Chapter 2 Can Africa Export Manufactures? Endowments, Exchange Rates and Transaction Costs

International Monetary Fund
Published Date:
August 2001
  • ShareShare
Show Summary Details


The key issue of how sub-Saharan Africa might build a strong comparative advantage in exports, especially of labour-intensive manufactures, preoccupies current debates on African development2. The concern stems from two turns of events. First, one of the most visible manifestations of the subcontinent’ s multifaceted development failures during the past 30 years has been its marginalisation in world trade, especially in the global market for manufactures. Second, recent development successes elsewhere have taught that export-oriented policies either have facilitated them, as in Korea and Chinese Taipei, or actually have generated export-led growth, as in Chile, Mauritius, Tunisia and the countries of Southeast Asia3.

Much recent research has focused on how the globalisation of trade and capital markets has affected Africa’ s comparative advantage in manufactured exports4, and on the subcontinent’ s resource endowments, location and geography5. This work—and, more important, the changing landscape of global trade and finance—have pushed the debate towards more specific strategic questions:

  • — Can Africa ever hope to have comparative advantage in manufactured exports?

  • — Can globalisation help, if not substitute partially for, traditional and usually complex strategies to achieve export-led economic transformation?

  • — Do poor African countries have scope, in a world of integrated capital markets, to jump-start their competitiveness in the “old-fashioned” way, with sustained real currency depreciation?

Export performance in many African countries has indeed responded to macroeconomic reforms, especially deep real exchange-rate depreciation in Anglophone Africa in the 1980s and in the CFA franc zone since 1994. Nevertheless, given the partial nature of the reforms and frequent adverse terms-of-trade shocks, the growth of both aggregate exports and especially manufactured exports has been neither deep nor stable (Rodrik, 1997). Even as Africa still works on regaining lost ground in international markets for its traditional exports, therefore, a consensus has begun to develop (although debate continues) that the ultimate policy goal should be to achieve significant export diversification by building new comparative advantage in non-traditional exports, including labour-intensive manufactures.

Manufactured exports (as well as some other non-traditional exports) can support sustained overall economic growth more effectively than traditional primary exports for at least three reasons. First, they likely will grow faster when the global economy expands, because they have higher income elasticities of demand. Second, with relatively higher price elasticities of both demand and supply, they are less susceptible to price swings. Third, the manufacturing sector offers much greater prospects for dynamic productivity gains. In the medium term and the long run, therefore, traditional primary exports should take the role of facilitating export diversification. In the short run, Africa should continue to consolidate recent gains by avoiding economy-wide indirect taxation of those exports, imposing only moderate and sector-specific taxes to finance export diversification.

This chapter looks closely at manufactured-export performance in a selection of African and other developing countries, taking account of endowments, geography and the potential effects of globalisation. It uses an empirical model to assess the implications of three views on development strategy, called here the endowment, transaction, and exchange-rate-led theories6.

The endowment theory, from Wood and Berge (1997), uses a version of the Hecksher-Ohlin model to argue that, under globalisation, human-capital and natural-resource endowments rather than labour and capital are the main determinants of comparative advantage in manufactured exports. The theory predicts that Africa, with its heavy natural-resource endowment and low stock of human capital, has basically no prospects in manufactured exports.

Taking another tack, and also using a modified Hecksher-Ohlin framework, Collier’s (1997) critique of the endowment theory argues that this prediction could be valid only if Africa had a massive Dutch-disease problem because of its rich natural resources, which the evidence does not support. Collier’ s alternative—the transaction theory—proceeds from an observation that manufacturing is one of the most transaction-intensive activities. It asserts that high transaction costs due to a poor policy environment have caused Africa’ s comparative disadvantage, at least in the short and medium terms. Collier proposes a strategy for building comparative advantage on the basis of increased integration of the African economies into the global trade and capital markets.

Elbadawi and Helleiner (1998) argue that, given Africa’ s current development levels, comparative advantage in exports should flow from sustained, policy-induced real exchange-rate competitiveness—until economies develop sufficiently to support a productivity-induced secular real appreciation. This real exchange-rate-led strategy recognises, following Collier, the dire need for re-capitalisation of the African economies to sustain export expansion and diversification. It suggests, however, that flexible, pragmatic approaches for integration with global capital markets may be needed, both to protect macroeconomic competitiveness and to avoid financial and currency crises.

Manufactured Exports in Africa and Other Developing Regions

Table 2.1 presents the basic patterns of change in manufactured-export (MNEX) performance in the 1980s and 1990s in a selection of 13 developing countries, of which seven are in sub-Saharan Africa, four in Asia and one each in North Africa and Latin America.

Table 2.1.Manufactured Exports in a Sample of Developing Countries(Amounts in millions of current US dollars; shares and growth rates in per cent)
Aggregate ExportsaManufactured

Share of Total

Exports in GDP
Share of Mfg.

Exports in GDP
Burkina Faso
1994/95 average274.9545.9013.102.19
Avg. annual growth (1984-95)8.5618.991.279.46
Côte d’lvoire
1994/95 average3 699.85494.5942.025.65
Avg. annual growth (1984-95)
1994/95 average2 815.06432.1736.086.50
Avg. annual growth (1984-95)6.0712.623.6014.09
1994/95 average2 179.551 013.4558.4127.18
Avg. annual growth (1984-95)15.9622.762.017.67
South Africa
1994/95 average31 122.2911 018.4558.4127.18
Avg. annual growth (1984-95)5.4612.420.168.07
1994/95 average898.3745.8719.511.31
Avg. annual growth (1984-95)15.043.0016.5510.59
1994/95 average2 677.20615.6626.359.88
Avg. annual growth (1984-95)7.86431.004.574.76
1994/95 average5 056.543 925.0044.7623.19
Avg. annual growth (1984-95)9.9315.873.149.87
1994/95 average13 814.552 027.9628.443.34
Avg. annual growth (1984-95)13.5219.341.868.33
Republic of Korea
1994/95 average132 762.61101 757.8031.5824.23
Avg. annual growth (1984-95)15.6114.68-0.47-1.27
1994/95 average72 462.5149 200.7792.5962.89
Avg. annual growth (1984-95)15.1326.365.3616.11
1994/95 average62 558.6236 892.3340.0523.63
Avg. annual growth (1984-95)20.7829.936.1414.30
1994/95 average49 849.5521 825.3826.3711.55
Avg. annual growth (1984-95)8.6524.340.4115.75
Source: World Bank data.Notes:

As measured simply by growth in their MNEX/GDP ratios, Kenya, Tanzania, Burkina Faso and South Africa were the best performers among the African countries. Taking levels of MNEX/GDP into account, Mauritius also made an impressive showing, as it maintained average annual growth of MNEX/GDP at 7.7 per cent between 1984 and 1995 with the already high share of MNEX in its economy exceeding 27 per cent in 1994/95. The same point applies, to a lesser extent, to Kenya and South Africa. Burkina Faso and Tanzania, on the other hand, achieved relatively fast growth but started from low levels. Côte d’lvoire and Zimbabwe saw much slower MNEX growth relative to GDP and began from relatively low to moderate levels as well.

Overall, the sub-Saharan group did not perform as well as the sample countries in other regions. Except for Mauritius, they all have much lower MNEX/GDP ratios than such world-class performers as Tunisia, Indonesia, Thailand, Korea and, especially, Malaysia. Their average MNEX growth rates (not measured relative to GDP) all lay considerably below those in the other countries of the sample. If manufactured exports—especially labour-intensive ones—likely offer the most efficient engine of growth for Africa, as they have for successful development elsewhere, the African countries need both to raise their MNEX growth significantly and to sustain it.

The remainder of this section reviews the extent to which the MNEX performances of countries in the sample co-varied with four sets of potentially determinant variables, to establish their analytical and policy relevance. The four are exchange-rate policy, transaction costs, stocks of skills relative to natural-resource endowments and aggregate investment. The first three correspond to the three strategic views or theories discussed above; the fourth is associated with overall economic performance, including manufactured-goods export growth7.

Figure 2.1 depicts average MNEX/GDP ratios (MXY in the figure) for 1990-95 for several countries, together with indexes of real exchange rates (RER), RER misalignments (RERMIS) and RER variability (RERVAR). It makes three important points. First, the six countries that maintained the highest ratios—Indonesia, Thailand, Tunisia, Korea, Mauritius and Malaysia—also had uniformly less RER variability. Among the many countries with MNEX/GDP ratios of less than 10 per cent, only Chile and South Africa achieved comparably high RER stability. Second, despite a tendency for national currencies to appreciate in real terms as MNEX/GDP moved to 20 per cent or more (as in Thailand, Tunisia, Korea, Mauritius and Malaysia), no clear pattern of rising appreciation appears as the ratio moves from very low levels to about 10 per cent. This may arise from the large representation in the sample of CFA countries, which have a fixed exchange rate vis à vis the French franc8. Third, and perhaps for the same reason, no evidence appears of any tendency for real exchange rates to become more overvalued or undervalued (RERMIS) as MNEX/GDP rises.

Figure 2.1.Real Exchange Rate and Manufactured Exports in Developing Countries, 1990-95


1. MXY is the ratio of manufactured exports to GDP, RERVAR is real exchange-rate variability, RERMIS is real exchange-rate misalignment, and RER is the real exchange rate.

2. NGA = Nigeria; GAB = Gabon; GHA = Ghana; TGO = Togo; TZA = Tanzania; ZMB = Zambia; CMR = Cameroon; MDG = Madagascar; MWI = Malawi; BFA = Burkina Faso; CHL = Chile; CIV = Côte d’lvoire; CAF = Central African Republic; SEN = Senegal; KEN = Kenya; ZAF = South Africa; ZWE = Zimbabwe; IDN = Indonesia; TH A=Thailand; TUN=Tunisia; KOR = Korea; MUS = Mauritius; MYS = Malaysia.

3. The index of real exchange-rate misalignment (RERMIS) is computed as (RER-ERER)/ERER* 100%, where ERER is a model-based index of the equilibrium real exchange rate. RERMIS is an index of the extent of undervaluation (negative of overvaluation) of the real exchange rate relative to the equilibrium level. Therefore, according to Elbadawi and Helleiner (1998), RERMIS should be positively and robustly associated with manufactured exports. The RERMIS and the ERER indexes are taken from Elbadawi (1998), which constructs these indexes for 63 developing countries, based on a panel-data model of the real exchange rate. Elbadawi’ s approach for modelling equilibrium real exchange rates is based on estimating RER levels consistent with “sustainable” current-account equilibrium (e.g. Edwards, 1997; Elbadawi, 1994; Williamson, 1994). Williamson (1994: p. 187), for example, recommends an approach for estimating “the set of real effective exchange rates (or paths) needed to achieve simultaneous internal and external balance by some date in the medium-run future, and to maintain balance thereafter”. This is the so-called fundamental equilibrium exchange rate (FEER). This concept calls for specifying (or assuming) behavioural specifications for the fundamentals and using the real exchange-rate equations in a bigger model to derive paths for the equilibrium real exchange rate, given the assumed paths of the fundamentals. The approach adopted by Elbadawi (1998) for estimating “sustainable” fundamentals resembles the FEER approach. It obtains the capital-account fundamentals using a model that links sustainable net capital flows and net foreign income to sustainable current account balance (Edwards, 1997), and sustainable change in reserves to long-term import requirements. In addition, it links sustainable foreign-aid ratios to levels judged as consistent with avoiding excessive aid dependency.

Next, consider the relationship between MNEX and aggregate investment. Even ignoring efficiency considerations, the share of gross investment in GDP is a useful broad indicator of an economy’ s potential to sustain high rates of both export and overall economic growth9. On this score, most of the African countries lag badly. Except for the 29 per cent investment ratio in Mauritius, all the others have investment rates lower than 25 per cent (Table 2.2). Investment performance in the four Asian countries in Table 2.2 provides a far superior support for exports. All have investment rates of 30 per cent or much more.

Table 2.2.Other Determinants of Manufactured Exports for a Sample of Developing Countries(See notes for units of measure. Ratios are in percentages, as are the growth rates, which are annual averages for 1984-95)
Ratio of Gross


Investment to GDPa
Ratio of School

Enrolment to

Land per Workerb
Fax Machines per

1,000 Peoplec
CorruptiondPaved Roadse
Burkina Faso
1994/95 average20.870.61n.a.417.35
Average annual growth4.2910.12n.a.-0.47
Côte d’lvoire
1994/95 average13.041.17n.a.2.799.50
Average annual growth3.42-0.04n.a.1.99
1994/95 average20.552.710.142.8113.70
Average annual growth1.240.1210.961.57
1994/95 average28.984.9517.003.1993.00
Average annual growth2.003.46177.08n.a.
South Africa
1994/95 average17.951.132.115.6441.50
Average annual growth-1.993.7024.44n.a.
1994/95 average23.373.280.072.564.20
Average annual growth3.14-2.5189.35n.a.
1994/95 average23.661.970.352.9451.45
Average annual growth3.686.1427.3546.09
1994/95 average24.301.352.532.9478.10
Average annual growth-2.915.0858.240.71
1994/95 average27.071.151.5513.80
Average annual growth6.90-7.0233.53n.a.
1994/95 average36.5511.378.672.3876.90
Average annual growth2.022.0510.821.40
1994/95 average41.963.983.974.7575.00
Average annual growth2.92-1.0070.811.41
1994/95 average40.941.771.483.1996.05
Average annual growth3.51.17126.2913.94
1994/95 average30.505.840.360.5645.85
Average annual growth2.023.9655.49-0.19
Source: World Bank data.Notes: n.a. = not available.

The transaction theory posits a negative relationship between transaction costs and MNEX. Figure 2.2 shows the association between the MNEX/GDP ratio and a composite index of transaction costs. The composite is a weighted index, for each country, of a qualitative indicator of corruption, a measure of paved roads and the number of fax machines. Table 2.2 contains the basic measures, and the weights are the corresponding coefficients from regression 4 in Table 2.3. The note to Figure 2.2 explains how the composite index was calculated. It can range from zero (no costs) to a maximum of one. The scatter fits a negative exponential curve, along which a value of about 0.5 for the transaction-cost index establishes a key threshold. Most countries with higher cost levels have MNEX/GDP values both low and fairly invariant with respect to changes in transaction costs. Eleven countries, all in sub-Saharan Africa, lie above this threshold. A second group has lower transaction costs, but also MNEX shares below the regression line; it includes Madagascar, Malawi, Central African Republic, Burkina Faso, Chile and, notably, South Africa. The dominance of mineral resources clearly contributes importantly to this outcome in both Chile and South Africa. Finally, the evidence from Tunisia, Thailand, Korea, Mauritius and especially Malaysia suggests a strong association between low transaction costs and high shares of manufactured exports in GDP.

Figure 2.2.Transaction Cost and Manufacturing Exports, 1990–95

Note: The composite index of the transaction cost is calculated as a normalised index of TC, where TC is given by: TC=1β1X¯1+β2X¯2+β3X¯3 where Xs are the average of LFAX, PROADS and LCORR, and βs are the estimated coefficients from Table 2.3.

Table 2.3.An Empirical Model of Manufactured Exports in Developing Countries


Log (MXY)
Equation 1

Equation 2

Equation 3

Equation 4

Log (INV/GNP)0.72662.21450.76372.3502
Log (TOT)-0.8786-1.7076-0.9398-1.8566-1.1567-1.9771-1.2207-2.1485
Log (SCH)1.08174.54450.88512.7426
Log (ARLAR)-12.6392-0.7393-11.6882-0.5641
Log (SCHLAR)1.07154.48270.85682.6891
Log (CORR)1.32655.63881.33045.7010
Log (PROAD)0.63334.78720.64194.9033
Log (FAX)0.46404.03910.48394.4217
Adjusted R20.86600.8722
P value0.00000.00000.00000.0000
No. observations82826464
No. countries41413232
Period of
1980-81, 1982-83,
1984-85, 1986-89,
Note: MXY, manufactured exports to GDP; RERMIS, real exchange rate misalignment; RERVAR, real exchange rate variability; INV/GNP, investment to GNP; TOT, terms of trade; TOTVAR, terms of trade variability; SCH, index of primary school enrolment; ARLAR, arable land to labour ratio; SCHLAR, schooling per land per labour ratio; OECYB, GDP of OECD countries; CORR, index of corruption; PROAD, paved roads; FAX, number of fax machines per 1 000 people; DSSA, dummy for sub-Saharan Africa; DEA, dummy for East Asia; DLAC, dummy for Latin America; P value refers to the Hausman test for fixed versus random effects models.

Finally, a version of the endowment thesis (see the first paragraph of the next section) predicts a positive relationship between MNEX/GDP and the stock of skills relative to natural resource endowments. Figure 2.3 indeed shows a strong, positive partial correlation between MNEX/GDP and a skills/resources proxy measure, albeit with a wide distribution around the mean at low levels of MNEX/GDP. The proxy is the ratio of school enrolment to land area, per 100 workers (the SCHLAR variable in Table 2.3).

Figure 2.3.Manufacturing Exports and Schooling per Land/Labour Ratio, 1990–95

Note: Schooling rate to land per labour ratio is measured by the index of primary school enrolment ratio to land area per 100 workers.

Econometric Analysis of Manufactured Exports in Developing Countries

Each of the three theories suggests a pivotal determinant of manufactured exports. First, the endowment hypothesis implies that a combination of high natural resources per worker (measured as land area per 100 workers)10 and low human capital per worker (measured by schooling per worker) should both be negatively associated with manufactured exports. Another version of the theory assumes that both of these factors have the same quantitative effect on exports, which leads to a restricted model with a single endowment variable, the ratio of human capital to land area per 100 workers. This model predicts that countries with higher human capital per worker relative to their per-worker natural-resource base have a comparative advantage in manufacturing. The analysis below assesses both versions of the theory.

Second, the transaction theory predicts that transaction costs dominate. The analysis uses three variables to account for them: an index of corruption, the length of paved roads and the availability of telephone and fax machines. It tests, at the first level, whether these three components have a stronger effect on manufactured exports than aggregate investment, to make the point that the components of investment which reduce transaction costs relax the most critical constraints that such exports face.

If the data corroborate the prediction of the transaction theory, the next step tests whether reducing transaction costs suffices for policy. A test for the significance of an index of real exchange-rate misalignment both accomplishes this and leads to a look at the third hypothesis, that an export orientation led by real exchange-rate policy can be effective. The test examines whether real exchange-rate misalignment matters for manufactured exports, regardless of whether the analysis controls for aggregate investment or transaction costs. Finally, and in addition to the pivotal variables that the three theories suggest, other variables included account for macroeconomic instability relevant to the export sector (real exchange-rate variability); external shocks (the level and variability of the terms of trade); external demand (per capita GDP in the OECD countries); and regional dummies for East Asia, Latin America and sub-Saharan Africa.

Table 2.3, introduced in the preceding section, provides estimates of manufactured export performance (the ratio of such exports to GDP, in logs to avoid picking up spurious effects) for a panel of 41 developing countries during 1980-9511. The first two equations are random-effects regressions, which include aggregate investment rather than transaction-cost variables12. The third and fourth regressions exclude aggregate investment and include the three transaction-cost measures13. All the right-hand-side variables other than relative prices are expressed relative to appropriate scale variables (see notes to the table).

All the regressions fit the data very well, explaining about 95 per cent of the variations in the MNEX/GDP ratio, MXY. First, regressions 1 and 3 suggest a significant and positive relation between the ratio of schooling per worker and manufactured exports, but land area per worker appears as insignificant. Regressions 2 and 4 reveal a positive and highly significant association using the ratio of schooling per worker to land area per worker as the independent variable. Because the schooling effect obviously drives the significance of this variable, this does not contradict the finding of insignificance for per-worker land area. Thus, adequately controlling for other relevant determinants, a high ratio of natural-resource endowment per worker does not appear to associate with manufactured-export performance across countries. This empirical evidence does not corroborate the endowment thesis. The result cannot suffice to reject the thesis formally, however, unless it is assumed, plausibly, that MXY correlates highly with aggregate (or primary) exports.

Second, the first two regressions find aggregate investment to be robustly and positively associated with manufactured exports. The third and fourth equations produce the same results for all three of the transaction-cost measures—and the significance levels for these variables are more than double those for aggregate investment. Regressions (not reported) accounting for the simultaneous effects of aggregate investment and transaction costs find only the latter to be significant. Third, RERMIS, measured as under-valuation of the exchange rate, is positively and significantly associated with exports in both the equations that control for investment and those that control for transaction costs. Therefore the combined results corroborate the basic prediction of the transaction theory: transaction costs are major determinants of exports of manufactures, and investment in reducing them generates the highest payoff in capacity to generate such exports. Moreover, the results support a view that real exchange-rate-based competitiveness is a pre-requisite for developing countries (especially low-income ones) to become successful exporters of manufactures.

Finally, all four regressions find real exchange-rate variability and the level of the terms of trade highly significant and negatively associated with manufactured exports. In addition, as expected, terms-of-trade variability has a deleterious effect on manufactured exports, although it was significant only in the aggregate-investment version of the model (regressions 1 and 2). Moreover, a less clear result from a theoretical perspective lies in the negative elasticity of the level of terms of trade. GDP per worker in the OECD countries (a proxy for external demand), however, was only marginally significant in the aggregate-investment version of the model and very insignificant in the transaction version, and it was dropped from regressions 3 and 4. Last, all regional dummies, especially the Africa dummy, were not significant—an important result, which suggests that Africa is on the regression line: the gap in performance between Africa and other regions, most notably East Asia, should be explained by differences in the global determinants of manufactured exports.

Why, then, was Africa marginalised in world manufactured exports? In the 1990s, manufactured exports by the four East Asian countries considered (Indonesia, Malaysia, Republic of Korea and Thailand) have accounted for more than 30 per cent of their GDP, while sub-Saharan African countries have managed to export only about 3 per cent of their GDP during the same period. Table 2.4—based on regression 4 of Table 2.3—simulates the sources that accounted for this outcome, and Figure 2.4 shows the net contribution of four categories of determinants: endowment, exchange-rate policy, transaction cost, and terms of trade.

Table 2.4.The Extent and Sources of Africa’ s Shortfalls in Manufactured Exports Relative to East Asia
VariableEast AsiaaSub-Saharan AfricaaDifferencebNet Contributionc
RER variability-0.1125-0.42380.31133.28
RER misalignment0.00700.0489-0.0419-0.44
Exchange-rate policy2.84
Terms of trade-2.4216-2.3919-0.0297-0.31
Terms-of-trade variability-0.0196-0.03490.01530.16
External terms-of-trade-0.15
Number of fax machines0.2189-0.28160.50055.28
Percentage of paved roads0.90240.62540.27702.92
Transaction benefits8.70
Skills-to-land ratio1.63931.51090.12841.36
Total predicted (MNEX/GDP)12.75
Source: regression 4 of Table 2.3.Notes:

Where yEA(yAFR) = (MANEX/GDP) in East Asia (Africa)

A simple Taylor’s expansion of log(yEAYAFR) around 1 leads to the following expression:

c) Column (4) gives the components of (yEAYAFR) on the basis of the above approximation.

The evidence very strongly corroborates the transaction theory. Lower transaction costs in East Asia relative to sub-Saharan Africa in the 1990s allowed its share of manufactured exports in GDP to reach as high as 8.7 times that in sub-Saharan Africa. The proxy measure of transaction costs, the number of faxes, accounts for half of the difference. One should interpret this result as a proxy for the overall effect on manufactured exports of communication-intensive inputs (such as managerial practices and flow of information). East Asia also outperformed sub-Saharan Africa in real exchange-rate stability, which more than compensated for Africa’ s advantage in exchange-rate competitiveness. The net effect of exchange-rate policy allowed East Asia to achieve manufactured export shares about 2.8 times those of sub-Saharan Africa. Assuming no differences between East Asia and Africa in other determinants, East Asia’ s superior performance in these two main sets of policy variables would predict its share of manufactured exports to be about 11.5 times that of sub-Saharan Africa.

Figure 2.4.The Extent and Sources of Africa’ s Shortfall in Manufactured Exports Relative to East Asia, 1990–95

Source: Table 2.4.

Conversely, East Asia’ s advantage relative to Africa in terms of the ratio of skills per worker to land per 100 workers (the endowment thesis) predicts the share of Asian manufactured exports to be about 1.4 times that of sub-Saharan Africa. The results also show that terms-of-trade effects favoured Africa, but the net effect was too small to make any measurable impact.


This chapter has analysed the determinants of manufactured exports in the countries of sub-Saharan Africa and other developing countries, guided by three pivotal views on the prospects for Africa in manufactured exports. First, according to Wood and Berge (1997), Africa cannot have a comparative advantage in exporting labour-intensive manufactures—even defining them broadly to include processed raw materials—because of its high endowment of natural resources relative to human capital. This thesis has implications for Africa’ s development dramatically different from the other two. Second, Collier (1997) argues that, for most of Africa, unusually high, policy-induced transaction costs are the main cause of its comparative disadvantage in manufactured exports. Both approaches flow directly from a specific interpretation of the Hecksher-Ohlin model, which makes the fundamental prediction that comparative advantage will reflect differences in relative endowments. The third view emphasises the necessity of stable and competitive real exchange rates for export profitability. The impact of globalisation heavily influences all three views.

The empirical results—based on a panel of 41 developing countries, with 11 in sub-Saharan Africa—suggest five important conclusions. First, after adequately controlling for other relevant determinants, a high relative natural-resource endowment does not associate robustly with the ratio of manufactured exports to GDP across developing countries. To the extent that GDP correlates strongly with aggregate (or primary) exports, this finding permits the conclusion that the empirical evidence does not corroborate the “endowment thesis”. Second, the results do support the basic prediction of the “transaction theory”, that transaction costs act as major determinants of manufactured exports and that investing in reducing them generates the highest payoff in capacity to export manufactures. Third, the results also lend support to the view that real exchange-rate competitiveness is a pre-requisite for a developing country (especially a low-income one) to become a successful exporter of manufactures. Fourth, Africa is not different. It lies on the regression line, which suggests that the gap between its performance and that of other regions, most notably East Asia, should be explained by differences in the global determinants of manufactured exports.

Fifth, the simulation exercise gives useful insight into why Africa is marginalised in world manufactured exports, with its shares of such exports to GDP in the 1990s at under one-tenth of the comparable East Asian share. Simulations of the net contribution of four categories of determinants—endowment, exchange-rate policy, transaction costs, and the terms of trade—provide very strong support for the transaction theory. The evidence suggests that bad policy, especially that affecting transaction costs, rather than adverse endowments, remains the most serious hurdle for Africa to leap before it can build comparative advantage in the international market for manufactured exports.


The author acknowledges, without imputation of responsibility, helpful comments from Mustapha Nabli and other participants at the workshop sponsored by the OECD, AERC and IMF, where this chapter was first presented. He also acknowledges research assistance by John Randa and Rajal Upadhyaya.

See, for example, World Bank (1998a and 1998b), Elbadawi (1998) and Sekkat and Varoudakis (1998).

Rodrik (1994, 1995) has shown that, perhaps unlike in many other recent development successes, sustained investment booms have driven both overall growth and phenomenal export expansion in Korea and Chinese Taipei; export orientation has helped to sustain high investment productivity.

See, for example, Collier (1997), Elbadawi and Helleiner (1998) and World Bank (1998b)

Examples here include Wood (1997), Wood and Berge (1997), Wood and Owens (1997) and Wood and Mayer (1998).

The analysis does not test these views directly, because the model estimates performance equations based on the ratio of manufactured exports to GDP, rather than comparative-advantage equations that would use, for example, shares of manufactured exports in aggregate exports as the dependent variable. For a detailed discussion and direct testing of the three views, see Elbadawi and Randa (1999).

Rodrik (1999) argues that a sustained rise in private returns to capital made possible the phenomenal export expansion of Korea and Chinese Taipei. The two economies engineered them using a range of strategic interventions, including investment subsidies, administrative guidance and the use of public enterprises. The same theme appears in his explanation of Africa’ s marginalisation in world trade (Rodrik, 1997), although he does not suggest that Africa should pursue similar strategies.

These countries (Burkina Faso, Cameroon, Central African Republic, Côte d’lvoire, Gabon, Senegal and Togo) have seen substantial real appreciation as well as RER overvaluation for most of the period since 1985 (Baffes, Elbadawi and O’ Connell, 1997).

According to robust evidence drawn from a vast set of developing countries, a 6 per cent real GDP growth rate would require about a 28 per cent rate of investment (Williamson, 1997).

This is a variant of population density, which also has been shown to be closely associated with the composition of exports (Perkins and Syrquin, 1989).

Data on manufactured exports and other related variables came from the World Bank’ s World Development Indicators. They allow estimation for a 41-country panel in regressions 1 and 2 and for 32 countries in regressions 3 and 4, for five periods: 1980–81, 1982–83, 1984–85, 1986–89 and 1990–95.

Hausman specification tests (reported in the table) suggest that random-effects results are superior to those based on fixed-effects regressions.

Because the number of telephones turned out to be consistently insignificant, it was dropped from the third of these measures, which retains only the number of fax machines.


    Baffes, J., I.Elbadawi and S.O’Connell (1997), “Single-Equation Estimation of the Equilibrium Real Exchange Rate”, Policy Research Working Paper 1800, World Bank, Washington, D.C.

    Collier, P. (1997), Globalization: What Should Be the African Policy Response?, Centre for the Study of African Economies, University of Oxford, Oxford.

    Edwards, S. (1997), “Exchange Rate Issues in Developing and Transition Economies”, inI.Elbadawi and R.Soto (eds.), “Foreign Exchange Markets and Exchange Rate Policies in Sub-Saharan Africa”, Journal of African Economies, Supplement to Vol. 6, No. 3.

    Elbadawi, I. (1994), “Estimating Long Run Equilibrium Real Exchange Rates”, inEstimating Equilibrium Exchange Rates, J.Williamson (ed.), Institute for International Economics, Washington, D.C.

    Elbadawi, I. (1998), “Real Exchange Rate Policy and Non-Traditional Exports in Developing Countries”, Research for Action, No. 46, WIDER, Helsinki.

    Elbadawi, I. and G.Helleiner (1998), “African Development in the Context of New World Trade and Financial Regimes: The Role of WTO and Its Relationship to the World Bank and the IMF”, paper presented at AERC Workshop on Africa and the New World Trade System, Mombasa, Kenya, April.

    Elbadawi, I. and J.Randa (1999), “Can Africa Export Manufactures? A Tale of Three Views”, World Bank, Washington, D.C.

    Perkins, D. and M.Syrquin (1989), “Large Countries: The Influence of Size”, inH.Chenery and T.N.Srinivasan (eds.), Handbook of Development Economics, North-Holland, Amsterdam.

    Rodrik, D. (1994), “Getting Interventions Right: How South Korea and Taiwan Grew Rich”, NBER Working Paper No. 4964, National Bureau of Economic Research, Cambridge, Massachusetts.

    Rodrik, D. (1995), “Trade Strategy, Exports and Investment: Another Look at East Asia”, Institute of Policy Reform Discussion Paper, Institute of Policy Reform, Washington, D.C.

    Rodrik, D. (1997), “Trade Policy and Economic Performance in Sub-Saharan Africa”, paper prepared for the Swedish Ministry for Foreign Affairs, Stockholm, November.

    Rodrik, D. (1999), “The New Global Economy and Developing Countries: Making Openness Work”, ODC Policy Essay No. 24, Johns Hopkins University Press, Baltimore.

    Sekkat, K. and A.Varoudakis (1998), Exchange Rate Management and Manufactured Exports in Sub-Saharan Africa, Technical Paper No. 134, OECD Development Centre, Paris.

    Williamson, J. (1994), “Estimating the FEERs”, inJ.Williamson (ed.), Estimating Equilibrium Exchange Rates, Institute for International Economics, Washington, D.C.

    Williamson, J. (1997), “The Washington Consensus Revisited”, inL.Emmerij (ed.), Economic and Social Development in the XXI Century, Inter-American Development Bank and Johns Hopkins University Press, Washington, D.C.

    Wood, A. (1997), “Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom”, World Bank Economic Review, Vol. 11.

    Wood, A. and K.Berge (1997), “Exporting Manufactures: Human Resources, Natural Resources and Trade Policy”, Journal of Development Studies, Vol. 34, October.

    Wood, A. and J.Mayer (1998), “Africa’s Export Structure in Comparative Perspective”, inEconomic Development and Regional Dynamics in Africa: Lessons from the East Asian Experience, UNCTAD, Geneva.

    Wood, A. and T.Owens (1997), “Export-Oriented Industrialization Through Primary Processing?”, World Development, Vol. 25, September.

    World Bank (1998a), “Africa Can Compete! A Framework for World Bank Group Support for Private Sector Development in Sub-Saharan Africa”, Private Sector Finance Group, Africa Region, World Bank, Washington, D.C.

    World Bank (1998b), “The Challenges of Globalization for Africa: Paper for Dakar Leaders Forum”, World Bank, Washington, D.C.

    Other Resources Citing This Publication