Repositório ISCTE-IUL

: Adopting a long term perspective, we evaluate the trade performance of less developed African countries. Besides some general trade indicators, we apply a constant market share analysis in order to decompose export performance into several components with specific economic interpretation. Our main conclusions are: (i) the sectoral specialization structure of exports has remained heavy in commodities but the composition of the basket of goods exported has changed considerably with a very strong concentration in crude oil (mainly in the last two decades), (ii) the geographical structure of exports has also changed, with an important increase of the relative importance of China and USA, (iii) the countries under analysis not only show a negative competitiveness effect, but are also penalized by their sectoral and geographical specialization, and (iv) the most favorable evolution is observed in the most recent sub-period (2000-2007), but it is insufficient to reverse the previous negative trend.


Trade Performance of the Less Developed African Countries 1. Introduction
We analyze the export performance of the Less Developed African Countries (LDAC) 1 over four decades, from 1967 to 2007.The period starts after most of these countries gained independence from colonial powers and covers the profound change in their economic strategies that occurred between the 1980s and the 1990s.In addition, the analysis extends to the commodity boom of the last decade.
Trade policy of the LDAC can be characterized by three major stages (UNCTAD, 2008).Before 1960, African countries' trade was mainly based on exports of primary products and imports of manufactures in a bidirectional relationship with the colonial powers.Subsequent to political independence, most of these countries adopted an import-substitution industrialization strategy, envisaged as a way to promote national production and reduce dependency on primary products.Nevertheless, faced with the poor results of this strategy and the ensuing loss of competitiveness, aggravated by the oil crisis of 1973 and 1979, which reduced demand for most primary products and resulted in decreasing prices, by the mid-1980s most of these countries began to adopt an export-oriented strategy based on market-oriented reforms, as advocated by the International Monetary Fund and the World Bank.By the mid-1990s, most African countries had gone through a structural adjustment programme designed by these institutions.With regard to the reform of the external sector, the guidance was to depreciate overvalued currencies, eliminate foreign exchange rationing, streamline import licensing and replace quantitative restrictions by tariffs (to reduce tariff dispersion and the overall level of tariffs), and to reduce or eliminate export taxes and de-monopolize trade.Many countries also adopted measures to promote non-traditional exports such as duty drawbacks, the creation of export-processing zones, and the promotion of foreign investment (UNCTAD, 2008).
Taking into account the heterogeneity of the period under review, the expectation is for significant changes in the pattern of trade and export performance of LDAC.The 1 This is a group a countries so designated in the Chelem database (which is used in this study), consisting of Angola, Benin, Burkina Faso, Burundi, Cameroon, Cape Verde, Central African Republic, Chad, Comoros, Ivory Coast, Democratic Republic of Congo (formerly Zaire), Djibouti, Equatorial Guinea, Eritrea, Ethiopia, Gambia, Guinea, Guinea-Bissau, Kenya, Liberia, Madagascar, Malawi, Mali, Mauritania, Mozambique, Niger, Rwanda, Sao Tome and Principe, Senegal, Sierra Leone, Somalia, Sudan, Tanzania, Togo, Uganda, and Zambia.export-oriented strategy was designed to benefit the tradable sector, diversifying the productive structure and increasing exports.
Nevertheless, LDAC's trade structure did not change substantially during the period analyzed, as it is made up basically of primary products, leaving those countries dependent on volatile global commodity prices.Indeed, the trend is an increasing concentration in some primary commodities, especially oil.Also, these countries decreased their market share in global trade during the period analyzed, as shown in this study.The strategy for promoting exports therefore did not produce the expected results with regard to improved export performance and diversified production, as amply illustrated in UNCTAD (2008).
While the above-mentioned empirical evidence on Africa's role in the international arena is uncontested, less agreement exists about its determinant factors.Some studies have sought to evaluate the role of protectionism in OECD markets.Indeed, Africa's major trading partners have very high non-tariff measures, such as sanitary and phytosanitary measures, social and environmental measures, or strict rules of origin (Mutume, 2006).However, evidence on whether Africa's protectionist policies has caused its marginalization in world trade is not clear-cut, as OECD trade preferences made market access for Africa more favorable than for many other exporters (Ng and Yeats, 1997).In turn, development economists and international organizations argue that inappropriate domestic policies are largely responsible for unfavorable trade and economic trends.In fact, many African countries have implemented the trade strategy reform in a very limited way.Tariffs remain high, trade monopolies continue to exist, export crops are still taxed, and trade procedures continue to be characterized by red tape and corruption.
Despite strong pressure from donor governments and multilateral agencies, the African leaders in general have been reluctant to open their economies and reduce the role of the state.For that reason, the reforms have progressed very little, with shy advances and setbacks (Rodrik, 1998).Overvalued exchange rates and high transaction costs (in part due to geography as these countries are far from destination markets (Redding and Venables, 2004), but also largely derived from very poor infrastructures and internal barriers to trade) have given Africa a comparative disadvantage in 'transactionintensive' activities, as is the case of manufacturing.With transport unreliable, firms typically need to carry very large stocks of inputs in order to maintain continuity and this problem is aggravated particularly in the case of manufacturing, since it tends to have a high share of intermediate inputs production (Collier and Gunning, 1999).
Weakness in institutions, namely at the level of property right protection and functioning of the courts, making the firms reluctant to invest, and insufficient human capital, are still other weaknesses commonly stressed.But it is also possible that LDAC are more specialized in export products and destination markets where demand is weak in comparison to other products and markets.If this is the case, their aggregate export share will tend to decline.This paper helps to understand the export performance of LDAC throughout the period analyzed by differentiating between competitiveness and structural factors, geographical and sectoral specialization patterns that drove the export market share growth of these countries in the period analyzed.For this purpose, we use the constant market share (CMS) accounting method in the version proposed by Cabral and Esteves (2006).This method allows decomposing the export performance at the level of the product/country of destination and for each individual market of destination (i.e., by product and country of destination).
The remainder of the paper is organized as follows.Section 2 characterises the LDAC export performance over the period analyzed through the use of some simple international trade indicators.Section 3 presents the methodology of constant market share, which is applied to LDAC in Section 4. Section 5 presents some final remarks.

The LDAC Export Performance: Some Preliminary Evidence
The empirical analysis performed in this paper is based on data drawn from the Chelem database.Specifically, we use a predefined group of countries -the Less Developed African Countries -and consider their exports over the period 1967-2007 to their 36 largest destination markets, covering almost all the exports of LDAC by the end of the period under analysis.The main contribution of our study is the application of a constant market share analysis to these data in order to provide an in-depth understanding of the factors behind the evolution of trade performance of this group of countries.
We begin our empirical exercise by considering some simple international trade indicators, aiming to provide an overview of the integration of LDAC in the international trade arena.Table 1 shows the evolution of the world market share of the LDAC over the period under scrutiny.LDAC's global export market share (MS) decreased from 1.82% in 1967 to 0.71% in 2007.This evolution represents a market share loss of 61.10% since 1967, a decreasing trend reversed only in the sub-period 2000-2007.Nevertheless, despite the recuperation in this last sub-period (with an increase of 92.03%), the year 2007 recovers only the 0.70% market share level of 1980.
[Insert Table 1 here] The main thing this table shows us, therefore, is that the relative importance of LDAC in world trade has been falling substantially over the period considered, even after trade liberalization, implemented in the 1980s, with the replacement of the importsubstitution industrialization strategy widely accepted in the 1960s and 1970s.
However, in the period 2000-2007 there was strong growth in the LDAC's world market share.As shown below, it is explained principally by the commodity demand boom of fuels, minerals, and other primary products that began in 2002.
Export diversification of the LDAC is very low, as shown in Table 2.These countries remain essentially primary commodity exporters and depend on a small number of products whose demand increased throughout the period analyzed.The weight of the top ten sectors in each year ranges from 86.09% in 1980 to about 94.23% in 2007, being very high since the beginning of the period studied.The export concentration reached the highest level at the end of the period analyzed, showing that these countries became increasingly dependent on a limited number of products. 2Insert Table 2 here] At the bottom of Table 2 we show the weight of the ten largest sectors in every year as well as the weight of the same sectors in the remaining years.From that, we see important changes in the composition of the exports.For example, while the top 10 sectors in 1967 represent 93.11% of total exports in that year, the exports of these sectors in 2007 have a value as low as 18.36%.The greatest changes relative to the beginning of the period analyzed occurred after 1990 and became more pronounced as we approach the end of the period.Noteworthy is the case of crude oil, which is not on the list of the ten most exported products in 1967.It occupies third place in 1980 and thereafter passes to the first position in the ranking.
To evaluate the degree of transformation of the sectoral export structure of the LDAC, we use the Lawrence index ‫ܮ(‬ ௌ ), which compares this structure at two different moments in time (which we define as 0 and 1): (1) j represents the product.This index ranges between 0 and 1, increasing with structural transformation.The results are reported in the first line of Table 3.This evidence shows that the decade with the least change in the sectoral structure was the last one, followed by the post-independence period.The biggest changes occurred in the 1980s and 1990s.
This may be a result of an economic strategy inflection, more favorable to exports, but it should be noted that the change is not very large relative to the previous period.
[Insert Table 3 here] The geographical pattern of destination countries of LDAC exports has changed considerably over the four decades analysed, as shown in Table 4 (which presents the destination countries ranked by their importance in 1967).
[Insert Table 4 here] The main conclusion emerging from this evidence is the decreasing share of the European countries and the importance of North America and Asia as destination markets.This trend reflects the gradual loosening of the economic ties of African countries with previous colonizers, in spite of the Preferential Agreements established by Europe with African countries (African, Caribbean, and Pacific Group of States).In turn, LDAC gradually reinforced their commercial relations with the USA, representing almost 24% of total LDAC exports in 2007.The growth of this market is a result of increased sourcing of oil and the implementation of the African Growth and Opportunity Act of 2000, the flagship of USA commercial and development policy with Sub-Saharan Africa, mainly based on trade preferences (Brenton and Hoppe, 2006).
Still more significant was the increased importance of Asian markets, mainly China, as these countries became importers of raw materials, amounting to 27.22% of total LDAC exports in 2007, followed by India, South Korea, Malaysia, Indonesia, Taiwan, and Thailand, with weights much smaller but still important.
Finally, it appears that intra-African trade remains marginal.Two reason are usually advanced for this: (i) African countries tend to export products similar in nature, and (ii) high transaction costs due to geography and poor infrastructures, limited regional integration, and high barriers to trade.
To evaluate the degree of transformation of the geographical export structure of the LDAC over the period analyzed, we once more turn to the Lawrence index, but now considering its spatial version ‫ܮ(‬ ீ ).We apply the measure presented in equation ( 1) but instead of ݆ for the product we consider the index ݅ for destination markets (݅ = 1, … , ‫.)ܫ‬ The results obtained from the application of this measure are in the second line of Table 3.
Comparing this evidence with that obtained through the Lawrence index applied to the sectoral structure, we see a greater stability in the present case.We also see that the LDAC underwent the most profound change in their geographical export pattern in the decade 1990-2000, followed by the last sub-period analyzed (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007).The 1980s is the period showing the most stable geographical pattern.This is in line with the increasing importance, in more recent sub-periods, of non-traditional African trade partners, like the USA and Asian countries.
In order to obtain a more complete understanding of the transformations in the exports from LDAC, we conclude this section by focusing our attention on the exports to each market individually, i.e., each combination country-sector (Table 5).

[Insert Table 5 here]
The evidence in Table 5 confirms the strong increase in the concentration of exports over the period analyzed, particularly crude oil.This trend is so strong that by 2007 the ten largest individual markets are related to the export of this product, a remarkable fact considering that oil accounted for only 0.18% of exports of LDAC in 1967.We should also note the entry of Asian customers from 2000 on, especially China, but also Korea The trend of concentration of exports identified above over the period under consideration is reflected in Table 6, which indicates the degree of concentration of exports, evaluated through the well-known Herfindahl index (for the three levels analyzed: destination countries, sectors, and individual markets).
[Insert Table 6 here] In all cases concentration increased, and is especially evident for sectors (0.55 at the end of the period), which corroborates the analysis above.
This methodology disaggregates the trade data of a given country (or group of countries as in the present case) and compares it with the trade flows of the rest of the world (Skriner, 2010).A spirited methodological debate during recent decades has produced a variety of versions of CMS analysis (Ahmadi-Esfahani, 2006).In order to decompose the export market share performance of LDAC, we follow the CMS analysis proposed by Cabral and Esteves (2006).This is a slightly adapted version of the formulation developed by Milana (1988).The formulation adopted here decomposes the variation registered in the export market share of a country (in our case, the LDAC) into several terms.
In terms of notation, we define ji X as the exports of LDAC of product ݆ to country ݅ and ji M as the imports of product ݆ by country ݅.Therefore, the market share of LDAC in country ݅ concerning product ݆ -‫ܵܯ‬ -is defined as: ( Total market share is expressed as: The next step involves calculating the percentage variation in the total market share as well as its decomposition into three main effects: the market share effect ‫,)ܧܵܯ(‬ the combined structure effect ‫,)ܧܵܥ(‬ and the residual effect ‫.)ܧܴ(‬ The ‫ܧܵܯ‬ captures the evolution of the share in each specific market (i.e., a country/product pair) weighted by the relative importance of that market in the total exports of LDAC.By capturing the effective variations registered in each individual market, this component can be interpreted as an indicator of the economy's competitiveness.For this reason, this is also called the 'competitiveness effect', expressed as follows: The combined structure effect ‫)ܧܵܥ(‬ captures the relative evolution of each destination market -translated into the variation in the relative weight of that market in the total imports -weighted by the relative importance of that market in the total market share of the country under analysis (in our case, LDAC).Thus, the ‫ܧܵܥ‬ measures the impact of the sectoral and geographical specializations on the variation of the LDAC's market share, being expressed in the following way: (5) Finally, the residual effect ‫)ܧܴ(‬ retains the cross-variations, which permits the full decomposition of the market share variation.It is expressed as: Thus, the total variation of the market share can be represented as follows: In order to deepen our understanding of the international trade dynamics of LDAC, we may improve the method applied until now through a more refined decomposition of the ‫.ܧܵܥ‬This will allow us to distinguish between the effect generated by the sectoral structure and that which arises from the geographical specialization.Consequently, the ‫ܧܵܥ‬ can itself be broken down into three components: the sectoral structure effect ‫,)ܧܵܵ(‬ the geographical structure effect ‫,)ܧܵܩ(‬ and a residual term -the mixed structure effect ‫.)ܧܵݔ݅ܯ(‬ The sectoral structure effect ‫)ܧܵܵ(‬ captures the part of the total variation of the market share that results from the specialization by products of the exports.This is expressed as: in which: In an analogous manner, the geographical structure effect ‫)ܧܵܩ(‬ captures the total variation of the market share that is due to the geographical specialization of the exports.It is defined as follows: in which: Since the two previous effects are not independent, we need to introduce the mixed structure effect ‫,)ܧܵݔ݅ܯ(‬ allowing a total decomposition of the ‫.ܧܵܥ‬Therefore, we have:

Decomposing the LDAC Export Performance -Evidence
Using the methodology presented in the previous section, we now evaluate the main factors that explain the evolution of LDAC export performance over the four decades considered.The method is applied at the most disaggregated sectoral level available using the CHELEM database (i.e., 72 sectors).
Table 7 presents the CMS results obtained for the total period considered as well as for the four sub-periods, in order to obtain a clearer understanding of the evolution of the trade performance of these countries.
[Insert Table 7 here] As shown in Table 7, the LDAC's global export market shrank over the period 1967-2007, mainly the result of the negative contribution of the competitiveness effect (-0.710) followed by a negative contribution of the ‫ܧܵܵ‬ (-0.525).This evidence allows us to conclude that during the period under review, the sectoral orientation of the LDAC's exports was mainly toward products with a lower growth rate relative to world trade growth, i.e., the sectoral specialization of these countries penalized their export performance.
Disaggregating this analysis by decades, we observe a positive ‫ܧܵܯ‬ in the 1980s and after 2000.While in the first sub-period the reason may be related to the reorientation of the strategy toward export promotion, the positive competitiveness effect in the more recent sub-period appears to be mainly explained by the price effect of the commodity boom.
Regarding ‫,ܧܵܵ‬ it is negative except for the last decade.Once more, this positive effect is directly related to the increased demand of commodities during this sub-period.
Finally, ‫,ܧܵܩ‬ which is related with the growth of the destination markets, is positive only in the first and last sub-periods.While in the case of the first sub-period a positive effect can be explained by the still important trade with the former colonizer, the most developed countries at this time, during the most recent sub-period it is basically due to the new Asian trade partners and, to a lesser extent, to increased trade with the USA, as described above (Section 2).
It should be noted that the last sub-period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007) not only has positive geographical (0.140) and sectoral (0.201) effects, but also a positive (and larger) competiveness effect (0.360).These positive developments in the last decade are the result of higher export concentration in oil for some dynamic markets.This, however, is an obvious and important fragility of a productive structure based almost exclusively on one natural resource.Moreover, it is a natural resource for which there are other important markets worldwide (some emerging, as is the case of Brazil) and that may be substituted by other energy sources.
Let us now consider, to conclude, the geographical and sectoral effects disaggregated by destination country and sector.This evidence is presented in Tables 8 and 9, respectively.For the sake of simplicity, we show the result of this disaggregation for only the five main positive and negative contributions for the global values of each effect.
[Insert Table 8 here] From the results shown in Table 8 it can be seen that there is considerable change in the relative importance of the main destination countries over the period chosen for the analysis.The trend we see makes clear the loss of importance of European countries (France and Italy stand out but only in the first sub-period, while France, the UK, Germany, the Netherlands, and Belgium-Luxembourg are part of the list of countries with the largest negative contributions during this period) and the increased importance of the USA, which is the country with the largest positive contribution between 1980 and 2000, and especially China, which will occupy a prominent place, increasing from 1980 to reach the top position in the last decade.We should also note the recent loss of importance of the USA, as in the last decade this country ranks first with respect to the largest negative contributions.
In terms of the geographical effect, it is also worth highlighting the loss of importance of the former colonizer.Thus, the UK, which was the first destination of exports in 1967 (Table 4), fell from 15.82% of LDAC's total exports in that year to 1.09% in 2007.From Table 8 it is clear that this country is the largest contributor to the negative geographical effect throughout the whole period analyzed, with most of this effect occurring in the first sub-period, i.e., 1967-1980.France also falls in relative terms, from a weight of 12.81% of total exports in 1967 to 3.93% in 2007.This is the country with the second largest negative contribution throughout the period studied, with emphasis on the 1990s.
Finally, disaggregating the contributions for the sectoral structure (Table 9), we see, as expected, the crucial importance of crude oil to the growth of LDAC's market share, mainly in the last two decades, in which it occupies the first place among the products with the largest positive contribution.

Final Remarks
Decomposing LDAC exports reveals that the declining importance in global trade of this group of countries since the beginning of the period under observation is in part due to the inability to remain competitive in international markets.With the turn of the century, a positive competitiveness effect is evident, but there are reasons to believe that this is the result of the commodity price boom in this period.Another reason for the poor export performance in the 1980s and 1990s is the weak demand of the destination markets: the geographical effect has a positive impact only during the first sub-period, in which the European partners, especially the former colonial powers, still had importance as trade partners, and in the 2000s, which is related to the commodity boom led by the dynamic Asian markets.
The specialization pattern gives a positive contribution to export performance only in the 2000s, i.e., precisely the period of the commodity demand boom.Indeed, the increased export performance in the last decade, in contrast to previously analyzed subperiods, is mainly the result of higher export concentration in oil and for some (few) dynamic markets.This, however, is the fragility of a productive structure based almost exclusively on just a single natural resource.
As mentioned in the Introduction, the weakness of the manufacturing sector and the poor export capacity of the LDAC appear to be the result of a low productivity trap, which only coherent programs to stimulate the supply response can solve.The commodity price boom of the oil and other primary products that drove LDAC exports has had the undesirable effect of increasing the concentration of exports of traditional products in terms of both products and target markets.But it also possible to envisage the current trend with optimism, as it opens to these countries a window of opportunity to finally transform their export pattern.The trick is to properly use the increased financial resources in order to broaden the export basket by sowing the seeds of a diversified industrial structure.1967-1980 -61.49 1980-1990 -25.99 1990-2000 -28.92 2000-2007 92.03 1967-2007 -61.10    Lawrence index 1967-1980 1980-1990 1990-2000 2000-2007 1967-2007 For sectoral export structure ‫ܮ(‬ ௌ )

and
Taiwan in 2000 and India in 2007.The bottom of Table5, which records the weight of the top individual markets of each year in the remaining years, puts into evidence the strong concentration of exports in the ten most important markets from 1980 to 2007, reaching 68.26% at the end of this period.The composition of the most important exports, however, varied substantially over the period.For instance, while the top individual markets represented 45.32% of total exports in 1967, this same basket represents only 3.81% of total exports at the end of the period.For their part, the ten most important individual markets in 2007 have a weight of 0% of total exports in 1967.A final observation concerns the strong concentration of oil exports in some target markets: in 2007, the two main destination countries of this product (China and the USA) represent around 46% of total exports of LDAC, suggesting a dangerous dependence upon these specific markets.The Lawrence index displaying the degree of transformation of the LDAC's export structure by individual markets ‫ܮ(‬ ெ ) can once again be obtained using equation (1) considering the index ݉ (݉ = 1, … , ‫ܬ‬ × ‫)ܫ‬ for individual markets instead of ݆.The last line of Table3presents the results for this index.It points to a similar structural change in all decades studied, with a slight decrease, revealing more stability, in the most recent sub-period.

Table 3 :
Lawrence Index