0
0
0
Angola, Ghana, Nigeria and Sudan also have a Burkina Faso
0
1
1
0
0
0
high ‘gain’ incidence. Only two SSA and two Cameroon
Chad
0
0
0
0
0
0
Latin American countries have a larger number Congo Rep.
0
2
2
0
0
0
of product groups in which they may Congo DR
1
0
1
0
0
0
1
0
1
0
1
1
potentially lose than those in which they may Ethiopia
Ghana
0
4
4
1
1
2
gain.
Guinea
1
1
2
0
0
0
Of course, this initial analysis captures only
one criterion for relative importance. But the
process has collected the data to look more
precisely at value of trade (both absolute and as
a share of total imports/exports). In this way a
more complex and complete picture can be
built up, and a start is made in the section on
product profiles.
A picture of what? Of the countries/sectors that
appear the strongest contenders for much more
focused (and resource-intensive) country and
sector oriented work. The methodology used to
undertake this trade data analysis will have a
strong bearing on which countries emerge as
the higher priorities. This is illustrated in Table
4 which makes a preliminary comparison
between the pointers derived from this analysis
and the ones in Jenkins and Edwards (2004).
They summarise (Table 16) the China effect on
the countries they have studied in terms of: the
impact on their target countries’ exports (to
China), their exports of competitive products
on to third markets, and the effect of imports of
Chinese goods. These can be related to the
four-way classification in Figures 1 and 2
above.
Kenya
0
2
2
1
1
2
Madagascar
0
1
1
0
1
1
Malawi
0
0
0
0
1
1
Mali
0
1
1
0
0
0
Mauritania
1
1
2
0
0
0
Mauritius
0
2
2
0
1
1
Mozambique
0
1
1
0
1
1
Niger
1
1
2
0
0
0
Nigeria
1
6
7
2
1
3
Senegal
0
1
1
0
0
0
South Africa
3
3
6
2
2
4
Sudan
1
4
5
1
0
1
Tanzania
1
2
3
0
0
0
Togo
0
1
1
0
0
0
Uganda
0
1
1
0
0
0
Zambia
1
0
1
0
1
1
Zimbabwe
1
0
1
0
0
0
Latin America
Argentina
2
3
5
1
3
4
Bolivia
0
2
2
1
0
1
Brazil
4
0
4
2
8
10
Chile
1
5
6
2
0
2
Colombia
1
3
4
2
1
3
Ecuador
1
3
4
1
0
1
Peru
1
1
2
1
1
2
Uruguay
0
0
0
1
1
2
Venezuela
3
3
6
0
2
2
Notes:
(a) Trade balance gain = being an exporter of a good
that China imports; being an importer of a good that
China exports (figures exclude imports of brown
goods).
(b) Trade balance loss = being an importer of a good
that China imports; being an exporter of a good that
China exports.
Table 4 shows, for the nine countries that are common to both studies, whether or not the
broad thrust of the conclusions are the same. A tick indicates that both studies expect the
same relative scale and direction of effect; a cross indicates that they do not. Out of the 27
combinations, almost half have a cross!
This does not mean that one paper is right and the other wrong. It means that we are all still
trying to shine a better light on the potential incidence of the Asian Drivers’ impact on
developing country trade. A corollary is that a premature jump into country studies could
waste a lot of money fast.
8
Table 4. Preliminary comparison with Jenkins and Edwards
Common countries
Export boost
Cameroon
Competition in third
markets
Imports
Ethiopia
Mozambique
Nigeria
South Africa
Uganda
Bolivia
Brazil
Peru
Key: = similar conclusions
= 14
= dissimilar conclusions = 13
Incidence of trade effect by type
Whilst still at the level of impact breadth, it is likely that a trade balance improvement that
comes about from cheaper imports will have different policy implications than one that
results from increased exports. Figure 3 takes all countries with a score of two or more in
either of the ‘gain’ columns of Table 3 and plots them in a simple four-cell matrix to indicate
whether the gains arise primarily from lower import costs, from greater export revenue, or
from both.
Unsurprisingly, all of the SSA states apart from South Africa are in the bottom right cell,
indicating that their trade balance gains arise primarily from lower import costs – but so, too,
are four of the seven Latin American states. Although there has not yet been any absolute or
relative quantification in this paper of the scale of these effects, this simple sorting reinforces
the expectation that many more countries will be affected by China on the import than on the
export front.
Figure 3. Gains by type and country
South Africa, Venezuela
Export gain
high
Argentina, Brazil
low
Angola, Benin, Bolivia,
Chile, Colombia,
Ecuador, Kenya,
Mauritius, Nigeria,
Sudan, Tanzania
low
high
Import gain
Figure 4 performs the same exercise for the two types of trade balance loss: greater
competition on world markets with China’s exports, and potentially higher prices for imports
that are being sucked into China. The key feature of Figure 4 is the absence of any SSA
states other than Nigeria and South Africa. This will partly be a reflection of small country
size, but not entirely since the same broad value thresholds were used when classifying
countries in relation to trade balance gains. It is likely also to be a function of the commodity
composition of trade. China’s imports are those associated with a rapidly industrialising state;
since few SSA states fall into this category they are not competing for world supplies of the
same products. China’s exports are of manufactures. About the only manufacture of SSA that
9
is significant across several countries is clothing. As Figure 4 covers only countries with two
or more loss products, those states for which clothing is the only substantial manufactured
export are overlooked.
Figure 4. Losses by type and country
Brazil, South Africa,
Venezuela
Export loss
high
Argentina
low
Chile, Colombia, Nigeria
low
high
Import loss
To deal with this the top left cell should also include all SSA (and Latin American) states that
export clothing – and only clothing (among manufactures). These are the states listed in
Figure 2, column 3, bottom pane – and there are not so many of them; only four apart from
South Africa. The impact of China for most SSA states will be on potential rather than actual
exports: its emergence reduces the likelihood of competitive clothing industries emerging in
most SSA states. This is hardly recent news! It has been recognised for the past decade that
the ‘window of opportunity’ for SSA to establish itself in clothing was a narrow one. So for
most SSA states the China clothing effect will be positive for the trade balance: it will lower
the cost of imports and whilst it will reduce the chances of an export industry emerging, the
failure of one to appear despite a quarter century of support from OECD protection/
preferences suggests that such chances were already slim.
This is not intended to diminish concern in those few countries that do export clothing. The
negative impact in Mauritius, for example, may well exceed by far all of the potential
positive effects. Also it does not deal with the case of SSA domestically oriented clothing
industries. But it does suggest that this is a ‘problem’ that is specific to identifiable countries
and not to the generality of SSA states.
Product profiles
The clothing situation is widely understood, but what about the other product groups
highlighted in Figures 1 and 2? How directly related are Chinese and SSA/Latin American
products? How substantial is each country’s trade?
A particularly intriguing ‘focus product’ is animal feed: it is not something that has been
widely picked up and it is of potential interest to a different range of developing countries
from those for which the more widely reported mineral imports of China would be relevant.
The middle pane of Table 5 shows that the exports of feed by SSA states, whilst tiny in
comparison with those of Argentina and Brazil, are not insignificant. An export of $49
million for Ethiopia, which has increased by 27 percent a year since 1999, is not trivial.
The ‘problem’ is that the feeds being exported by Africa are not the same as those that form
the bulk of China’s imports and are being exported by Argentina and Brazil. Ninety-four
percent of China’s imports of the product group that we have dubbed ‘feed’ are of soya
10
beans. Most of the Latin American exports are also in soya beans, but almost none of the
SSA exports. For Africa the main export is sesame seeds. So large are China’s imports that
‘only six per cent’ of its feed imports is a sizeable amount: the increase alone in China’s
imports of sesamum seed between 1999 and 2003 was equivalent to over one-quarter of the
total 2003 exports of the product by the five SSA states in the table!
This is prima facie evidence to justify further research. The next step is to investigate the feed
market to determine whether the grains being exported from SSA are, indeed, direct inputs
into feed (or indirectly affected by feed because they are substitutes for feed grains) and
whether there are knock-on effects from the trend in the dominant grain. On the import side
the overlap between what China and developing countries are importing is identical. Bolivia,
Colombia and Uruguay, shown in the bottom pane of Table 5, have significant and rapidly
growing imports, primarily of soya beans. Any pressure on world prices could affect the
profitability of their meat industries.
Table 5. Feed
Total feed
Value 2003 ($000)
Avg. annual change
1999–2003 (%)
5,785,368
57
China imports
Exports by:
Argentina
2,115,067
Brazil
4,942,242
Burkina Faso
11,526
Ecuador
16,096
Ethiopia
48,640
Nigeria
21,536
Sudan
81,729
Tanzania
13,923
Imports by:
Bolivia
43,120
Colombia
56,363
Uruguay
111,297
Source: UNSD Comtrade database.
Of which soya beans
Share of 2003 value Avg. annual change in
(%)
value 1999–2003 (%)
94
57
42
31
7
20
27
-4
-3
3
100
100
—
90
—
0.03
—
—
42
31
—
27
—
7
—
—
18
40
30
100
98
99
18
41
30
Tables 6 and 7 cover metals, which are pretty straightforward. The main purpose of the tables
is to determine the absolute importance of each product for the SSA/Latin American
countries that trade in them and whether or not exports/imports are sustained (indicated by
the average annual change columns).
Exports of alumina by all three countries Table 6. Aluminium and alumina
Total alumina/aluminium
in Table 6 are significant and sustained. At
Value 2003
Avg. annual
$111 million, the export must be
($000)
change 1999–
2003 (%)
considered significant for Guinea. The
alumina imports
1,375,761
42
very rapid annual change in Mozambique Chinese
Alumina exports by:
aluminium exports probably reflects the
Brazil
301,550
12
Guinea
110,571
10
fact that production seems to have come
Venezuela
120,038
18
on stream in 2001. If it is the case that it is China aluminium exports
2,413,824
60
a relatively new plant, the implications of Exports by:
Argentina
153,068
24
the growth in China’s exports on to the
Brazil
876,750
-2
world market (up 60 percent a year!)
Cameroon
74,967
-7
probably need to be taken into account
Ghana
94,867
-5
Mozambique
564,393
1,500
when assessing long-term viability. The
South Africa
652,534
-2
same applies to investment on stream in
Venezuela
526,504
3
the other countries.
11
Ferrous metals, copper articles and cobalt are covered in Table 7, which serves merely to
confirm what might be expected. With the exception of ferrous metals (which need to be split
into component parts) the picture seems unambiguously to suggest growing demand from
China for minerals of which the countries in the table are significant and sustained exporters.
One might query, however, the position of copper in South Africa, given that exports have
been declining rapidly.
Table 7. Ferrous metals, copper articles and cobalt
Ferrous metals
Value 2003
Avg. annual
($000)
change 1999–
2003 (%)
14,622,995
55
Chinese imports
Exports by:
Argentina
340,680
Brazil
4,035,545
Chile
Colombia
394,879
Congo DR
Congo Rep.
Mauritania
24,258
Peru
South Africa
1,861,415
Venezuela
788,255
Zambia
Zimbabwe
116,685
Source: UNSD Comtrade database.
Copper articles
Value 2003
Avg. annual
($000)
change 1999–
2003 (%)
3,074,780
43
Cobalt
Value 2003
Avg. annual
($000)
change 1999–
2003 (%)
94,684
65
17
11
4,658,909
8
22
54,340
22,811
2
204
21,683
11
74
8
24
763,027
55,852
5
-22
240,378
-1
4
Table 8 deals with leather and leather goods. Three-quarters of China’s exports are in leather
goods rather than unprocessed leather, whereas most of the developing countries’ exports are
in the latter. Although leather does not feature in Table 1 as one of the product groups for
which China’s role as importer is of interest (because it fell below our threshold of absolute
value in Step 1), its imports of bovine leather are none the less significant ($1.3 billion in
2003) and rapidly increasing (up 56 percent a year since 1999).
The table suggests that the next area for research is into the various leather value chains, to
determine whether SSA/Latin American countries are benefiting from increased demand for
their exports of leather or whether differentiation within the market means that China’s
processed exports are based on raw materials sourced in distinct sub-markets. If it is the
former, which seems the more likely, then the countries represented in the table will
experience a shift in the relative price of the unprocessed and processed product in favour of
the former. This could lead to a change in domestic industrial structure.
Table 8. Leather
Total leather
Value 2003 ($000)
Avg. annual change
1999–2003 (%)
3,160,426
49
China exports
Exports by:
Argentina
591,324
Brazil
667,683
Ethiopia
6,237
Nigeria
56,181
Uruguay
239,516
Note:
(a) Excluding items classified elsewhere, e.g. leather footwear
Source: UNSD Comtrade database.
3
41
37
32
16
a
Of which leather goods
Value 2003 ($000)
Avg. annual change in
value 1999–2003 (%)
73
44
8
2
0.1
0.02
10
34
24
-38
3
13
In addition to the products covered in Tables 5–8 there are the cases of footwear, clothing and
textiles to consider (as well as the miscellaneous categories of white and brown goods – too
complex to deal with in this introductory paper – plus the uncontentious product of sulphur).
12
Clothing has been considered above. The issues raised by China’s emergence on to the world
market following the expiry of the Multifibre Arrangement (MFA) have been well rehearsed.
The research that needs to be done now is to look at the textile industry and its link to
clothing. In particular it would be interesting to undertake work on value chains and
preference arrangements to determine whether or not the case for amending OECD rules of
origin (that, in the main, prohibit the use of non-originating cloth) has been strengthened by
the growth of Chinese exports. Far more SSA/Latin American countries are net importers of
textiles than are exporters. Indeed, with the exception of Zambia (cotton yarn) only Brazil is a
significant net exporter.
Whilst attention has been given (correctly) to the competitive threat posed by China’s
clothing exports, perhaps equal attention now needs to be given to the possible competitive
advantage resulting from cheaper cloth for Latin American/SSA clothing industries. If these
industries are prevented from using the cloth solely by onerous origin rules in their export
markets, then the case that such rules are prejudicing development becomes even stronger.
In the case of footwear, further analysis is required at the micro level within the context of
value-chain research. Any negative ‘China effect’, if there is one, is less likely to be on the
trade balance than on the division of labour within value chains. Of the countries considered,
Brazil is the only significant net exporter (with Ecuador also having very modest net exports
in 2003), although these ‘net figures’ probably mask a certain value of exports from a much
wider range of countries (which in all except the two cited are smaller than their imports).
Although this would not alter the conclusion that for all countries except Brazil (and to a very
minor extent Ecuador) the trade balance effect of China in footwear is positive, one would
want to check also the effect on industrial structure in significant producing states.
13
References
Jenkins, R. and Edwards, C. 2004. ‘How Does China’s Growth Affect Poverty Reduction in
Asia, Africa and Latin America? Expanded report to DFID’, December (mimeo).
Norwich: University of East Anglia, Overseas Development Group.
14