Chapter

9 Trade and Growth in Sub-Saharan Africa

Editor(s):
Zubair Iqbal, and Mohsin Khan
Published Date:
December 1998
Share
  • ShareShare
Show Summary Details
Author(s)
Benno J. Ndulu and Njuguna S. Ndung’u 

The relationship between trade and growth has occupied the development debate for a protracted period. Indeed, the vigor and interest characterizing the debate reflect both its perceived importance and continued elusiveness in settling the main contentious issues on the theoretical and empirical fronts. What appears to be gaining wide acceptance from cross-country evidence is that those developing countries that have been most successful in pursuing growth are also the ones that have taken most advantage of trade opportunities. These countries have experienced high rates of economic growth in the context of rapidly expanding exports and imports. The converse is that countries that have rigidly stuck to import substitution policies and maintained barriers to export have lagged behind. At least this is the long-term association that has been supported by a large range of cross-country studies including Edwards (1993), Tanzi (1995), Sachs and Warner (1995, 1997), and Papageorgiou, Michaely, and Choksi (1991). What is far less clear is the extent to which trade and trade policies have played a causal role, rather than being a facilitator of other more fundamental factors of growth (Rodrik, 1997).

Two significant observations have been made concerning sub-Saharan Africa’s trade and growth performance. One is that sub-Saharan Africa is increasingly being marginalized in world trade, as is evident from the decline of its share in world trade from 3 percent during the mid-1950s to the current 1 percent (Yeats and others, 1997). This marginalization in trade is blamed variously on the closedness of trade regimes (Sachs and Warner, 1995, 1997), the fact that the region’s economies have become more inward-looking at the same time the rest of the world is integrating into the world economy (Collier, 1995) and persistence of structural and trade policies that militate against international competitiveness (Yeats and others, 1997). A rather important finding in this regard is that by Rodrik (1997), which shows that the relatively slow growth of Africa’s GDP more than anything else explains the marginalization of the region in world trade. In an analysis that assesses Africa’s trade orientation status, vis-à-vis the region’s size (population), income per capita, and distance to the world’s leading trading nations, he finds that Africa trades as much as is to be expected given its geography and the level of income per capita; that is, it lies on the regression line. East Asia trades more than expected, while Latin America trades less. This is not to downplay the importance of trade policy, which he finds significant in explaining ratios of trade to GDP, but the finding underscores his earlier assertion that the other factors influencing growth augur well for trade performance.

Further information on sub-Saharan Africa’s trade performance can be gleaned from Figures 1 through 8. For the period 1970–94, export shares and import shares in GDP for 45 sub-Saharan African countries averaged 22 and 32 percent, respectively, compared with a sample of 99 developing countries as a group. The dispersion of performance among African countries is also very large, ranging from 3 to 60 percent for export shares and 8 to 114 percent for import shares (Figures 1 and 2). What is also revealing is the fact that the regional average trade performance relative to its output did not register as steep a decline over time, in contrast to its share in world trade. In this respect, the decline of export share during 1970–79 and 1990–94 was by approximately 10 percent (Figure 5), import shares remained more or less constant (Figure 6), while that of the share in world trade was much steeper as pointed out earlier. The gap with East Asia’s performance is however much more. One obvious conclusion from the above is that even if Africa’s shares of trade in GDP had remained constant over time, the fact that its output growth was much lower than that of other developing countries (see Figures 3 and 7) would still have resulted into the region’s marginalization in world trade. The reasons behind the much slower export growth relative to other countries or regions should be the focus on the link between openness as more broadly defined and growth rather than the trade shares per se.

Figure 1.Sub-Saharan African Countries’ Percentage Shares of Exports in Gross Domestic Product

(Percent)

Note: Mean is 22 percent; median is 18 percent. The variable plotted is XCDP.

Figure 2.Sub-Saharan African Countries’ Percentage Shares of Imports in Gross Domestic Product

(Percent)

Note: Mean is 32 percent; median is 25 percent. The variable plotted is MGDP.

Figure 3.Sub-Saharan African Countries’ Percentage Growth of Real Gross Domestic Product

(Percent)

Note: Mean is 3.8 percent; median is 3.6 percent. The variable plotted is RGDPGR.

Figure 4.Sub-Saharan African Countries’ Percentage Growth of Exports

(Percent)

Note: Mean is 3.0 percent; median is 2.7 percent. The variable plotted is XGDPGR.

Figure 5.Comparative Percentage Shares of Exports in Gross Domestic Product for Sub-Saharan Africa, Selected Periods

(Percent)

Note: The variable plotted is XGDP.

Figure 6.Comparative Percentage Shares of Imports in Gross Domestic Product for Sub-Saharan Africa, Selected Periods

(Percent)

Note: The variable plotted is MGDP.

Figure 7.Comparative Percentage Growth Rates of Gross Domestic Product in Sub-Saharan Africa, Selected Periods

(Percent)

Note: The variable plotted is RGDPR.

Figure 8.Comparative Percentage Growth Rates of Exports in Sub-Saharan Africa, Selected Countries

(Percent)

Note: The variable plotted is XGDPGR.

The second observation, which has been more forcefully made recently, is that the most significant factor behind the slow growth performance of sub-Saharan Africa is a lack of openness to trade, which constrains the region from exploiting opportunities to trade and stifles efficiency and technological progress associated with engaging in open trade (Sachs and Warner, 1997). This observation borders causality rather than the more modest association which other cross-country studies pose. In fact, Sachs and Warner conclude from their empirical analysis based on counterfactuals that, if African countries had pursued greater openness (say at East Asia’s level) between 1965 and 1995, they would on average have achieved an extra 1.4 percentage points in per capita income growth, which significantly exceeds the actual average of 0.9 percent for the same period. The fundamental policies singled out in this respect are keeping tariffs on imports at less than 40 percent, black market exchange rate premiums at less than 20 percent, average quota and licensing coverage on imports at less than 40 percent and ridding the economies on extreme controls on exports—via trading monopolies. The other study, which took a similar approach to explaining growth performance focusing on trade-influencing factors, is that by Rodrik (1997). Using an Africa-only sample, he finds that trade policies played a much smaller role in long-term growth performance but exert a much more significant influence in the medium term. Wide variations across countries are discerned in sources of growth decomposition for each country, underscoring the caution against generalizations. What is also different in this study is that the individual components of trade-affecting policies were decomposed, allowing distinct channels of influence on growth to be assessed.

In light of the above observations, this chapter will focus on four dimensions in explaining the relationship between trade and growth in sub-Saharan Africa. The first is to determine what explains trade performance in sub-Saharan African countries. We do this by identifying the key policy and nonpolicy factors that influence trade shares. This is supplemented by a review of recent trade liberalization experience in the region to seek an understanding of motivation behind it, the extent to which it has taken place, and in what forms.

Second is to separate between what we consider to be experiences of trade policy “proper” (for example, trade taxes) and endogenous trade policy measures (for example, quantitative restrictions and policies governing exchange) with the view to establishing the relative impacts of these on growth. The contention here is that endogenous trade policy dominates the out-turn of trade openness and therefore affects growth. This is particularly so when one considers the fact that actual trade taxes applied (revenue from taxes and the tax base) are much lower than statutory levels on account of exemptions offered, particularly to the dominant importer—the public sector and imports for projects funded by official development assistance and sweeteners provided to investors. The bulk of these exemptions are statutory, and therefore they can be anticipated.

The third area we focus on is the transmission channels from trade openness to growth, by distinguishing to the extent possible the separate effects of import liberalization, export promotion, and the liberalization of exchange regimes. The weights and intensity of application of these vary significantly across countries and over time to enable us to distinguish their impacts. In addition, we consider the technological enhancement impact of openness by incorporating the interaction between trade and FDI as a vehicle for technological diffusion.

The fourth area concerns bringing out the dynamics between trade and trade policies and growth in the medium term, using panel data analysis. We consider this to be the closest approximation of intertemporal country specificity in the context of cross-country studies, and it allows us to obtain some insights on medium-term growth effects to add to the bulk of the literature, which focuses on the long-term growth effects of trade openness. In carrying out this analysis, we exercise caution because trade openness and growth influence each other in both directions and hence require that we check for simultaneity in the relationship.

The second section will review the wisdom from the literature on the influence of openness on growth. In the third section, we review recent experience with trade liberalization in sub-Saharan Africa to establish whether or not it has occurred and in what form, and to highlight the potential channels of influence on growth. In doing so, we are mindful of the fact that a large number of studies on this issue focusing on sub-Saharan Africa have typically not covered the more recent period (post-1990), when more intensive trade liberalization efforts were made and therefore may have missed the fine points of the process of liberalization since the late 1980s. The fourth section presents the conceptual framework and empirical analysis of the relationship between trade and growth in sub-Saharan Africa. It reviews the empirical findings and draws conclusions from them. This includes pursuing (in the empirical analysis) the relative importance of openness to explain Africa’s growth, as well as revisiting what we conjectured in the previous sections to draw out policy wisdom in light of the findings. The fifth section briefly concludes.

Linking Trade and Growth: What the Literature Tells Us

Edwards (1993) provides a comprehensive review of the key issues and empirical evidence on the link between trade and growth in developing countries. What is particularly notable from this review is the continuing difficulties in obtaining reliable measures of trade policy and identifying more precisely the channels through which outward orientation facilitates growth. The difficulties also relate to assigning relative importance to trade policy reforms in comparison with consistent macroeconomic policy reforms and investment strategies in explaining growth performance. Furthermore, what complicates matters even more is the fact that some trade policy measures such as quantitative restrictions have been applied for both protective purposes (trade function) and for balance of payments purposes (protecting reserves).

Import substitution strategy was widely embraced by developing countries during the 1960s and 1970s following the influential thoughts of Prebisch and Singer in the 1950s, which propagated protection as a means to industrial development. Its critique rested on the now well accepted assertion that excessive protection of local industry stifled productivity growth, encouraged inefficient industrialization at the expense of agriculture and exports, reduced savings, increased unemployment, and led to very low rates of capacity utilization. The critique began with Little, Scitovsky, and Scott (1970) and was continued by Papageorgiou, Michaely, and Choksi (1991).

The same critics and other more recent studies (e.g., Sachs and Warner, 1997) have instead strongly advocated openness to international trade and finance complemented by consistent macroeconomic policies as the key to successful export-led industrialization and rapid growth. East Asia’s phenomenal success in this regard is the centerpiece of evidence based on country experiences, and more recently evidence from cross-country growth regressions identified trade openness to be a critical explanatory factor. We mentioned above that most recent of these cross-country studies as applied to explaining African growth performance to be those by Sachs and Warner (1997) and Collier and Gunning (1997) and Rodrik (1997). The first two studies place trade openness at the top of explaining the slow growth performance of the region. Rodrik finds that a small number of variables—country size, per capita income, geography, and taxation of trade—explains 82 percent of variations in trade shares. Furthermore, Sachs and Warner (1997) find openness to trade to be the single most important influence on long-term growth in Africa, while Rodrik (1997) finds openness to trade as measured by the Sachs and Warner index to be a strong determinant of African growth only in the medium term, using an Africaonly sample (p. 21). In Rodrik’s long-term growth equation (again for Africa only), the trade variable that enters with statistical significance is export taxation. Using a sources-of-growth decomposition for the 31 countries included in the regression, Rodrik finds that export taxation did play a more significant role in explaining the growth performance of worst performers.

It is instructive to note that earlier studies focused on searching for evidence by estimating impacts of liberalized trade regimes (with neutrality as the desired state) on export growth, and in turn linked export growth to overall growth of output and employment, thus claiming an indirect causation. This effort proved less successful because empirical evidence linking export and overall growth was rather weak. A few others (for example, Feder, 1983) added the impact of higher productivity of the export sector growth on overall growth via externalities and productivity differential effects with a little more empirical success. Yet others, Esfahami (1991) and Helleiner (1990), sought the link between export-growth and economywide growth through the role of export earnings in alleviating the foreign exchange constraint to imports in the context of import compression, which held back expansion in production and investment, also with some empirical success. The main argument here is that, given the imperfect substitution between imported and home goods, a cut in capital and intermediate goods imports will result in a reduction of real activity and growth (Khan and Knight, 1988; Ndulu, 1986, 1991). Since import capacity has been found to be a relatively more important determinant of imports under a foreign exchange constraint than relative prices (Moran, 1990), overall growth performance, and for that matter export growth, will be stimulated by relaxing this constraint.

We can distinguish three key channels of influence from trade to growth drawing from the foregoing. The first is that of efficiency-enhancing advantages of a more open trade regime. These relate, on the one hand, to enhanced competitive environment, and on the other to improved access to better technology. They operate through productivity growth resulting from more efficient allocation of resources, exploitation of comparative advantage and scale economies, and improved technological competence. Foreign competition undermines local monopolies and weeds out inefficient local producers. Transitional costs of de-industrialization may be high, but these are quickly compensated by the growth of new and restructured enterprises, and the expanded market opportunities that a liberalized trade regime offers. A more open trade regime also permits faster technological diffusion through improved access to a larger menu of innovation worldwide. The public nature of such innovation makes it possible to access new technology without typically incurring costs of developing it (Baldwin, 1996).

The second channel is linked to improved incentives for production of exportables and reduction of the high transactions costs associated with import barriers. Barriers to imports for protection of local industries are effectively a tax on exports because they raise the domestic resource costs for exports and lead to the appreciation of the real exchange rate, thereby reducing international competitiveness and incentives to export (De Rosa, 1990). Rodrik (1997) further confirms this assertion by empirical results, which show that barriers to imports adversely affect exports in Africa. We show below that import expansion has a residual positive impact on growth, though it is weak.

Improved availability and choice of consumer goods, particularly those considered as incentive goods, which is permitted by import liberalization, plays an important role in enhancing production of the self-employed, the majority of whom are engaged in agriculture. Long periods of frustrated consumer demand were associated with shortages occasioned by quantitative restrictions, which were strongly biased against consumer goods in many African countries. These restrictions were applied in a bid to cope with foreign exchange shortages and were implemented mainly through administrative allocation of the scarce foreign exchange. Import liberalization also reduces enterprise and trade transactions costs associated with measures to circumvent controls, thus incurring additional costs of underground trade, and costs associated with maintaining excessive inventories to avoid production and trade disruptions due to uncertainties linked to trade restrictions. The complexity of the tariff regime introduces opaqueness and uncertainty in business planning and has added to transactions costs associated with rents extorted by authorities through corruption practices.

The third channel relates to reducing explicit and implicit barriers to export or outward orientation of production. Explicit barriers here are export taxes, which reduce the profitability of exports and bias the production structure toward nontradables. We did refer above to import barriers being an implicit tax on exports. Other forms of implicit taxation are excessive margins exacted by marketing intermediaries and exchange controls, which for a protracted period forced African exporters to surrender their export proceeds at grossly overvalued exchange rates, thus reducing profit margins when measured in local currency terms. The parallel market exchange rate premium in a sense thus measures the extent of taxation of exports by way of transfer of incomes to the authorities (Pinto, 1989). Dismantling of marketing monopolies and the freeing up of exchange regimes are thus important avenues for raising the profitability of export activities and encouraging outward-orientation of production structures.

There is, however, one important qualification we wish to raise here regarding the effect of import liberalization on growth. A rise in imports is effectively a leakage of aggregate demand and thus (through the Keynesian multiplier) will lead to a short-run reduction in aggregate output. Depending on whether the positive supply-side effects of import liberalization discussed above exceed this negative effect, the net effect of a rise in import share in the short run may thus be ambiguous. This qualification may be particularly important in the case where annual panel data analysis of growth determinants is applied.

A Review of Trade Policy Experience

Preliberalization Trade Policy Stance

For a period of nearly three decades after independence, African countries mounted high trade barriers for three main purposes. The first was to provide protection to infant industries under the widely adopted ISI strategy. High and complex tariff walls, import controls, subsidies via controlled interest rates, and overvalued currencies were among the main instruments applied for the purpose. These were supplemented by trade confinement to public entities covering a wide range of import substitutes and “no objection” certificates for controlling importation of competing goods.

The second purpose was revenue generation. External trade taxes were and continue to be a significant source of government revenue.1 Import and export duties were raised from year to year to meet budgetary gaps in a scenario of rapidly rising public expenditures. This financing strategy was adopted notwithstanding the fact that there is little if any correlation between high tariff rates and actual revenue collection (Pritchett and Sethi, 1994). As a consequence of frequent annual changes of tariffs to meet revenue shortfalls, large variations in the levels and structure of trade taxes ensued. To a significant extent, this phenomenon was to account partly for frequent reversals of trade liberalization episodes in the future.

Third, import controls and exchange controls were also used for managing foreign exchange reserves, particularly during periods of balance of payments crises. With exchange rates kept virtually fixed before the period of reforms, this approach to balance of payments management was widely applied. Tightening of import restrictions and exchange controls during crises is widely documented by an AERC study involving 10 country cases (see Oyejide, Ndulu, and Gunning, 1996).

There are two important qualifications to this general scenario of high trade barriers. Large tax exemptions, particularly for public entities and donor-funded projects, significantly lowered effective tariffs. This was done partly to provide further support to import substitution industrialization dominated by public enterprises. As a consequence, a wide gap between scheduled and effective tariffs ensued. Although effective tariff rates in sub-Saharan African countries were higher than in other developing regions, they were not too different from those in Latin America, for example. Table 1 presents a comparison of effective tariff and non-tariff measures applied in a selected number of African countries with a selection of countries from Latin America and East Asia based on data presented in Rodrik (1997). The differences with Latin America particularly since mid-1980 is negligible. There are larger differences, however, in the coverage of nontariff measures of nearly 25 percent for all product categories. This difference is larger for manufactured products pointing to higher trade barriers linked to ISI. One important observation from the table is that of the overall decline in effective import tariff rates from its peak of 32.8 percent during 1980-83 to 22.5 percent associated with the early import liberalization measures in the sample of countries included. We will return to this phenomenon in greater detail later below.

Table 1.Tariff and Nontariff Barriers in Selected African Countries
Sub-Saharan Africa1980–83

13 countries
1984–87

13 countries
1988–90

10 countries
Primary products
Weighted average tariffs (percent)24.420.118.9
NTBs (coverage, in percent)48.447.4
Manufactured products
Weighted average tariffs (percent)32.823.522.5
NTBs (coverage, in percent)42.745.4
All product categories
Weighted average tariffs (percent)30.222.621.3
NTBs (coverage, in percent)45.546.1
Latin America4 countries11 countries8 countries
Primary products
Weighted average tariffs (percent)16.821.117.3
NTBs (coverage, in percent)42.848.6
Manufactured products
Weighted average tariffs (percent)23.625.122.7
NTBs (coverage, in percent)28.420.9
All product categories
Weighted average tariffs (percent)21.323.920.9
NTBs (coverage, in percent)32.930.3
East Asia5 countries7 countries7 countries
Primary products
Weighted average tariffs (percent)10.510.011.1
NTBs (coverage, in percent)31.118.8
Manufactured products
Weighted average tariffs (percent)21.618.118.0
NTBs (coverage, in percent)
All product categories
Weighted average tariffs (percent)18.215.815.7
NTBs (coverage, in percent)25.611.8
Source: Rodrik (1997), Tables 4 and 5.
Source: Rodrik (1997), Tables 4 and 5.

A second qualification relates to partial compensatory measures to correct for anti-export bias resulting from import barriers and exchange controls. These were offered in the forms of cash grants to exporters, export subsidies, foreign exchange retention schemes, and tax rebates or duty drawbacks. The corrective measures reflected the inconsistencies in the ISI strategy. On the one hand, the strategy vastly expanded the demand for intermediate and capital goods imports, while on the other it stifled growth in import capacity through barriers to export. This approach to resolving the inconsistency of the strategy was largely not successful, even though it partially reduced the anti-export bias.

Trade Liberalization Experience

Much of the discussion below draws from the AERC study on trade liberalization experience in sub-Saharan Africa. The study involved indepth review of the experiences in 10 countries.2 A synthesis of the experiences is contained in an overview paper by Oyejide, Ndulu, and Gunning (1996).

Although all of these 10 countries have had episodes of sporadic trade liberalization since the 1970s, the most consistent efforts occurred from the late 1980s. Previous attempts were linked either to commodity price booms, which relaxed foreign exchange constraints, or to attempts at compensating for anti-export barriers through subsidies or export promotional efforts. Most notable, however, was a drastic reduction of export taxes in a few countries during the early 1980s in response to the foreign exchange crunch that set in after the second oil crisis of 1979. These countries included Cote d’Ivoire, Kenya, and Tanzania.

The main stimuli for the more persistent liberalization episodes were twofold. The most important was conditionality imposed for gaining access to external finance under the structural adjustment programs. The policy required changes, and the associated benchmarks were set in the policy framework papers and sector lending programs. They invariably included import and export liberalization, related liberalization of domestic trade environment (particularly dismantling public trade monopolies and decontrol of prices) and liberalization of the payments systems. Import liberalization entailed reduction of tariff levels, compression of the tariff structure, and removal of import licensing. Export liberalization entailed further reduction of export taxes, removal of mandatory surrender requirements for export earnings, and simplification of export procedures. These measures were complemented by macroeconomic adjustment in the negotiated programs. Apart from freeing up the foreign exchange market, reduction of fiscal deficits and containment of monetary expansion were considered as consistent measures for sustaining liberalization. The typical sequence of action across the countries was tariffication of quantitative restrictions, removal of domestic trade barriers, and then reduction in tariff rates.

A second stimulus operating in different countries and at different points in time was “own initiative.” Most significant among the cases studied is that of South Africa. Various policy reviews conducted between 1983 and 1993 led to an adoption of a new industrial strategy under a more neutral trade regime with tariffication of QRs and a simpler, more transparent tariff structure. Subsequent binding of lower tariff levels under GATT and negotiated free trade agreements consolidated these changes. Mauritius also pursued own initiatives, successfully focusing mainly on compensatory mechanisms to ameliorate the anti-export bias arising from ISI. Most notable among these measures was the setting up of export processing zones free of trade barriers. In other countries, the own initiatives were largely in the form of Own Funded Import Schemes, which served as launching pads for a more comprehensive subsequent liberalization under the adjustment programs (Kauffman and O’Connell, 1991).

The efforts and consequences of these liberalization measures can be gleaned from Table 2, which is borrowed from Oyejide, Ndulu, and Gunning (1996), It presents the most recent review of status of trade liberalization in sub-Saharan Africa, in a manner which consolidates measures, outcomes and impacts of liberalization. It uses five indicators to measure progress with liberalization: tariffication of QRs, payment liberalization, tariff liberalization, reducing effective rates of protection and export taxes. The link with policy accounts helps to map out attribution of changes in the indicators to policy actions. The impacts are assessed using changes in trade shares, output, employment and industrial efficiency. The table also evaluates the sustainability of liberalization through the number of reversals and whether they were partial or total. This is complemented by a forward-looking assessment of confidence in sustained liberalization by means of responses from a survey of industrial enterprises.

Table 2.Trade Liberalization in Sub-Saharan Africa, Selected Countries
FindingKenyaUgandaMauritiusNigeriaGhanaSouth AfricaTanzaniaZambiaCôte d’IvoireZimbabwe
Trade liberalization:Yes, veryYes
Did trade liberalization take place?YesYesYeslimitedYesYesYesYesYes(limited)
Tariffication of QRs:100100909997
Exchange rate premium decline (percent)
Imports covered by licensing
Reduction (percent)1006410010010010022
Payment liberalization: (reduction of exchange controls)?YesYesYesYesYesYesYesYes
Tariff liberalization?YesYes
Reduction in tariffs (percent)76855–10 pts.38706736
Tariff structure compression (percent)768541705044
Effective rates of protection:
Reduction?YesNoYesNoYesYesYesYesYesYes
Percent605058
Export taxes:
Reduction?YesYesYesNoYesYesYesYes
Percent100100100
Macroeconomic adjustment:
Did macroeconomic adjustment accompany liberalization?
Exchange rate depreciation:
Depreciation (Nominal local)?YesYesYesYesYesYesYesYesYesYes
Foreign (percent)36443645801700455304800100165
Fiscal reform?YesYesYesNoNoYesYesNoNo
Reduction of deficit/GDP (percent)-28441
Impetus for liberalization:
What was the impetus for liberalization?
Positive commodity price shocks?YesNoYesNoNoNoYesNoNoNo
When?1976/771976/77
Commodity linked to reform?YesYesYesYesYesNoYesYesYesYes
Beginning when?198019871979198619831986198319841990
Membership in a regional scheme?YesYesNoNoNoNoYesYesNoYes
Own initiative to better performance?NoNoYesYesYesYesYesNoNoYes
Impact of liberalization:
What impact did liberalization have (+ or -) on:
Imports (imports/GDP)*3812335-671088min*Real –4*Real –36-29mix
Exports (exports/GDP)62946min723min26-31-14+
Output (real GDP growth)56.46min4min4.2-1.13.4
Employment+min33minmin18
Industrial efficiencyn.s.c.+
Sustainability of liberalization: Was liberalization sustained?
Number of episodes7324234234
Any reversals?YesYesNoYesYesNoYesYesYesNo
Were reversal partial (P) or total (T)?PTPPPPP
Did survey results indicate confidence/ support for liberalization?YesYesNoNoYesYesYesNoYes
Source: Oyejide, Ndulu, and Gunning (1996).Notes: Where information is not quantified, qualitative indicators are used, if available. This table summarizes the findings of the Regional Integration and Trade Liberalizaton Project. Abbreviations: min: negligible; mix: mixed; pts: percentage points on all imports except luxury goods; n.s.c: no significant correlation; *: nominal unless otherwise stated.
Source: Oyejide, Ndulu, and Gunning (1996).Notes: Where information is not quantified, qualitative indicators are used, if available. This table summarizes the findings of the Regional Integration and Trade Liberalizaton Project. Abbreviations: min: negligible; mix: mixed; pts: percentage points on all imports except luxury goods; n.s.c: no significant correlation; *: nominal unless otherwise stated.

The following broad observations can be made from Table 2. In the 10 countries, quantitative restrictions have been completely or mostly tariffied. Coverage of import licensing is virtually nil in five of the countries for which data were provided. There has been a switch from long positive lists of permitted imports to very short negative lists of prohibited or luxurious items. This has been strengthened by a virtual abolition of exchange controls and complete liberalization of the current account transactions. Four out of five countries for which information was available have signed Article VIII with the IMF to bind this status.

Tariff structures in all the 10 countries have been compressed from the previous high number of more than 60 (for example, Mauritius) to a range of 3–6 categories. Duty rates (scheduled) have been significantly lowered, again virtually across all 10 countries. The reductions are from the maximum, ranging between 150 and 250 percent, down mostly to 30 to 50 percent. A few items remain above these levels. Protection levels (measured by effective rates of protection) likewise have been significantly reduced in 7 out of 10 countries. The individual case studies also present evidence showing significant movements toward more neutral trade regimes. The measurement applied in this regard was the ratio of domestic to world terms of trade, with neutrality implying nonexistence of wedges related to trade policies. This is partly confirmed by the steep decline in black market foreign exchange premiums, which typically reflect the size of the wedges between domestic and border prices of imports and foreign currency.

A critical concern from the review of these experiences pertains to the sustainability of these achievements in light of the fact that they have not been bound. Although the majority of sub-Saharan African countries are signatories to the GATT, they have bound their tariffs at levels that are significantly above actual. The bound tariffs range from 15 to 280 percent against actual applied average rates ranging from 7 to 47 percent. Furthermore, the frequency of partial reversals noted in Table 2 shows that pressures for reversals abound. Oyejide, Ndulu, and Gunning (1996) make the following assessment of the sustainability of the trade liberalization in the 10 countries in relation to compatibility of macroeconomic policy stances: Exchange rate adjustment appears critical for ameliorating pressures of reversals. The large devaluations that accompanied trade liberalization partly reduced pressures from a potential excessive rise in import demand as liberalization proceeded. Furthermore, the higher revenues from trade taxes, associated with the large change in the valuation of the trade tax base in domestic currency terms, helped to stem potential fiscal problems resulting from the reduction in tariffs and hence pressure for reversal. Replacement of quantitative restrictions likewise helped to reduce potential fiscal pressures. Although in many countries trade reform has been accompanied by high real interest rates resulting from vastly increased government borrowing from the public, as yet there is no evidence that the resultant slowing down of investment, and hence resource reallocation expected from liberalization, has undermined credibility. It is nevertheless a source of concern if benefits from liberalization fail to materialize quickly. Support for it may wane. In 6 out of the 10 country cases, the support for liberalization was quite strong based on surveys of industrialists (see Table 2).

During the relatively short periods of trade liberalization in the case study countries, there has been generally a shift in the allocation of resources toward tradables and away from nontradables and import substitutes. Export orientation has increased, mainly due to a rise in the profitability of exports caused by real depreciation of local currencies and to a much smaller extent on account of trade policy changes. Although de-industrialization occurred as a result of opening up to competition, it appears that the growth of activity from surviving and new firms more than compensated for the collapsed ones.

For the African region as a whole, export growth is rebounding after nearly a decade of decline between mid-1970 and mid-1980. Export earnings grew at an average rate of 3.5 percent between 1986 and 1993 following various reform measures. During 1995 and 1996, a significant acceleration of export growth was achieved. Export earnings grew at an average rate of 8 percent during the 2 years, with half of this expansion attributed to an increased volume of exports (UNCTAD data). This may be the beginning of the more substantial response to the recent intensification of liberalization.

Relationship Between Trade and Growth

Framework of Analysis

This section presents some empirical assessments of the links between international trade and growth. The available literature relates trade regimes to export growth, and in turn export growth to income growth. This shows that there is a direct as well as an indirect link between trade regimes, export performance, and growth, but the direct link has been found to be weak. Part of the problems related to these results is the inherent simultaneity between export performance and growth. Furthermore, data are limited that would empirically help to account for the dynamism between trade performance, export composition, and export growth, on the one hand, and real income growth and the quality of this growth, on the other.

Whereas this section does not claim success in incorporating these shortcomings, it nevertheless starts by explaining the factors behind trade performance. This is accomplished by first estimating cross-country panel regressions for a group of countries that explains trade shares (export and import) in GDP. We estimate two equations for the trade shares, that is, export share and import share. In addition, we show estimates from a growth equation for the same group of countries. Furthermore, to take into account the simultaneity problem alluded to above, these three panel regressions are reestimated simultaneously using the full information maximum likelihood method and the results compared. We use these results to further show the impact effects in an attempt to assess the relative importance of the links between trade and growth.

The main question asked in the empirical analysis is what are the links between trade and growth for the group of countries in the sample for the period 1970–94. Can these links be empirically assessed? The transmission channels recognized in the literature include:

  • Openness of the economy. This enhances the efficiency and overall effectiveness of macroeconomic policies and investment.

  • Export-led growth, which tends to be more efficient beyond the openness argument. This is because export-based growth will bring in (1) technology transfer, (2) efficient allocation of resources due to international competition, and (3) cost-efficiency and related factors. These effects provide a further impetus to growth beyond what openness can provide, that is, there are dynamic multiplier effects.

  • Through the real exchange rate as a policy tool for ensuring competitiveness and minimal price distortions.

  • Alleviation of import constraints (foreign exchange supply constraints), which is necessary to ensure supply of intermediate and capital goods to domestic firms. Import liberalization thus contributes to overall growth through reducing import supply constraints and reduces barriers to entry.

We recognize two further channels that may affect growth:

  • Foreign exchange liberalization, where it has occurred, has helped to alleviate foreign exchange constraints. This builds up importing capacity and thus reduces import constraints.

  • Quality of growth from export orientation brings with it export externalities, which are partly related to technological advancement, competitiveness, and productivity differential aspects.

To take these factors into account, we introduce in the estimated equations trade-related policies that influence or hamper trade. These include:

  • Export taxes: these may be direct taxes levied on export or implicit in the form of real exchange rate misalignment.3

  • Tariffs and nontariff barriers: exporters have to buy their inputs at higher than world prices but have to sell their goods at world prices. Thus there is a wedge between world and domestic prices of imports, while exports have to be traded at world prices. This is captured by import price index adjusted for domestic duties and real exchange rate misalignment.

  • Macroeconomic stability: these are trade-related, in view of the fact that they affect investors (foreign and domestic) and thus technological development. Thus volatilities in macroeconomic variables and the outcome of the macroeconomic stance become crucial in directly and indirectly affecting export performance. For example, macroeconomic instability in general will be reflected by rising inflation domestically, which thus renders exports uncompetitive.

  • Uncertainty in the future flow of foreign exchange leads to stockpiling of imports to hedge against future shortfalls. This encompasses wastage in the use of imported raw materials, and there is an element of overpricing. Thus there is a premium due to scarcity of foreign exchange. This leads to the distortionary effects discussed above and thus misalignment of the real exchange rate.

  • Finally, institutional variables are likely to be important in facilitating export and import activities.

These channels are, however, difficult to map out in an empirical exercise. This is because the relationship between export performance, alleviation of import constraints, and overall economic growth may provide direct as well as feedback effects. This may perhaps be captured by simultaneously estimating the three equations and may thus capture the dynamics among them that could otherwise not be possible in a single equation estimation. To take this into account, we first estimate panel regressions for trade shares and growth and then reestimate them simultaneously. The interdependencies between them will thus show up in a reduced number of regressors and the efficiency of the parameters left in the model.

Empirical Results

The results of the panel regression equation that take into account some of these links are shown in Table 3. The table shows the variables used in the first column, the export share equation in the second column, the import share equation in the third column and the real income growth equation in the fourth column. The appendix defines each of these variables.

Table 3.Panel Regression Results (Random Effects Model)
VariableExport Share EquationImport Share EquationGrowth Equation
Constant2.769 (4.536)1.325 (1.56)
EXPGDPt–10.0519 (4.031)0.156 (2.75)
MGDPt–10.0416 (3.206)-0.039 (–0.689)
TOT-0.126 (–7.68)
TOTSHK0.0689 (5.728)0.0556 (4.642)0.387 (24.80)
XPI0.0425 (7.26)
MPITX-0.000013 (–0.126)
ΔRGDPGR0.229 (11.778)0.257 (13.323)
RGDPGRt–10.511 (27.242)0.517 (27.35)0.265 (18.71)
RER0.0572 (5.778)0.0354 (4.629)-0.047 (–4.57)
RERMIS-0.049 (–4.11)
RERMISt–1-0.016 (–1.96)
FDIOPEN0.0195 (2.247)
XPODUTY-0.0078 (–1.218)
ICRGE-0.0318 (–2.850)-0.030 (–3.068)
WARSSA-0.078 (–3.94)-0.0485 (–2.438)
ΔDY0.0431 (3.740)0.054 (5.338)-0.0027 (0.161)
GINV0.620 (32.19)
LSCHOOL0.017 (2.31)
INFVR-0.097 (–8.56)
INF0.171 (12.71)
DSX-0.066 (–4.27)
GDPRR-0.020 (–1.99)
R20.8200.8130.82
Note: The figures in parentheses are the t-values.
Note: The figures in parentheses are the t-values.

In the export share equation, EXPGDPt–1 is export share lagged one period. This variable captures the past performance of exports and hence constraints related to export bias. If export incentives are appropriate, then past export performance should be positively related to the current export level and should positively stimulate growth. This is because a successful export environment (culture) will simulate further export production, and this stimulates growth with important feedback effects. In the import share equation, the parallel to this is import to GDP lagged (MGDPt–1), which stimulates import demand in line with past levels but shows that it is negatively related to growth. There are two competing effects here. One is on the demand side, where imports are seen as a leakage, and thus should be negatively related to growth. The other is on the supply side, where imports constraints are eased with liberalization coupled with efficiency gains. The effect of the latter influence should stimulate growth, and thus be positive. The results show that the leakage effect is stronger, but it is, however, not highly significant in the growth equation.

The second variable in the export share equation is TOTSHK, terms of trade shocks. They show that positive shocks have positive effects on exports, imports, and growth, but terms of trade (TOT) has a negative effect on growth. The export price index (XPI) is positively related to export share. This is consistent with a theoretical prediction of a supply function. This is also supported by the results in the table. The next set of variables are lagged growth of real income (RGDRt–1) and speed of this growth (ΔRGDCR). These two variables are important in both import and export equation, and certainly lagged growth (or past growth performance) explains the observed current growth of incomes. This confirms the observed regularity in the literature that strong growth will stimulate the further growth in future and that a current recession may not stimulate future growth. We come back to this aspect when we look at the impact effects of these variables.

The next set of variables introduces trade-policy-related variables. These include real exchange rate (RER) and real exchange rate misalignment (RERMIS), as defined above. Real exchange rate depreciation stimulates export supply, while an appreciation stimulates import demand. The results also show that real exchange rate depreciation stimulates growth. In addition, there are tax variables that enter into the trade share equations. The first is computed from taxes on imports, (MPITX), which enters into the import demand equation with negative effect (though not statistically significant). The second is a direct tax on exports (XPODUTY)/which deters exports, and the results corroborate with the expectations (even though they are not highly significant).

The next set of variables looks at the quality of institutions (ICRGE) and sudden disruptions due to civil unrest (WARSSA), which is an indicator of the number of civil wars crossed with a dummy for sub-Saharan Africa. These two variables are consistently negative for export supply and import demand equations but have no effect on the growth equation. This underscores the importance of institutions and political tensions in supporting or hampering trade.

The final variable that relates to the three panel regressions is the flow of external debt to GDP (ΔDY). This is the first difference of the stock of the ratio of external debt to GDP. This should be positive because it reflects some aspect of financing (that is, official aid flows) and can thus be seen to alleviate foreign exchange constraints. The results support this view, but it is not significant in the growth equation. Even though external debt enters the model, and in first difference is not statistically significant, we have further included the crowding out effect of external debt that would work through investments to reduce growth. This is the debt service ratio (DSX), which is negative and highly significant in the growth equation. In addition, the export equation includes a proxy for the technological enhancement impact of openness, as argued in the introduction of this chapter. This is FDIOPEN, which is the ratio of foreign direct investment to GDP crossed with openness. This variable is positively related to export supply, which confirms the arguments put forward above. It is, however, not significant in the other equations.

We recognize the fact that growth is affected by other factors besides trade or openness to trade. For this reason, we have included a number of additional variables. They include gross investment (GINV), which is a sum of public and private investment rates and which is positively related to growth; human capital, which is proxied by LSCHOOL, the number of initial schooling years. This is in line with the recognition of human capital in endogenous growth models, and the regression results support this view.

The outcome of macroeconomic policies effect growth. To take this into account, we include inflation and its variability in the growth equation (INF and INFVR). Unstable macroeconomic policies are refleeted by rising and volatile inflation, which retards growth, while moderate inflation would stimulate growth through revenues from a moderate inflation tax. It appears that the results confirm these links.

Finally, the neighborhood contagion effect is proxied by regional GDP (GDPRR), which is negatively related to growth and is significant. Overall, the panel regression results conform to the predictions. But a caution, as argued in the introductory section of the chapter, is that simultaneity may be a problem. To remedy this, we need to reestimate the panel regressions simultaneously. We do this in full-information, maximum-likelihood estimation (FIML), and the results are shown in Table 4.

Table 4.Regression Results: Trade Shares and Growth
VariableExport EquationImport EquationGrowth Equation
Constant2.38 (1.16)2.47 (1.26)2.26 (2.12)
EXPGDPt–10.0959 (9.79)
TOT-0.120 (–12.38)
TOTSHK0.046 (1.68)0.0447 (1.55)0.379 (36.59)
ΔRGDPGR0.268 (6.98)0.264 (5.97)
RGDPGRt–10.495 (15.79)0.500 (13.938)0.314 (32.08)
RER0.029 (1.33)0.028 (1.14)-0.057 (–4.35)
WARSSA-0.071 (–2.302)-0.0053 (–2.47)
ΔDY0.0667 (2.109)0.067 (2.08)0.030 (2.36)
GINV0.470 (48.44)
LSCHOOL0.021 (2.88)
INFVR-0.117 (–11.18)
INF0.166 (14.87)
DSX-0.067 (–5.60)
R20.7950.7940.787
Note: The results are a full-information, maximum-likelihood estimate. The figures in parentheses are t-values.
Note: The results are a full-information, maximum-likelihood estimate. The figures in parentheses are t-values.

The variables in the FIML estimation were retained in the model as long as they were significant so that the results that emerged were achieved through a process of reduction.

The first thing to note in Table 4 is that the number of regressors is significantly reduced. This is consistent with the time series models where one starts with a large, overparameterized statistical model, and the reduction process ensures efficiency until a parsimonious model is achieved. The second thing to note from this table is that even though the number of regressors have been drastically reduced, the R2 for these three equations has not significantly declined.

The results show that export success or success of export policy, reflected by lagged export share, is positively related to real income growth. Growth performance, both in terms of lagged performance and its speed, are an important drive to both export and import share equations as well as to the growth equation. Thus, a booming economy also promises growth in the future, so the observed growth level is also a function of past growth performance. The results in this table indicate that 31.4 percent of past growth performance is carried into the current growth record. This is an important stimulus to both trade shares, as well as growth with positive externality effects on all the three variables.

Notably, import share lagged neither affects growth nor its own equation. This is consistent with earlier argument on the net affect from both supply and demand sides. In this case, the import liberalization effect is picked up by growth feedbacks to the import equation.

The next set of variables is the real exchange rate and a dummy for civil war, which have maintained their significance and direction, except that now civil wars do not affect growth directly, but via disruptions of export supply and import demand. Changes in the ratio of external debt to GDP is consistently positive, just like before and is now significant in the growth equation.

From the FIML results, we note that trade policies affect imports and exports indirectly. This is because no trade policy variable effects are significant in the trade share equations. The results show that trade policy measures are indirect, and we can make three observations in relation to these results. First, trade matters to growth, but macrovariables like the real exchange rate have a strong influence on growth and then work indirectly to affect exports and imports. Second, gross investments, GINV, affects growth directly. But investments are usually affected by trade policies. Trade reforms affect investment by improving their profitability. Thus the effect of trade policy on growth is indirect and strong through the channel of investment. Finally, it should be noted that trade liberalization effects occur through capital formation and thus affect growth. These effects do not enter the production functions directly. This perhaps helps to explain why trade shares were not significantly affected by trade policy variables.

We next assess the relative importance of the explanatory variables in explaining sub-Saharan Africa’s trade and growth performance by way of contrast with the sample as a whole and with East Asia in particular. The main question asked in these computations is what explains sub-Saharan African countries’ trade performance relative to the overall sample of developing countries and the best performers in East Asia. To accomplish this, the sample means of the explanatory variables were computed and used together with the means for the sub-Saharan African and East Asian regions. We then use the regression coefficients of the respective variables from the FIML results to obtain the contribution of each variable in explaining the differential performance effects. In the computations, we subtract the sub-Saharan African mean of the variable from the sample mean (and East Asian means) and multiply by the coefficient from the regression results. The last two columns show the differential contribution effects from the sample and East Asia compared with sub-Saharan Africa, and the last row gives the overall contribution of the included variables to explaining the estimated difference.

These results have thus utilized the interdependence between the equations and the feedback effects. The results are shown in Table 5 for the export share equation, Table 6 for the import share equation, and Table 7 for the growth equation.

Table 5.Comparative Assessment: Export Share Equation
MeanSample Versus Sub-Saharan AfricaEast Asia Versus Sub-Saharan Africa
VariableSampleSub-Saharan AfricaEast AsiaCoefficient
TOTSHK0.00590.00490.03380.0460.00460.133
ΔRGDPGR-0.0012-0.0016-0.001050.268-0.0107-0.0147
RGDPGR0.03910.03430.07290.4950.2381.911
RER0.97550.95490.98060.0290.05970.0745
WARCIV0.18160.13330.350-0.071-0.3429-1.5385
DDY0.03570.04980.00670.0667-0.0941-0.287
Overall differential-0.14540.2783

These tables show those variables that simultaneously determine trade shares and growth and also the lagged influence of export share on growth. The contributions are arrived at by multiplying the coefficient from the regression results in Table 4 and the difference between the mean of the variable in question, for sub-Saharan Africa and the overall sample of the 99 countries and the mean for East Asia. The results in these two tables are summarized by the respective last rows in Tables 5, 6, and 7.

Table 6.Comparative Assessment: Import Share Equation
MeanSample Versus Sub-Saharan AfricaEast Asia Versus Sub-Saharan Africa
VariableSampleSub-Saharan AfricaEast AsiaCoefficient
TOTSHK0.00590.00490.03380.04470.00470.129
ΔRGDPGR-0.0012-0.0016-0.001050.264-0.0106-0.0145
RGDPGR0.03910.03430.07290.5000.2401.93
RER0.97550.95490.98060.0280.05770.0270
WARCIV0.18160.13330.350-0.0053-0.0256-0.1149
DDY0.03570.04980.00670.067-0.0945-0.289
Overall differential0.1721.668
Table 7.Differential Effect Assessment: Growth Equation
MeanSample Versus Sub-Saharan AfricaEast Asia Versus Sub-Saharan Africa
VariableSampleSub-Saharan AfricaEast AsiaCoefficient
EXPGDPt–10.2270.2030.4460.09590.2302.100
TOT0.7780.7750.757-0.120-0.0360.252
TOTSHK0.00590.00490.03380.3790.03791.057
RGDPGRt–10.03910.03430.07290.3140.1511.061
RER0.97550.95490.9806-0.057-0.117-0.0291
DDY0.03570.04980.00670.030-0.0423-0.087
GINV0.2080.19980.2790.4700.38543.337
LSCHOOL1.2970.92971.6780.02120.7790.808
INFVR0.5580.6960.540-0.1171.6150.211
INF0.4940.3090.08380.1663.071-3.738
DSX0.1550.1530.181-0.067-0.0134-0.174
Overall differential6.0614.885

A positive overall effect shows that sub-Saharan Africa performs worse than the sample average or the average for the East Asian countries. In all cases, when we use the trade shares equations, sub-Saharan Africa performs worse than East Asia and performs slightly better than the average countries in the sample. From these results, it appears that the driving force is real growth, which affects trade shares equally importantly.

Table 7 shows the comparative assessment of growth performance for the two regions with sub-Saharan Africa with similar but this time stronger conclusions. Trade and trade policies affect growth, and growth in turn affects trade performance. This conclusion seems to support the observations made about the marginalization of sub-Saharan Africa in world trade and its poor output growth performance.

Although export shares affect growth, as we saw above, the feedback effects from growth to export are much stronger. The results, which seem to utilize the interdependences of the trade shares and real income growth, seem to point to the fact that output growth performance is key to the successful link between export and growth. The direct effects from trade and trade policies are not as strong as the growth impetus. This result is supported by the overall differential effect shown in Table 7. Sub-Saharan Africa is thus disadvantaged in the growth process, investment, and human capital, and this constrains export performance. This is shown in the differential effect to growth of 6.1 percent favoring the average countries in the sample.

We seem to arrive almost at a similar conclusion as Rodrik (1997), who argues that stimulating economic growth is the key to affecting other aspects that contribute to growth. This is in regard to the important contribution made to trade by successful stimulus from growth. But we qualify our results by utilizing the links we have outlined above and seek evidence from the results we have obtained in these two tables. In addition, we use the growth equation to assess the relative importance of factors contributing to growth and thus assess the role of trade in growth.

The next stage of analysis looks at the export growth equation. We include most of the variables as before, but in addition we now include lagged export growth and import growth in the equation, and the growth rates of the variables instead of the levels of the variables. In addition, some variables enter in both contemporaneous and lagged values. The results of the export growth equation are shown in Table 8 and Table 9 for the FIML estimation.

Table 8.Export Growth Panel Equation
VariableCoefficientt-ratio
EXPOGRt–10.85952.13
MGDPGR0.42932.61
ΔRGDPGR0.16111.37
ΔDY0.26020.64
TOTSHK-0.116-12.14
XPI0.01892.67
XPIt–10.06218.914
INF0.02232.821
ΔRER-0.0422-5.955
REPMISt–1-0.0135-1.834
DSX-0.157-13.273
LSGHOOL0.01262.434
FDIOPEN0.05639.650
R20.910.91
Note: Random effects model.
Note: Random effects model.
Table 9.Regression Results: Trade and Growth
VariableExport Growth EquationGrowth Equation
Constant-5.184 (–5.349)0.714 (0.983)
EXPOGRt–10.862 (91.15)0.0355 (3.835)
ΔRGDPGR0.161 (19.59)0.609 (73.02)
XPI0.021 (3.89)
XPlt–10.054 (11.14)
ΔDY0.254 (27.11)
TOT-0.157 (–20.67)
TOTSHK-0.117 (–16.33)0.345 (49.56)
ΔRER-0.0429 (–6.320)-0.0072 (–0.774)
RERMIS-0.0061 (–0.963)
RERMISt–1-0.047 (–6.031)
MGDPGR0.427 (64.40)-0.117 (–14.00)
INF0.0185 (2.74)0.189 (27.93)
INFVAR-0.132 (–16.063)
DSX-0.156 (–18.640)-0.091 (–11.88)
LSGHOOL0.0111 (1.84)
GINV0.234 (20.96)
R20.910.88
Note: Likelihood function = –18,865.1. NOBS = 2,475. The results are a full-information, maximum-likelihood estimation. The figures in parentheses are t-values.
Note: Likelihood function = –18,865.1. NOBS = 2,475. The results are a full-information, maximum-likelihood estimation. The figures in parentheses are t-values.

The real exchange rate enters in its first difference, indicating that it is the real exchange rate movements that are important to export growth. The results show that, to the extent that the real exchange rate changes are favorable to profitability and external competitiveness, they stimulate growth of exports. Real exchange rate misalignment enters with a first lag and is consistent with prediction: it penalizes export performance. In this equation, human capital enters into the growth equation even though it was not significant in the export share equation. The technological-enhancing variable, FDIOPEN, also turns out to be significant in the export growth equation. This variable plays the same role as argued previously. Two variables that have now become significant in the export growth equation are the debt service ratio (DSX) and inflation (INF). DSX was significant to the growth equation but not in the export share equation.

To the extent that significant export receipts flow out to service external debt, they are likely to check export growth. The results in Table 8 confirm this. For inflation, it shows that it stimulates export production and also growth. This is consistent with the arguments that moderate inflation is likely to stimulate export production through reduced pressure on appreciation of the exchange rate and growth. We added inflation variability UNFVR), but this was only significant in the growth equation and not in the export growth equation. The results in Table 8 are reported only for the significant variables. This equation was again reestimated simultaneously with the growth equation. The results are reported in Table 9.

Using the same variables for the export growth equation, we can also assess the feedback effects from growth to export growth. The results in Table 9 show that the speed of growth of real income, lagged export growth, and moderate inflation most significantly explain sub-Saharan African performance relative to the overall sample. Misalignment in the real exchange rate deters both export growth and real income growth. In addition, terms of trade shocks and debt service and import growth make negative contributions to real income growth.

The results of the computed beta coefficients showing the relative importance of the variables in the export growth and output growth equation are shown in Table 10. We have also shown the total contribution of the variables in the model by showing their explained proportion and hence indicating the unexplained proportion. The results give a lot of weight to the export environment, captured by lagged export growth, an inertial effect, which feeds into output growth with strong effects. This underscores the importance of incorporating the simultaneity arguments considered above. The other factors behind export growth are external debt flows, import growth, and export prices. On the growth side, lagged effects of growth have the most influence, followed by investment and export growth, while debt service obligations are a strong deterrent to growth.

Table 10.Trade and Growth: Relative Contributions (Estimated Beta Coefficients)
VariableExport GrowthGrowth
EXPOGRt–126.7514.44
ΔRGDPGR4.9375.77
XPI6.51
XPIt–16.82
ΔDY23.65
TOT-2.93
TOTSHK-6.980.345
ΔRER-1.53-1.04
RERMIS-0.028
RERMISt–1-0.874
MGDPGR35.9-39.99
INF0.239.618
INFVAR-9.054
DSX-9.01-21.385
LSGHOOL2.186
GINV32.323
Explained85.56978.6
Unexplained14.43121.39
Total100100
Note: The beta coefficient helps to assess the relative importance of the variables and factors in the regressions in explaining export growth and trade. The beta coefficients are estimated by using the size of the respective regression coefficient (x) multiplied by the ratio of the standard deviation of x divided by the standard deviation of the dependent variable. These computations are reproduced in this table.
Note: The beta coefficient helps to assess the relative importance of the variables and factors in the regressions in explaining export growth and trade. The beta coefficients are estimated by using the size of the respective regression coefficient (x) multiplied by the ratio of the standard deviation of x divided by the standard deviation of the dependent variable. These computations are reproduced in this table.

These results further reinforce the arguments in this chapter that trade performance will be reinforced by output growth as well as growth being sustained by trade performance. Thus export performance contributes to output growth, but more important, output growth explains more of the trade performance in general and has entered significantly in both trade shares and export growth equations.

However, even though a strong conclusion emerges that export growth contributes to overall output growth with strong feedback effects, there are still weaknesses into how much the chapter can contribute toward understanding the transmission mechanisms or channels from export growth to output growth and vice versa. Three issues may be pointed out. First, composition of exports and output may determine the strength of the transmission mechanism and the strength of feedback effects and furthermore the quality of growth. Second, primary exports will have a different impact on output growth from that of manufactured exports.4 Finally, we cannot answer the question to what extent trade reforms have affected the composition of exports and thus the transmission channels. Thus, even though we come to a strong conclusion about the relationship between trade and growth, we still need disaggregated export data to map out policy pointers on trade and growth and the strength of the transmission channels. Primary exports will have a differential impact on real income growth and a different transmission mechanism.

Concluding Remarks

There is now increasing evidence that those developing countries that have been most successful in achieving and sustaining high growth rates are also the ones that have taken most advantage of trade opportunities. Openness to trade and international finance has facilitated the effectiveness of investment and macroeconomic conditions that are conducive to growth. In this respect, the recent trend toward opening up the sub-Saharan African economies and reducing barriers to trade and finance augurs well for higher, sustained growth in the future. If we go by the recent encouraging upturn (1995 and 1996) in growth and export performance in the region, there is a good reason to believe that this may be a beginning of a virtuous circle from policy reforms to growth. This particular view is supported by the strong positive inertial influence of past growth and export performance for future improved performance. What is perhaps also a significant finding is the confirmation of the strong influence of overall growth on trade performance. In this regard, Africa’s marginalization in world trade will have to find an answer from creating conditions for improved overall growth, which is the base for participation in world trade. That is, poor performance in the past has been due to two factors: Africa did not trade enough, and its trade policy stance is to blame for these outcomes.

The influence of trade-enhancing policies on exports and overall growth has been shown to matter. Explicit trade policies, including trade taxes and nontariff barriers, play an important but less prominent role in this regard than macroeconomic policies that influence trade performance. Most significant among the latter are exchange rate policies, reduction in the exchange controls, price stability, and investment response. These of course operate more effectively in an open trade environment, which permits enterprises to make best and uncumbered use of global opportunities for trade and investment.

A major hindrance to strengthening the virtuous circle as confirmed by the results is that of the debt service burden, which crowds out domestic expenditures needed for supporting expansion of productive capacity and infrastructure for growth. An expeditious solution to this problem arising from the past is key to stemming coordination failures that hamper response of investment and indeed strengthening of the export base. Another structural bottleneck is the underdeveloped state of human capacity, which stifles absorption of technological competence availed through interaction of openness and foreign direct investment. Investment in human capital, particularly education, is thus an important complementary measure for realizing the benefits from opening up African economies.

Finally, the sustainability of the various liberalization measures that have been put in place is fundamental for achieving sustained improved performance in trade and growth. Pressures from fiscal gaps and from interest groups concerned with maintaining the protection of ventures put in place during the ISI era are most worrisome in this regard. Binding achievements toward a more open economy are critical for sustenance. Whether this is done through regional arrangements, WTO provisions, or local constituencies (for example, export lobbies), it is necessary to embark on them before to pressures for reversals mount to critical levels.

Appendix: Variable DefinitionsEXPGD

Export share in GDP

MGDP

Import share in GDP

TOT

Terms of trade

TOTSHK

Terms of trade shock

TOTSHK for year is t is calculated as: [(Pxt /Pxbase – 1)* (X/GDP)t–1] – [Pmt/PM – 1)* (M/GDP)t–1], where PX and PM are export and import prices indices respectively deflated by the U.S. GNP deflator; PXbase and PMbase are the average PX and PM respectively for the preceding 3 years; X and M are respectively the exports and imports of goods and nonfactor services.

XPI

Export price index

MPITX

MPI* [1 + IMPODUTY + RERMIS], where MPI is the import price index and IMPDUTY are import duties.

RERMIS

Real exchange rate misalignment, defined as (RERERER)/ERER

FDIGDP

Ratio of foreign direct investment to GDP

OPEN

Openness: imports + exports as a ratio of GDP

XPODUTY

Export duties

ICRGE

Measure of quality of institutions

WARSSA

Number of civil wars crossed with the sub-Saharan African dummy

DY

Ratio of external debt to GDP

LSCHOOL

Number of years in school

GINV

Gross investment (sum of ratio of public and private investment to GDP)

INF

Rate of inflation

DSX

Ratio of debt service to exports

GDPRR

Average regional GDP

EXPOGR

Growth of exports

MGDPGR

Growth of imports

RGDPGR

Real income growth

RER

Real exchange rate

INFVAR

Inflation variability

References

    BalassaB.1978“Exports and Economic Growth: Further Evidence,”Journal of Development EconomicsVol. 5 pp. 18189.

    BaldwinR.E.1996“Review of Theoretical Developments in Regional Integration,” in Regional Integration and Trade Liberalization in Sub-Saharan AfricaVol. 1Framework Issues and Methodological Perspectivesed. byA.OyejideI.Elbadawi and P.Collier (New York: St. Martin’s Press).

    CollierPaul1995“The Marginalization of Africa,”International Labour ReviewVol. 134No. 4–5 pp. 54157.

    CollierPaul and Gunning1997Explaining African Economic Performance (Oxford: Centre for the Study of African Economies).

    De RosaD. A.1990“Protection and Export Performance in Sub-Saharan Africa,”Weltwirtsafliches ArchivVol. 128No. 1 pp. 88124.

    EdwardsS.1993“Openness, Trade Liberalization and Growth in Developing Countries,”Journal of Economic LiteratureVol. 21 pp. 135893.

    EsfahamiH.S.1991“Exports, Imports and Economic Growth in Semi-industrialized Countries,”Journal of Development EconomicsVol. 35No. 1 pp. 93116.

    FederGershon1983“On Exports and Economic Growth,”Journal of Development EconomicsVol. 12 pp. 5973.

    HelleinerG.K.1986“Outward Orientation, Import Instability and African Economic Growth: An Empirical Investigation,” in Theory and Reality in Development: Essays in Honour of Paul Streetened. byS.Lall and F.Stewart (New York: St. Martin’s Press).

    HelleinerG.K.1990“Trade Strategy in the Medium-Term Adjustment,”World DevelopmentVol. 18No. 6 pp. 87997.

    KauffmanD. and S.O’Connell1991“The Macroeconomics of the Unofficial Foreign Exchange Market in Tanzania,” in Economic Reforms in Sub-Saharan Africaed. byA.Chhibber and S.Fischer (Washington: World Bank).

    KhanM. and M.D.Knight1988“Import Compression and Export Performance in Developing Countries,”Review of Economics and StatisticsVol. 70No. 2.

    LittleI.M.D.T.Scitovsky and M.Scott1970Industry and Trade in Some Developing Countries (New York: Oxford University Press).

    MoranC.1990“Imports under a Foreign Exchange Constraint,”World Bank Economic ReviewVol. 3No. 2 pp. 27995.

    NduluB.1986“Investment, Output Growth and Capacity Utilization in an African Economy: The Case of the Manufacturing Sector in Tanzania,”East African Economic ReviewVol. 2No. 1.

    NduluB.1991“Growth and Adjustment in Sub-Saharan Africa,” in Economic Reform in Sub-Saharan Africaed. byA.Chhibber and and S.Fischer (Washington: World Bank).

    OyejideT.A.B.J.Ndulu and J.W.Gunningeds. 1996“Country Case Studies of Trade Liberalization” (unpublished; Nairobi: African Economic Research Consortium).

    PapageorgiouD.M.Michaely and A.Choksi1991Liberalising Trade 7 vols. (Oxford: Blackwell).

    PintoB.1989“Black Market Premia, Exchange Rate Unification, and Inflation in Sub-Saharan Africa,”World Bank Economic ReviewVol. 3No. 3 pp. 32138.

    PritchettL. and G.Sethi1994“Tariff Rates, Tariff Revenue and Trade Reform: Some New Facts,”World Bank Economic ReviewVol. 8No. 1 pp. 116.

    RodrikD.1992“Closing the Productivity Gap: Does Trade Liberalization Really Help?” in Trade Policy Industrialization and Development: New Perspectivesed. byG.K.Helleiner (Oxford: Clarendon Press).

    RodrikD.1997“Trade Policy and Economic Performance in Sub-Saharan Africa” (unpublished; Cambridge, Massachusetts: Kennedy School of Government, Harvard University).

    SachsJeffrey and AndrewWarner1995“Economic Reform and the Process of Global Integration,”Brookings Papers on Economic Activity: 1 pp. 1118.

    SachsJeffrey and AndrewWarner1997“Sources of Slow Growth in African Economies” (unpublished; Cambridge, Massachusetts: Harvard Institute for International DevelopmentMarch).

    TanziV.1995“Taxation in an Integrating World,”Integrating National Economies Series (Washington: Brookings Institution).

    YeatsAlexander J.1997 with AzitaAmjadiUlrichEincke and FrancisNgDid Domestic Policies Marginalize Africa in World Trade? (Washington: World Bank).

The authors acknowledge research assistance from Radha Ruparel. They also thank Peter Montiel, Lawrence Harris, and participants in the seminar on which this book is based for useful insights and comments on the earlier draft of this chapter. They, however, remain responsible for any errors and omissions.

Recent research shows that the share of trade taxes in total revenue for African countries ranged from 4 to 52 percent in 1992.

These include Cote d’Ivoire, Ghana, Kenya, Mauritius, Nigeria, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe.

The preferred measure here (if available) is the parallel market exchange rate premiums. This is because exporters download their foreign exchange earnings using the official rate but have to purchase their inputs in the market that includes the parallel market. However, in countries where liberalization has taken place, real exchange rate misalignment is adequate in caphiring temporal deviations from the perceived real exchange rate equilibrium. However, data availability may constrain the choice to using a simple proxy of real exchange rate misalignment as a deviation from the average real exchange rate. This is what is used in the chapter.

It is quite difficult to assess this proposition due to data requirements. This could be feasible for individual countries rather that the 99 countries used in this study.

    Other Resources Citing This Publication