With Africa's growth finally picking up, the challenge is to accelerate and sustain the pace to reduce poverty
After decades of economic stagnation, growth in sub-Saharan Africa (SSA) has picked up considerably, reaching an eight-year high of 5.6 percent in 2004. Only a slight dip in this rate was seen in 2005, with one-third of the oil-importing countries achieving rates above 5 percent. But most SSA countries will still not be able to achieve the Millennium Development Goal (MDG) of halving the 1990 level of poverty by 2015 without doubling their rate of growth—and about half the region's population now lives below the poverty line. Are the improvements in growth sustainable? And what types of growth strategies will contribute most effectively to reducing poverty? This article explores these questions, drawing on new analytical methods to assess Africa's growth performance over the past 40 years.
A snapshot of growth
Between 1960 and 2003, real GDP per capita in SSA grew an average of 1.1 percent a year. This means that real per capita income was approximately the same in 2003 as in the mid-1970s, as the region steadily lost ground relative to both industrial and developing countries. But, since the mid-1990s, growth has accelerated. Average real GDP per capita growth increased to 2 percent in 1995–99 from –1.1 percent in 1990–94, with many countries sharing in the improvement (see Chart 1). However, during 2000–03, growth slackened for all of the subgroups in Chart 1 except the oil producers and resource-intensive countries.
What spurred growth after 1995? First, in the fast growers, macroeconomic performance was stronger and average inflation was almost half that of the slow growers. Their fiscal deficits were also lower because of higher revenue collections, and they were more open to trade, as shown by higher ratios of exports plus imports to GDP (see Chart 2). And, in a major departure from the past, total factor productivity (TFP) increased markedly—most likely reflecting efficiency gains stemming from countries' implementation of macroeconomic and structural reforms. But total and private investment barely rose, except in the oil-producing countries.
Should countries continue to lower budget deficits? Probably not. Studies have shown that reducing the deficit below 2–2.5 percent of GDP in countries that have achieved macroeconomic stabilization is not beneficial for growth. Our study confirms that countries whose deficits exceeded 2.5 percent of GDP were able to boost their growth rate by strengthening their fiscal position, whereas in countries with deficits lower than 2.5 percent of GDP, growth declined.
To shed more light on Africa's growth performance, it is useful to identify fast- and slow-growing countries. One way is through growth benchmarking, which ranks countries by their actual growth relative to their potential. A country's growth potential is determined by circumstances over which it may have little control, although its subsequent policy choices will affect the growth outcome. Benchmarking analysis quantifies this relationship by comparing actual and "expected" growth; the latter is derived from an estimated global relationship between growth and a set of factors beyond the country's control. This set includes variables that capture country location, exposure to trade, growth in trade partners and terms of trade, level of income in 1960, and historic mortality rates for European settlements. This last variable has assumed considerable importance in studies of growth in low-income countries because it is meant to capture the quality of institutions that countries inherited from colonial powers and may also capture current health conditions.
This exercise reveals that a number of SSA countries have fared relatively well since the mid-1990s despite unfavorable circumstances (such as being landlocked or resource-poor). Among those that performed well against their benchmark, Mozambique and Uganda's progression from conflict recovery to sustained growth provides lessons for the nascent recoveries in the Democratic Republic of the Congo, Liberia, and Sierra Leone. The best performers of the 1990s, in addition to the better-known cases of Botswana and Mauritius, are those that have grown slowly but steadily, like Benin, Burkina Faso, and Ghana. Other countries—such as Cameroon, Ethiopia, and South Africa—underperformed relative to the benchmark.
If SSA is to have a realistic prospect of halving income poverty by 2015, its real per capita GDP growth rates will have to increase to about 5 percent. Although knowledge about what leads to high, sustainable growth in SSA is limited, it is instructive to analyze recent success stories.
In a 2004 paper, Harvard University's Ricardo Hausmann, Lant Pritchett, and Dani Rodrik proposed that the traditional long-term focus of empirical growth research (based on regressions) could camouflage important patterns in a country's growth. By examining jumps in countries' medium-term growth trends, they argued, one could gain insight into the sources of growth spurts, or accelerations. For developing countries as a group, they compared per capita GDP growth rates seven years before and seven years after a given base year. They found that growth accelerations were largely unpredictable and that major changes in policies or external conditions led to sustained growth in surprisingly few instances.
Building on their paper, we sought to identify the determinants of growth spurts that took place in SSA during the 1980s and 1990s. For our purposes, a spurt occurred in a year when the per capita growth rate over five years was at least 2 percent higher than the growth rate in the previous five years and when the growth rate during those five years was at least 2 percent. In addition, per capita income had to be higher at the end of the acceleration than before it began: thus, a recovery from a crisis or a war did not count unless income eventually exceeded the precrisis level. We found that, since 1980, SSA has experienced 34 growth accelerations, with more occurring in the 1990s than in the 1980s, including several now under way (see table). This shows that, despite their relatively low average growth, SSA countries are capable of short- to medium-term spurts of growth.
Four factors appeared to accompany those growth spurts (see Chart 3). First, policies were not lax—that is, inflation and budget deficits did not rise. In fact, the World Bank's Country Policy and Institutional Assessment, a broad measure of the quality of the policy stance, remained steady or improved during an acceleration. Second, trade was important for promoting growth accelerations, which were associated with real exchange rate depreciations, export growth, and trade liberalization, although the importance of the variables in the two decades varied. Third, the quality of political institutions—such as a move toward democracy—was linked to accelerations. Finally, TFP and investment were key to growth spurts. One notable difference between the accelerations of the two decades is that debts increased in the 1980s and fell in the 1990s. The decline could be explained, in part, by the launching of initiatives in the mid-1990s to reduce the debts of the most heavily indebted poor countries.
What triggers an acceleration? It is more difficult to find supporting variables that move at about the time that growth begins to accelerate than to identify variables that distinguish growth episodes from nonepisodes. Our analysis suggests that economic liberalization increased the probability of an acceleration in the two decades by nearly 10 percent. But the models did a poor job of predicting the timing of accelerations, indicating that, at this stage, they still fall short of providing policy guidance. One encouraging finding for policymakers is the link between the quality of a country's policies and the propensity for an acceleration. The analysis showed a link between the occurrence of an episode and indicators of macroeconomic stability, institutional quality, trade openness, and productivity. But it did not show a link between resource availability or geography and an episode, suggesting that a growth spurt is feasible for most countries in the region.
What makes an acceleration last for at least 10 years—that is, what makes it a sustained acceleration? Sustained accelerations are associated with lower debt relative to unsustained accelerations, confirming our finding for the 1990s accelerations. This finding has two complementary implications: debt relief for heavily indebted poor countries has the potential to spur sustained growth, at least when other growth triggers are operative, and countries experiencing growth need to avoid accumulating excessive debt.
Economic and political institutions also influence the incidence and timing of growth accelerations. Studies point to a complex interaction of growth with institutional quality: good institutions promote growth, but growth also supports institutional improvement. Institutional indices can be classified into those that measure the quality of political institutions (for example, accountability mechanisms) and those that measure the quality of economic institutions (such as security of property rights and contract enforcement). Evidence indicates that both types of indices improve during growth accelerations but also that improvement in economic institutions has been sluggish in recent years following progress in the 1990s. Nevertheless, IMF researchers have found that even though the quality of institutions in SSA today may not be high by global standards, it is high enough in significant parts of the region when compared with the starting points of other countries that have succeeded in generating sustained growth (see "Levers for Growth," page 28 of this issue).
Ensuring that growth is pro-poor
Although economic growth is essential for achieving the MDGs, especially for poverty reduction, policymakers may wonder if growth alone can improve the welfare of the poor. For that reason, recent studies have focused on the elements of pro-poor growth. There is no single definition of pro-poor growth, but a recently completed multidonor study specified that pro-poor growth could be measured by the growth in incomes of those below the poverty line (Agence Française de Développement and others, 2005). However, other researchers (for example, the UNDP Poverty Centre) have argued that pro-poor growth should reduce the income gap between the poor and the nonpoor. The former measure has the advantage of directly linking growth to poverty reduction, which is embodied in the MDG for poverty.
Three components can alter poverty levels over time: the rate of economic growth, the response of poverty to that growth (known as the elasticity of poverty to growth), and changes in income distribution. However, studies show that almost all change comes from just the rate of growth.
In Africa, the role of inequality is worth examining, given the sizable changes in income distribution since 1980—an indicator that in other regions tends to be quite stable over time. We know that changes in inequality may either help reduce poverty further or offset some of the poverty reduction delivered by growth. A recent study on the contributions of growth and changing inequality to the total change in headcount poverty for five SSA countries since the early 1990s illustrates this point (see Chart 4). A statistical technique is used to break the actual change in the headcount measure of poverty into a component attributable to overall economic growth (the rate and the elasticity of poverty to growth) and one attributable to the change in income distribution. This allows for the fact that a narrowing gap in incomes may lift some people above the poverty line in the short term. A rising income gap, though, does not necessarily mean higher observed poverty: the beneficial impact of growth on poverty reduction must also be taken into account. When the two growth and inequality components shown separately for each country are added, they equal the total change in poverty.
For example, Uganda's impressive growth led to a large reduction in poverty, but an increase in inequality over the same period worked against further poverty reduction; if inequality had not changed, poverty would have been lower in 2003 with the same rate of growth. Burkina Faso showed more modest overall growth and, therefore, a smaller contribution of growth to poverty reduction, but the decline in inequality also reduced poverty. Poverty reduction in Ghana was due almost entirely to growth because the change in inequality over the period was small; Senegal is very similar to Ghana, but the offset to poverty reduction resulting from rising inequality is slightly larger. Zambia, which experienced sluggish overall growth, struggled during the 1990s to achieve any poverty reduction at all.
What elements of growth promote additional poverty reduction through stable or declining inequality? First, growth in agriculture. In the past, the sector was hamstrung by a policy bias against farmers, although much of the underlying policy framework has now been removed. Policymakers have been concerned that real exchange rate appreciation attributable to aid inflows might put agricultural exports at a disadvantage and undermine their efforts to rebuild the sector's competitiveness. But there is little empirical evidence that aid inflows have crowded out export-oriented agriculture. Second, allocating a significant portion of aid inflows to infrastructure helps to raise the productivity of all sectors in the economy and to offset the need for domestic prices to rise as demand increases, thus protecting competitiveness.
The main challenge facing SSA is how to accelerate and sustain growth. Our analysis suggests that accelerations are spurred by strong trade growth, improvements in broad measures of policy and institutional soundness, and political liberalization, and that they have been accompanied by higher investment and TFP—one reason why the limited investment response to SSA's reforms is so worrying. Encouragingly, a fair number of the countries that experienced accelerations succeeded in sustaining them for 10 years. They had stronger real exchange rate depreciations, higher investment, and lower debt burdens than countries that did not sustain their accelerations. But even countries that have sustained a 10-year growth acceleration need to do more if Africa is to dramatically reduce poverty over the decade ahead.