An analysis of issues shaping Africa’s economic future

u The economic outlook for Sub-Saharan Africa remains robust, but growth is vulnerable to lower commodity prices and a slowdown in capital flows u The frequency and strength of growth spurts have increased u Growth has shifted the structure of African economies in favor of the resources and services sectors

Summary u Global growth has been weak, with divergent trends in high-income countries, and below long-run growth levels in developing countries.
u Sub-Saharan Africa is growing at a moderate pace, reflecting in part a slowdown in some of the region's large economies. Public infrastructure investment, a rebound in agriculture, and a buoyant services sector are key drivers of growth in the region.
u Prospects for the region remain favorable, despite headwinds. External risks of higher global financial market volatility and lower growth in emerging market economies weigh on the downside. In several Sub-Saharan African countries, large budgetary imbalances are a source of vulnerability to exogenous shocks and underscore the need for rebuilding fiscal buffers in these countries.
u A key risk on the domestic side is a contagion of the Ebola outbreak. Without a scale-up of effective interventions, growth would slow markedly not only in the core countries (Guinea, Liberia, and Sierra Leone), but also in the subregion as transportation, cross-border trade, and supply chains are severely disrupted.
u Sub-Saharan Africa is lagging sharply in achieving the Millennium Development Goals (MDGs); for example, the region has achieved only a third of the poverty target of halving the proportion of people living under $1.25 a day, while globally this target has already been met. In addition, there is considerable variation across countries in how much progress is being made on the MDGs.
u The region's pattern of growth and economic transformation has implications for poverty reduction.
In Sub-Saharan Africa, growth in agriculture and services has been more effective at reducing poverty than growth in industry.
u Structural transformation has a role to play in accelerating poverty reduction in Sub-Saharan Africa.
Increasing agricultural productivity will be critical to fostering structural transformation. Boosting rural income diversification can facilitate this transformation, as well. Investments in rural public goods and services (for example, education, health, rural roads, electricity and ICT), including in small towns, will be conducive to lifting productivity in the rural economy.
u Although Sub-Saharan Africa's pattern of growth has largely bypassed manufacturing, growing the region's manufacturing base, especially by improving its fundamentals-lower transport cost, cheaper and more reliable power, and a more educated labor force-will benefit all sectors.
Section 1: Recent Developments and Trends u Following weaker-than-expected growth in the first half of 2014, a modest pickup in global growth is expected in the second half of the year, lifting growth to around 2.6 percent in 2014, to 3.2 percent in 2015, and to an average yearly rate of 3.3 percent during 2016-17.
u Despite headwinds, medium-term growth prospects for Sub-Saharan Africa remain favorable. Regional gross domestic product (GDP) growth is projected to strengthen to an annual rate of 5.2 percent during 2015-16 from 4.6 percent in 2014, and to rise to 5.3 percent in 2017.

GLOBAL ECONOMY
Global growth has been weaker than expected, amid mixed performance in a number of major countries (figure 1). In the United States, the Euro Area, and Japan, it averaged 0.6 percent in the first half of 2014, but growth across these countries has diverged considerably. Growth in the United States has been gathering momentum, but the Euro Area and Japan appear to be stagnating. Supported by rising employment and investment growth, a still accommodative monetary policy, and easing fiscal consolidation, U.S. growth recovered strongly in the second quarter from a sharp contraction in the first quarter. The recovery in the U.S. economy is expected to gather pace in the second half of 2014 as better employment prospects support real income growth and confidence, fiscal consolidation pressures ease, and investment rises in line with strong profits and favorable financing conditions. GDP growth for 2014 is projected at about 2.1 percent, rising above trend in 2015 to around 3.0 percent.
Meanwhile, growth appears to have stagnated in the Euro Area and Japan. With the strength of the recovery continuing to be impaired by weak domestic demand, ongoing balance sheet adjustments, and a fragmented banking sector, Euro Area GDP was flat in the second quarter, following a small uptick in the first quarter. A slow improvement in credit and labor market conditions should provide some momentum ahead, but investment prospects remain subdued and precautionary savings still high. Exports should gradually pick up, supported by strengthening demand from the United States and a weakening euro. Growth is expected to resume in the second half of the year, with overall growth for 2014 projected to reach about 1.0 percent. In Japan, a sales tax hike in April caused a more significant contraction in activity than initially expected, while exports failed to pick up despite a weak yen. Monetary policy accommodation and reform commitments will provide ongoing support, but fiscal consolidation is expected to keep domestic demand subdued throughout 2015, with exports recovering only slowly. Real GDP growth is projected to average 1.0 percent in 2014, down from 1.5 percent in 2013.
A modest pickup in global growth is expected in the second half of 2014, lifting growth to 2.6 percent for the year

Forecasts
In developing countries, growth is expected to remain below long-run average levels in most regions. This is partly due to weak external demand and withdrawal of fiscal stimulus, especially in major emerging market countries. In China, growth is expected to slow from 7.7 percent in 2013 to 7.4 in 2014 and to an average of 7.1 percent during 2015-17, as it makes the transition away from an investment-led growth strategy toward greater emphasis on domestic consumption. However, developing countries are expected to benefit from the projected pickup of economic activity in high-income countries in the second half of 2014, which is expected to lift trade growth modestly. This should provide some impetus to developing countries, particularly those for which the United States is a major trading partner. In addition, still accommodative U.S. monetary policy and the European Central Bank's announced policy measures to ease credit are supporting capital flows to most developing regions.
Overall, a modest pickup in global growth is expected in the second half of 2014 to raise annual growth to around 2.6 percent for the year. Growth is expected to strengthen to 3.2 percent in 2015 and to an average yearly rate of 3.3 percent during 2016-17. High-income country growth is expected to come in at 1.8 percent in 2014, up from 1.3 percent in 2013, and to average 2.3 percent a year during 2015-17. Developing-country growth is expected to edge lower to 4.4 percent in 2014 before rebounding to 5 percent in 2015, and to average 5.4 a year during 2015-17.
Monetary policy tightening in the United States, deflation in the Euro Area, and geopolitical tensions in a number of regions present risks to the global outlook. Monetary policy in high-income countries is expected to diverge. The U.S. Federal Reserve is projected to start raising policy rates in mid-2015, which carries the risk of financial market volatility. In contrast, in the Euro Area, where inflation continues to drift downward, deflation risks are increasing, and the European Central Bank has announced additional easing measures to begin in October. In Japan, where inflation expectations are still weakly anchored, loose monetary policy is projected to continue. Developing countries and emerging market economies are especially vulnerable to bouts of financial market disruptions and volatility as a result of changes in monetary policy in high-income countries or weakening investor sentiment if geopolitical tensions (for example, in Russia or Iraq) or health concerns (for example, from the Ebola virus in West Africa) escalate. Such disruptions can trigger a sharp withdrawal of funds from these countries and place strong downside pressure on domestic currencies.

RECENT ECONOMIC DEVELOPMENTS IN SUB-SAHARAN AFRICA
Growth has moderated in Sub-Saharan Africa, reflecting in part a slowdown in some of the region's largest economies. Growth slowed notably in South Africa, the region's second-largest economy, due to structural bottlenecks, labor unrest and low investor confidence. The South African economy expanded a modest 1.0 percent year-on-year in the second quarter of 2014, its lowest growth rate since the 2009 financial crisis, slowing from an already weak 1.6 percent expansion in the first quarter as strikes in the platinum sector dragged mining and manufacturing output down. In Angola, oil production declined with mature oil fields coming off stream, causing a marked deceleration in growth. By contrast, economic activity strengthened in Nigeria, the region's largest economy. GDP advanced 6.5 percent year-on-year in the second quarter, up from a 6.2 percent expansion in the first quarter. Growth also remained robust in many of the region's low-income countries including, notably, Cote d'Ivoire, Ethiopia, Mozambique, and Tanzania. However, the Ebola outbreak severely disrupted economic activity in Guinea, Liberia, and Sierra Leone, causing growth to slow in these countries.
Overall, GDP growth in the region is projected to be 4.6 percent in 2014, the same as in 2013 (figure 2). Excluding South Africa, average GDP growth for the rest of the region is expected to hold steady at 5.6 percent, a faster pace than other developing regions excluding China. GDP per capita growth is expected to hold steady at 2.1 percent in 2014.
Public infrastructure investment, a rebound in agriculture, and a buoyant services sector were key drivers of growth in the region. Substantial infrastructure investment, including in ports, electricity capacity, and transportation continued to be undertaken across the region, helping to sustain high growth rates in many countries. In Cote d'Ivoire, a strong increase in cocoa production and rice output boosted agriculture growth, and Ethiopia's robust growth continued to be supported by the agriculture sector and public investment, particularly in infrastructure. Services sector expansion, led by transport, telecommunications, financial services, and tourism, is spearheading overall economic growth in countries such as Nigeria, Tanzania, and Uganda. However, the region is seeing a slowdown in foreign direct investment (FDI) flows, an important source of financing of fixed capital formation: FDI is expected to be around $29 billion in 2014, down from nearly $32 billion in 2013. The slowdown reflects subdued global demand and weaker commodity prices, especially of metals. Overall, net capital inflows to the region are projected to amount to 4.3 percent of regional GDP in 2014 compared with 5.1 percent in 2013 as FDI and portfolio investment flows register significant declines.
The fiscal policy stance remained expansionary across the region during 2014, resulting in large budgetary imbalances. The fiscal deficit for the region as a whole is projected to widen to 3.6 percent of GDP from 3.1 percent of GDP in 2013, with significant variations among countries. The fiscal position of many countries deteriorated due to increasing current and capital expenditures, but also to declining revenues, notably among oil-exporting countries facing both declining production and lower oil prices. In Angola, for example, the overall fiscal balance is projected to deteriorate sharply from a surplus of 0.3 percent of GDP in 2013 to a deficit of 4.1 percent of GDP in 2014, with declining oil revenues and an increase in the wage bill as the main factors behind the deficit. At the same time, several countries took measures to control expenditures that are helping to stabilize or reduce their fiscal deficits. Nigeria's overall deficit is projected to narrow to 2.9 percent of GDP in 2014, driven in part by higher oil and gas revenues, but also by expenditure restraint, including a decline in subsidies as a percent of GDP. In Senegal, the overall fiscal deficit is projected to decline to 5.1 percent of GDP in 2014 from 5.5 percent in 2013, mainly on the back of expenditure restraint as the authorities implement measures aimed at streamlining less productive spending, including on wages and salaries and goods and services.  Notwithstanding ongoing efforts to control spending, strengthen revenue administration, and undertake tax policy reforms, the widening fiscal deficit for the region as a whole suggests uneven progress in rebuilding fiscal space, and underscores the need for countries in the region to take advantage of the benign global financial conditions to rebuild depleted fiscal buffers. This will entail curbing current spending, which has contributed to the expansionary fiscal stance, and focusing on quality and efficiency of investment spending. Fiscal consolidation measures would help contain the rising primary fiscal deficits, which were a major factor behind the buildup in debt-to-GDP ratios observed in 2013, notably among frontier market countries.
Inflation edged up in the region, but was not a major concern in most countries (figure 3). Overall, inflation rose from 6.2 percent (y/y) in the beginning of 2014 to 6.9 percent (y/y) in July. The uptick in inflation was most visible in the frontier market countries that also sustained large currency depreciations-notably Sub-Saharan African countries are turning to sukuk bonds for financing infrastructure. A sukuk (Islamic bond) is a financial instrument representing a risk-sharing way to provide capital, typically to governments. Through this mechanism, sukuk (bond) holders own shares of assets relating to specific projects or investments for which the capital was provided. Sukuk holders are entitled to revenue generated from these assets. Global sukuk issuance was $120 billion in 2013, down 21 percent from 2012, yet still over five times the 2008 total (Gelbard et al. 2014). Most sukuk activity is concentrated in the Gulf Cooperation Council countries and Malaysia, but African countries are beginning to access this source of funds.
Earlier this year, Senegal issued the continent's first major sovereign sukuk at 100 billion Senegal francs (US$208 million). The nation's sukuk will be used to invest in infrastructure for improved water and power distribution. While Senegal might have issued Sub-Saharan Africa's largest sukuk, the popularity of Islamic financing as a source for government debt has increased across the region. In July 2012, Sudan raised 955 million Sudanese pounds (US$165 million) through this bond instrument, and Gambia routinely sells this financial instrument. This September, South Africa issued a US$500 million sukuk. The nation's five-year sukuk was priced at 3.9 percent. Other countries in the region have expressed an interest in issuing this financial instrument, as well.
Through sukuk, African countries can diversify their investor base, while deepening local capital markets. Since funds from sukuk can be used only to finance projects reflecting real economic activity, most such instruments are used to fund infrastructure projects. Thus, sukuk provides an important mechanism for long-term financing of large public projects.

BOX 1:
Sub-Saharan African countries are tapping the sukuk market basis of macroeconomic imbalances and the pace of reforms (figure 8). This highlights some of the risks inherent in borrowing in global financial markets. Some of the region's countries are turning to the sukuk market as a source of long-term financing for infrastructure needs (box 1).
Sovereign bond spreads fell across the board, but they remain relatively high in some countries

Basis points
Source: Bloomberg.
The region's main currencies have generally stabilized (figure 9), following significant volatility early in the year, which could help contain inflationary pressures going forward. The Zambian kwacha continued to slide in the first half of the year, weakening by over 20 percent, before rebounding. Concerns about loose fiscal stance and low external reserves led to bouts of renewed pressure on the Ghanaian cedi, which depreciated by more than 40 percent against the U.S. dollar in the first nine months of the year.

ECONOMIC OUTLOOK FOR SUB-SAHARAN AFRICA
Despite headwinds, medium-term growth prospects for Sub-Saharan Africa remain favorable. Regional GDP growth is projected to strengthen to an average annual pace of 5.2 percent during 2015-16, from 4.6 percent in 2014, and to rise to 5.3 percent in 2017. Under this baseline scenario, GDP per capita will rise steadily from an estimated 2.1 percent in 2014 to 2.6 percent in 2015, reaching 2.8 percent in 2017.
Public investment in infrastructure, increased agricultural production, and a buoyant services sector are expected to continue to support growth in the region. The growth pickup is expected to occur in a context of reduced support from commodity prices and net FDI flows as global demand remains subdued. Overall, Sub-Saharan Africa is forecast to remain one of the fastest-growing regions. Private consumption in the region is expected to remain strong during 2015-17. Reduced imported inflation, aided by a benign global inflationary environment and stable exchange rates, and adequate local harvests are expected to help contain inflationary pressures in most countries, which should allow for some gains in real disposable incomes. However, currency-induced price pressures, which could weigh on consumer sentiment and slow private consumption growth, remain a concern for countries with a heavy reliance on portfolio flows, including Ghana and South Africa.
Government consumption is projected to grow at a moderate pace as governments across the region strive to restrain current expenditures, allowing for some fiscal consolidation to take place. The > 1 0 expansionary fiscal policy stance that has led to budgetary imbalances has been found to be linked more systematically to current expenditures than to capital spending, with insignificant correlation to capital expenditures. Ongoing efforts to contain wages and salaries and streamline less productive expenditures on goods and services are therefore important consolidation steps. In this context, steps by Ghana and Zambia to control spending will be key to correcting their large fiscal and external imbalances.
Net exports are again projected to make a marginal contribution to GDP growth in the region over the forecast horizon. The contributions of net exports will be constrained by lower commodity prices, which could be exacerbated by low output in countries such as Angola, where production is stagnating. In metal-exporting countries, increased output would mitigate the weakness of metals prices. On the import side, the demand for capital goods is projected to remain strong, as governments continue to frontload infrastructure investments and private consumption remains strong. Reflecting these trends and the weakening of commodity prices, the current account deficit in the region is projected to widen from an estimated 2.4 percent of GDP in 2013 to an average of 3.0 percent of GDP in 2015 and 2016.
At the country level, growth is expected to remain robust in Nigeria, supported by the continued expansion of nonoil sectors, particularly the services sector, which now accounts for more than 50 percent of GDP. South Africa is expected to experience steady but slow economic growth as gradually improving net exports help mitigate the drag from monetary and fiscal policy tightening; and infrastructure bottlenecks, especially in the electricity sector, are progressively alleviated in the coming years and help to lift investment sentiment. Among other middle-income countries, high interest rates and rising inflation due to currency depreciation are expected to slow economic activity, including notably in Ghana.
In low-income countries, political stability and continued investment in infrastructure should keep growth rates high. The Ebola outbreak is expected to severely disrupt activity in key economic sectors in Guinea, Liberia, and Sierra Leone and to slow growth in these countries in 2014. Economic spillovers are, however, expected to be modest and contained to Ghana and Nigeria, the main transportation hubs in the West Africa subregion, provided further contagion can be controlled rapidly. The economic dislocation will be more adverse under a pessimistic scenario of containment.

RISKS TO THE ECONOMIC OUTLOOK
The outlook is subject to significant downside risks stemming from both domestic and external factors. Key domestic risks include a more widespread extension of the Ebola outbreak, a deterioration of the security situation in a number of countries, and risks associated with expansionary fiscal policy and currency weaknesses. A sudden increase in volatility in financial markets and lower growth in emerging markets are among the major external risks to the region's outlook.

Domestic risks
Widespread contagion of the Ebola outbreak: Without a scale-up of effective interventions, the virus could spread more rapidly than assumed in the baseline, and become harder to control as it reaches large urban centers and new countries with a weak public health infrastructure and low institutional capacity to deal with the outbreak. In addition to the loss of lives, affected countries would suffer a sharper decline in output, with growth slowing markedly not only in the core countries but also in the subregion as transportation, cross-border trade, and supply chains are severely disrupted. See box 2.
The 2014 outbreak of Ebola in West Africa is exacting a heavy humanitarian toll. Beyond the tragic loss in human lives and suffering, the Ebola epidemic is already having a measurable economic impact in terms of foregone output, higher fiscal deficits, rising prices, and lower real household incomes and greater poverty. These economic impacts include the costs of health care and foregone productivity of those directly affected but, more important, they arise from the aversion behavior of others in response to the disease.

Channels of impact
The impact of the Ebola epidemic on economic well-being operates through two distinct channels. First are the direct and indirect effects of the sickness and mortality themselves, which consume health care resources and subtract people either temporarily or permanently from the labor supply. Second are the behavioral effects resulting from peoples' fear of contagion, which in turn leads to a fear of association with others and reduces labor force participation, closes places of employment, disrupts transportation, and motivates some government and private decision makers to close seaports and airports.

Short-term impact on economic activities
Since the escalation of the Ebola outbreak in July 2014, there has been a sharp disruption of economic activities across sectors in Guinea, Liberia, and Sierra Leone. The largest economic effects of the crisis are those resulting from changes in behavior (driven by fear), which have resulted in generally lower demands for goods and services and consequently lower domestic income and employment.
The services sector has been hit particularly hard across all three countries. In Liberia, where services comprise approximately half of the economy and employ nearly 45 percent of the labor force, wholesale and retail traders have reported a 50 to 75 percent drop in turnover relative to the normal amount for the trading period. The domestic transport sector has been severely affected, with gasoline and diesel sales down by 21 and 35 percent, respectively. Average hotel occupancy across the three countries has fallen from 60 to 80 percent year-round before the crisis to 13 to 40 percent, resulting in many layoffs of hotel workers. In Sierra Leone, a sharp reduction in international flights servicing the country has increased its isolation from global markets.
Significant impacts have also been felt in agriculture, which is the mainstay of much of the population in these countries. In Guinea, which is among the poorest countries in West Africa, agriculture in Ebola-affected areas has been hit by an exodus of people from these zones, affecting the production of key export commodities, such as coffee, cocoa, and palm oil. Although robust price data are not yet available, reports indicate rice price spikes of up to 30 percent in Ebolaaffected areas in Sierra Leone. A Food and Agriculture Organization rapid assessment in the district of Kailahun, Sierra Leone, indicates that at least 40 percent of farmers have either abandoned their farms and moved to new, safer locations or have died. In the most productive agroecological areas, about 90 percent of the plots have not been cultivated. This is due in part to current restrictions on movement, and to an expressed fear by farmers of meeting or even sharing working tools.
Mining has not yet been severely affected by the Ebola outbreak in West Africa, and companies have indicated they intend to maintain their originally planned production levels to the extent possible. Nonetheless, in Liberia, investments to expand capacity to 15 million tons per year have been put on hold, and one major mining company closed its operation in August, contributing to an expected contraction in the mining sector of 1.3 percent in 2013 compared with an initial projection for growth above 4 percent.

BOX 2:
The economic impact of the Ebola epidemic a > 1 2 Deterioration of the security situation: The conflict in South Sudan and security concerns in northern Nigeria could deteriorate further, with significant regional spillovers. Since the outlook for a political settlement remains poor, the South Sudan conflict could escalate further and significantly disrupt trade in the subregion, including in Kenya and Uganda. An intensification of the Boko Haram insurgency in Nigeria could further disrupt agricultural production in northern Nigeria and spur violence in the subregion, prompting governments in neighboring countries, including Cameroon, to divert additional budgetary resources to security-related expenditures.
Heightened fiscal vulnerabilities: Budgetary concerns and currency weaknesses will remain sources of vulnerability for many countries in the region. Notably, a continuation of loose fiscal policy could lead to a further deterioration of already weak fiscal positions in some countries. Fiscal buffers would be further depleted, increasing the vulnerability of the countries concerned to exogenous shocks. In Ghana and Zambia, budget risks will remain salient given public sector wage pressures. In Zambia, for example, 37 percent of the 2014 budget is allocated to the public wage bill, thanks to the 2013 public sector wage award. Currency concerns are also likely to continue. A combination of weak export growth, high import demand, and negative investor sentiment could cause the currencies of frontier market economies to weaken significantly against the dollar, contributing to inflation in these countries. While moderate

BOX 2: Continued
Preliminary estimates indicate that GDP growth in 2014 could be sharply lower in Guinea (from 4.5 percent to 2.4 percent) and Liberia (from 5.9 percent to 2.5 percent), with a loss of over 3 percentage points for Sierra Leone (from 11.3 percent to 8.0 percent). In terms of foregone output, this amounts to a total of US$359 million across the three countries, already a major loss. The fiscal impact of the crisis has also been enormous, emanating from the combination of revenue shortfalls from reduced economic activities and increased expenditures, particularly for health and social protection. Financing gaps in 2014 for the three core countries range from US$80 million to US$120 million, totaling over US$290 million. Slow containment and exponential growth of the disease will lead to even greater financing gaps in 2015.

Medium-term impacts
In light of the considerable uncertainty about the future trajectory of the epidemic, two epidemiological scenarios were used to estimate the medium-term economic impact, extending to the end of 2015. One scenario corresponds to rapid containment within the three most severely affected countries, while a second scenario corresponds to slower containment in the core three countries, with some broader regional contagion.
The likely economic impact of the Ebola epidemic will be significant for the affected countries in any plausible scenario. However, the scenario in which the epidemic is not swiftly contained promises to leave a much deeper adverse economic impact. Even in the absence of broad epidemiological contagion, those countries in the region with tourism are already feeling the economic impact. For example, in Gambia-where tourism accounts for nearly 11.5 percent of GDP-an estimated 65 percent of hotel reservations have been cancelled since the beginning of the crisis. Economic modeling suggests that the economic impact for the region as a whole is likely to run into the billions and potentially into the tens of billions.

a. This section is adapted from The World Bank Group, "The Economic Impact of the 2014
Ebola Epidemic: Short-and Medium-Term Estimates for West Africa," September 17, 2014. food prices and prudent monetary policies have seen inflation remain low in many countries in 2014, currency-induced price pressures will pose a persistent threat.

External risks
Higher market volatility: A sudden increase in risk premia and volatility in global financial markets from their current low levels remains a significant downside risk for the region. It would not only adversely affect South Africa, which depends heavily on portfolio capital flows to finance its current account balance, but also frontier market countries such as Ghana, Nigeria, and Zambia, which have increased their reliance on external market financing. Recent episodes of capital market volatility suggest that countries with large macroeconomic imbalances would face strong downward pressure on the exchange rate and high currency-induced inflation.
Lower growth in emerging market economies represents another significant downside risk to the regional outlook. A sharper slowdown in economic activity in emerging markets, particularly in China, would most likely lead to a lower demand for commodities, which could see a significant decline in their price, especially where supply is abundant. A further decline in the already depressed price of metals, particularly iron ore, gold, and copper, will severely affect a large number of countries in the region.
In countries such as Mauritania, Mozambique, Niger, Tanzania, and Zambia, metals account for a large percentage of exports, while their extraction has led to significant FDI flows. A protracted decline in metals prices could lead to a significant decline in exports and cause foreign investors to scale down their operations in these countries, which could adversely affect their growth momentum. Actual outcomes will, of course, depend upon other growth opportunities. For example, in Tanzania, the development in the gas sector could well mask a potential declining trend in metals in the country in the long run.
Simulation results suggest that the income effects of a sharper decline of commodity prices on Sub-Saharan African economies could be significant. A scenario is considered where the prices of metals (aluminum, copper, gold, iron ore, and silver) and agricultural commodities (cocoa, coffee, cotton tea,  Sixty-five percent of the maternal mortality target has been met, with regional progress outpacing global progress. Nevertheless, women in Sub-Saharan Africa face a lifetime risk of maternal death that is about 90 times greater than for women in high-income countries.   The water and sanitation targets remain a problem for people in most developing countries. Sub-Saharan Africa is lagging the most, with 36 percent of its population lacking access. Access to improved water sources and improved sanitation facilities is correlated with wealth. In Sub-Saharan Africa, almost 90 percent of the richest fifth of the population use improved water sources, while only 35 percent of the poorest fifth of the population do (WHO and UNICEF 2013). However, this region also had the worst starting position, with an even worse situation in rural areas, where only 23 percent of the population has access to improved sanitation; in urban areas the access rate is 20 percentage points higher. This large disparity in Sub-Saharan Africa is the principal reason the MDG sanitation target is unlikely to be met. Within-country progress on the MDGs is uneven, with considerable inequity in outcomes. For example, on the primary school completion rate, many children that start school drop out before completing the primary stage, discouraged by cost, distance, physical danger, and failure to progress. Access to primary education has also been inequitably distributed across households. The result is unequal outcomes on the education MDG. For example, in Mali, Niger, and Uganda, primary completion rates for children in the lowest two quintiles of the income distribution are between 20 and 30 percent, while for the highest quintiles, primary completion rates range between 60 and 100 percent (figure 13).

Sub-Saharan
> 1 6 Section 2: Economic transformation and poverty reduction in Sub-Saharan Africa u Nearly two decades of strong growth are transforming the structure of Africa's economies, but not as expected.
u Sectoral composition of output has shifted in favor of services, with this sector's growth outpacing that of agriculture and industry: Between 1995 and 2011, annual per capita growth in the services sector averaged 2.6 percent compared to 1.7 percent in industry and less than 1 percent in agriculture.
u While agriculture's share in output has declined, so has that of industry. However, there are divergent trends within industry, with extractives gaining output share and manufacturing's share declining.
u Export diversification has been limited, as well; primary commodities continue to account for threefourth of Sub-Saharan Africa's total goods exports, and the share of the region's top five exports in total exports has climbed to 60 percent in 2013 from 41 percent in 1995.
u Labor shifts reflect sectoral changes, but almost 60 percent of Africa's jobs and 78 percent of its poor workers continue to have a foot in agriculture, the sector with the lowest productivity. The extent of reallocation of labor to high-productivity, nontraditional activities has been limited; the movement of workers has been out of agriculture and into services, not manufacturing.
u The pattern of growth and economic transformation has implications for poverty reduction. In Africa, growth in agriculture and services has been more poverty reducing than growth in industry. In the rest of the world, by contrast, industry and services have a larger impact on reducing poverty.
u Structural transformation has a role to play in accelerating poverty reduction in the region. Increasing agricultural productivity will be critical to fostering structural transformation. Boosting rural income diversification can facilitate this transformation, as well. Investments in rural public goods and services (for example, education, health, rural roads, electricity, and ICT), including in small towns, will be conducive to lifting productivity in the rural economy.
u Although Sub-Saharan Africa's pattern of growth has largely bypassed manufacturing, growing the region's manufacturing base, especially by improving its fundamentals-a better business climate, lower transport cost, cheaper and more reliable power, and a more educated labor force-will benefit all sectors.

AFRICA'S GROWTH AND ECONOMIC TRANSFORMATION
African countries have seen a dramatic turnaround in economic expansion beginning in the mid-1990s.
Between 1995 and 2013, economic growth averaged 4.5 percent per year in real terms, comparable to the rate in the rest of the developing world and more than double the pace of growth of the previous 20 years. On a per capita basis, output growth has been more modest at 1. In the region's economic transformation, the services sector has been the big gainer, growing its share in GDP from 49 percent in 1995 to 55 percent in 2011.
The sectoral trend in other developing countries differs from that of Sub-Saharan Africa in important ways. A notable difference is that per capita growth in industry in other developing countries is much higher (4.8 percent) than in Sub-Saharan Africa, and matching the pace of growth in services.
Consequently, industry's share in output has climbed from 36 percent in 1995 to 39 percent in 2011

Sub-Saharan Africa
Resource-rich   In Sub-Saharan Africa, services have grown as a share of output over the last two decades, while agriculture and manufacturing have declined Sub-Saharan African countries diverges from this; although the share of agriculture in GDP is declining, so is that of manufacturing. The (informal) and formal services sectors have been absorbing workers and capital, respectively, in most countries, and their share in economic activity is on the rise broadly across the continent. But much of the growth in services has been in low-productivity activities.
The lack of industrialization in Africa's boom has fueled a debate on whether sustainable growth requires a shift in favor of manufacturing (McMillan and Rodrik 2011;McMillan, Rodrik, and Verduzco-Gallo 2013) or whether moving up the quality ladder in sectors where countries can exploit and build on their current patterns of comparative advantage can sustain growth (Hausmann and Hidalgo 2011). India's growth pattern suggests that a shift into high-productivity services, bypassing manufacturing, represents another path to sustainable growth (Ghani, Goswami, and Kharas 2012). Modern services, such as software development, call centers, and outsourced business processes, represent high value-added activities (similar to manufactured products) that can be important drivers of growth for innovative and technology-savvy countries. Similar opportunities may exist in staple and nonstaple agriculture by moving up the value chain as in Nigeria (cassava flour) and Kenya (flowers).

Diversification of exports:
Diversification of exports away from raw materials and commodities and toward better quality of existing products or new products often supports structural transformation in lowincome countries. There is some evidence that in the early stages of development, at per capita income levels below US$10,000, a country's basket of exports tends to diversify as it exports a wider product range-that is, diversification along the extensive margin (Klinger and Lederman 2009). At much higher incomes (such as in industrial economies), specialization leads to a smaller basket of exports-that is, specialization along the intensive margin (IMF 2014).
Sub-Saharan Africa's overall pattern of trade shows little export diversification during two decades of rapid growth, and exports remain concentrated in a narrow set of products. The share of the top five exports in the region's total merchandise exports increased from 41 percent to 60 percent between 1995   Evidence shows that several countries have been successful in growing the share of manufacturing in total goods exports. Among these are Senegal, Togo, and Uganda. In Uganda, the share of manufacturing exports has risen from an average of 3 percent of exports during 1990-2000 to 11 percent during 2001-12. The improvement reflects growth of both basic manufacturing goods and more technology-intensive products such as steel and iron rods and plastic tubes and pipes. In Senegal, chemicals and manufacturing products have played a role in lifting the country's share of manufactured exports in total exports to 33 percent from 16 percent over the same period. Other strong performers are Kenya, Madagascar, and Rwanda. Intraregional trade has contributed to the boost in manufacturing exports.
Trade in services offers an important path for diversification, as well. Modern services, such as software development, call centers, and outsourced business processes, are high value-added activities that hold the potential to be an important driver of growth in technology-savvy countries. Favorable global trends, such as the rising shares of services in global trade, and of modern services in total services, provide opportunities for export diversification. Some African countries, such as Kenya (with its successful expansion of mobile banking) and Mauritius (which has grown its tradable business and financial services), are well positioned to benefit from this global trend.

Change in sectoral composition of labor: What has this transformation meant for employment and jobs?
Although agriculture's output share has fallen to well under a fifth of GDP, 59 percent of the labor force continues to be employed in this sector ( figure 19). Moreover, agriculture employs more poor people than any other sector in Africa. After nearly two decades of economic growth, most Africans continue to earn livelihoods in the traditional economy or the informal sector. In short, the majority of African workers are engaged in nonwage employment-namely, farming and nonfarm small household enterprises. By contrast, labor is more evenly distributed across sectors in other developing countries, with agriculture accounting for 37 percent of employment; manufacturing for 24 percent; and services for 40 percent, representing the largest share.
Moreover, agriculture employs more poor people than any other sector in Africa. While just 5 percent of the poor are employed in industry and 16 percent in services, 78 percent of the poor rely on agriculture for their livelihoods ( figure 19). In other developing countries, the distribution of labor is similar among the poor, albeit with a higher proportion working in industry and services than is the case in Sub-Saharan Africa.
Where people work is largely correlated with a country's level of development. Figure 20, shows that the share of labor in agriculture declines with increasing GDP. The opposite is true of employment in services, which rises as GDP per capita rises ( figure 21). This is consistent with the agricultural sector employing more of the poor across countries, as well as within them (box 3). Controlling for income levels, the    19). This fact, however, is derived from looking at a snapshot of the composition of labor by sector. To understand the drivers of poverty reduction and sectoral transformation, it is also essential to look at the movements in the composition of labor over time and across the income distribution. Ideally, this would mean analyzing the employment trajectory of the poor (and nonpoor) using panel data to obtain estimates by country. Given the scant availability of such surveys in the region, an alternative is to track income ventiles instead of individuals, using crosssections from the Survey-Based Harmonized Indicators Program (SHIP).
To observe shifts in the sectoral composition of labor at different points in the income distribution, ventiles were constructed by dividing the total distribution of per capita monthly food and nonfood consumption expenditure (2005, purchasing power parity), with each group containing 5 percent of the population. Labor composition was categorized into that in the agriculture (agriculture and fishing), industry (mining, manufacturing, electricity, and construction), and services (commerce and transport, financial, insurance and real state, public administration, and other services) sectors. Rwanda (2005-10) and Senegal (2001-05) were selected for the analysis, because both these countries have experienced substantial reductions in poverty over the last decade. In Rwanda, the poverty headcount at US$1. Indeed, the conversion of growth into poverty reduction has been slower in Africa than in the rest of the developing world, with average growth elasticities of poverty reduction of -0.7 and -2, respectively (Christiaensen, Chuhan-Pole, and Sanoh 2013). Beyond the broad trends and averages, there is considerable diversity of experience. The growth elasticity of poverty reduction measured as changes in poverty headcount over changes in mean per capita income is illustrated in figure 26 for a subset of Sub-Saharan African countries. There is substantial heterogeneity in this indicator across countries; for example, a 1 percent increase in GDP in Cameroon is correlated with a reduction in poverty more than twice that seen in Zambia. In Burundi, the poverty-reducing effect of growth is only half that of the regional average. The growth elasticity of poverty reduction is positive in some cases, meaning that upward or downward movements in income are associated with poverty movements in the same direction. Data quality issues and lack of availability of recent estimates may be affecting these results but, more important, this reflects the fact that poverty reduction also depends on other conditions, such as inequality, both levels and changes, and sectoral and geographic patterns of growth.

Sub-Saharan Africa Rest of the world
An issue that arises is whether the pattern of growth has implications for poverty reduction. Namely, does poverty respond differently to growth in different sectors? There are several reasons why there could be differential impacts of sectoral growth on poverty reduction. One obvious reason is due to differences in size of sectors. Thus, even if two sectors have similar growth rates, the effect on poverty can vary because of differences in sector size. But there are also other reasons. Importantly, for output growth to lift people out of poverty, poor people need to be able to participate in the growth-either by contributing to it directly, or by benefiting from it through redistribution. But poor people often face constraints to moving to places where the growth takes place (such as, the capitals), and the political economy constraints to redistribution are typically equally challenging. Thus, poor people are more likely to participate in and benefit from growth if it happens in the activities and areas where they work or live. Consequently, the geographic and sectoral patterns of growth are likely to matter in reducing poverty.
Since poor people are generally engaged in agriculture and concentrated in rural areas, responsiveness of poverty reduction to agricultural growth and rural economic growth is likely to be higher than to growth in other sectors. Another reason is that poor people usually have low labor skills. Growth in sectors with a higher intensity in the employment of unskilled labor is more poverty reducing than growth in other sectors (Loayza and Raddatz 2010). Differences in inequality of assets across sectors can also impact the poverty-reducing effects of sectoral growth. For example, unequal distribution of land ownership may constrain the poverty-reducing effect of agricultural growth (Christiaensen, Demery, and Kuhl 2011). In sum, the pattern of growth matters for poverty reduction because of heterogeneity in the participation of the poor by sector. The importance of agriculture in reducing poverty in low-income countries, especially in Africa, stems from the fact that a majority of the poor-three-quarters of the world's poor live in rural areas-depend on agriculture for their livelihoods (Filmer and Fox 2013;World Bank 2007). It follows, therefore, that growth in this sector will be more beneficial for the poor than growth in sectors in which the poor exhibit lower participation. Christiaensen, Demery, and Kuhl (2011) find that agricultural growth has a greater impact on reducing poverty than nonagricultural growth. Diao, Thurlow, and Fan (2012) find (for six Sub-Saharan African countries) that the impact on the poverty rate of a 1 percent annual increase in GDP per capita driven by agricultural growth is between 53 percent to 127 percent larger than from an equivalent increase in GDP fueled by nonagricultural growth.
But there is also evidence to suggest that not all agricultural growth is equally poverty reducing. For example, growth that positively affects smallholder staple crop productivity, as opposed to export crops, has been found to be more poverty reducing (Diao, Thurlow, and Fan 2012). This follows from the larger multiplier effects and growth elasticities of poverty for such crops-1 percent growth in agriculture driven by cereal or root/tuber productivity growth generates a larger decline in national poverty than a 1 percent growth in agriculture driven by growth in export crops. While export crops typically have higher value and growth potential than food crops, the latter are usually more effective at generating economywide growth and reducing national poverty.  Ferreira et al. (2010), for China from Ravallion and Chen (2007), and for India from Ravallion and Datt (1996).
Note: The results refer to time periods of analysis from 1985to 2004for Brazil, 1981-2001for China, and 1951to 1991

Sectoral composition of growth and poverty reduction
Country-level results show substantial differences in the sectoral impact of growth on reducing poverty.
We want to examine from a regional perspective how the sectoral composition of growth has mattered for Africa's poverty reduction. Using cross-country analysis, the link between the composition of growth GDP is disaggregated into three broad output sectors: agriculture, industry, and services. 6 This level of disaggregation allows us to include 29 countries in Sub-Saharan Africa and 31 in the rest of the world in our analysis. Sectoral output is value added in GDP by sector, and is from the World Development Indicators database. A finer disaggregation, especially of the manufacturing sector, would be desirable, but issues of data quality are a concern. Data quality issues suggest that results need to be interpreted cautiously.
The empirical analysis confirms that in addition to overall growth of GDP per capita, what matters for poverty reduction is where growth comes from-that is, agriculture, industry, or services (table 2).
Growth in the agricultural and services sectors is strongly associated with aggregate poverty reduction, but growth in industry does not have a significant effect on lowering poverty. Statistical tests reject the hypothesis that the sectoral composition of growth does not matter. 7 The impact of GDP originating in the agricultural and services sectors (controlling for size) on national poverty is fairly comparable. A 1 percent increase in GDP per capita led by agricultural growth reduces poverty by 0.67 percent, and the same increase led by services reduces it by 0.96 percent. These poverty-reducing effects come from two sources: sectoral growth in GDP per capita, and the share of each sector in total GDP per capita. If, on average, the share of agriculture in the Sub-Saharan Africa sample were half of what it is, the agricultural sector would have to grow twice as fast to achieve a 0.67 percent reduction in poverty. Likewise, if the services sector share is increased, it would require a correspondingly smaller growth rate to achieve the same reduction in poverty. Larger shares of GDP per capita compensate for lower growth rates, or demand less of growth, for poverty reduction than in sectors with declining shares such as agriculture.
4 For the empirical analysis the following model, from Ferreira et al. (2010), was estimated separately for headcount, poverty gap, and poverty gap squared using a standard OLS: The empirical results also show that differences in the impact of growth originating in agriculture and services mostly disappear when looking at the effects of the poorest of the poor (as captured by the poverty gap squared measure), underscoring that growth in agriculture is especially beneficial for the poorest. Overall, given limited growth in agricultural GDP over the past two decades (0.9 percent per capita per year), the somewhat muted reduction in poverty does not surprise. Most of it has likely been driven by the service sector (not unlike in India in the 1980s and 1990s (Ravallion and Datt, 1996). It also highlights the enormous potential for accelerating poverty reduction if the performance of the agricultural sector can be boosted. Higher world food prices, increased domestic demand for food following solid economic growth and urbanization and increased investment both by the public and private sector should well position Africa for harnessing this opportunity. Ethiopia's and Rwanda's experiences are illustrative. Cross-country evidence from the rest of the world also shows that the sectoral composition of growth matters for poverty reduction. For countries outside the Africa region, industry and services growth exerts a much stronger poverty response than agricultural growth. For both services and industry, an increase of 1 percent in the growth in GDP per capita led by each of these sectors decreases the poverty headcount by almost 2 percent. The gap between the poverty-reducing effects of these two sectors is uneven for poverty gap and poverty gap squared, with industry having larger impacts. The growth elasticity of poverty reduction of the services sector is 40 percent larger than that for industry.
The link between agricultural growth and poverty reduction in the rest of the world is weak for poverty headcount and poverty gap, but significant for squared poverty gap. Overall, the results for other developing countries differ substantially from that for Sub-Saharan Africa, where agricultural growth is more effective in reducing poverty.
One striking result is that industrial sector growth does not have a significant effect on poverty reduction in Africa, while it has larger and significant effects in the rest of the developing world. One key point to bear in mind is that the share of poor labor in industry is low compared to agriculture and services (  income of US$4 a day, but it is still not statistically significant. The above analysis shows that subsectors, like manufacturing, can lift incomes, but they would have to increase dramatically in size to have a substantial impact on lowering poverty in the region. Recent analysis by Rodrik (2014) suggests that African growth in the coming years is in fact much more likely to stem from agriculture or services than from manufacturing. If this pattern of growth materializes, combined with the observed higher growth elasticities of poverty reduction with respect to aggregate GDP growth originating in these sectors, it may provide a window of opportunity for poverty reduction in Africa.

Two case studies
Country-specific poverty analyses can provide detailed insights on the importance of the sectoral composition of growth in reducing poverty. The alignment of findings with those from the crosscountry analysis will vary by country. For example, results from a country study on Ethiopia find support for the importance of agricultural growth in reducing poverty, but not of services. Hill and Tsehaye (2014) examine the poverty-growth links in the context of Ethiopia. Ethiopia has grown rapidly for over a decade, registering annual average per capita growth rates of over 8 percent. The country's growth has been accompanied by a sharp drop in the poverty rate. The study examines zone-level variations in sectoral growth and provision of public goods to explain the reduction in poverty between 1996 and 2011. 9 The aim is to see whether the sectoral composition of growth mattered for poverty reduction in Ethiopia.
They find that agricultural growth is significantly related to the decline in poverty in Ethiopia. Thus, zones with the fastest increase in agricultural production experienced the largest decline in poverty. Growth in agricultural output per capita lowered the average poverty rate, with a 1 percent increase in growth reducing poverty by 0.9 percent. Controlling for other variables raises the growth elasticity of poverty reduction to 2 percent. The impact of agriculture is especially found close to urban centers (50,000), pointing to the role of improved access to markets. By contrast, growth in manufacturing and services has not exerted a statistically significant impact on poverty reduction (figure 29). In Ethiopia, growth in rural and small nonfarm town services is closely related to agricultural growth, with the majority of these businesses (64 percent) established on the back of earnings from agricultural production (Jolliffe et al., 2014). Given this correlation between activities, it is possible that the coefficient on agricultural growth also captures some of the poverty-reducing 9 Panel data for 50 zones between 1996 to 2011, four observations over this 15-year period.

SO WHAT IS THE ROLE FOR STRUCTURAL TRANSFORMATION IN ACCELERATING AFRICA'S POVERTY REDUCTION?
The evidence reviewed so far suggests a strong continuing role for increasing productivity within sectors in accelerating Africa's poverty reduction, especially in the so-called low-productivity sectors (agriculture and services). But would a sole focus on the fundamentals to foster sectoral growth not ignore key features of the historical patterns of development, that is the reallocation of activities and labor from low-to high-productivity sectors as countries develop? The realization that low-and high-productivity sectors often co-exist and the opportunities this offers for fostering growth (and poverty reduction) by moving people from low-to high-productivity sectors (at least in an accounting sense) has long formed the basis for the dual economy models of economic growth (going back to Arthur Lewis in the 1950s). Following concerns about the sustainability of Africa's recent growth path, they have recently also seen a revival over the one sector neoclassical growth models. 10 Which view of the world is taken also matters for policy. While the latter models emphasize incentives to save, accumulate human and physical capital, and innovate (that is, the fundamentals), the former emphasize the removal of barriers to (labor) movement across sectors instead (Rodrik, 2014).
Reallocation of labor out of agriculture over time is consistent with long-term trends in demand behavior, that is, the long-term proportional decline in spending on food as income increases (Engel's Law). Nonetheless, dual economy models have simultaneously also emphasized the need to meet food subsistence needs in the early stages of development, before labor can be released productively (Gollin, Parente, and Rogerson 2002).
Within the spirit of the dual economy models, growth in overall labor productivity can thus be achieved in two ways 11 : 1) by increasing labor productivity within existing economic activities (through technological change (increasing total factor productivity), capital accumulation, or shifts in the terms of trade; and 2) by moving labor from low-to high-productivity sectors (the structural transformation).
Contributions of the latter depend on the labor productivity gap between both sectors, and the speed with which labor is reallocated. These considerations have not been fully captured in the cross-country analysis presented above. So, what is the role of structural transformation in accelerating Africa's poverty reduction?
Sectoral productivity gaps are real, and there is also huge heterogeneity within sectors Estimating labor productivity differences across sectors and the contribution of structural change to overall labor productivity growth is methodologically and empirically challenging. 12 This holds even more in data-constrained environments like Africa, where solid empirical evidence is scarce. Two recent studies shed some new light. Taking agriculture and nonagriculture as the low-and high-productivity sectors, and using data on their value added and employment shares from the International Labour Organization and the United Nations National Account Statistics, Gollin, Lagakos, and Waugh (2014) calculate the labor productivity gap across 72 countries (figure 31). They find that the value added per worker is on average 3.5 times larger in nonagriculture than in agriculture; the median ratio is 3.1.
10 See for example the 6th international conference organized by the Global Development Network, in Accra, 2014, fully devoted to the theme of Structural transformation in Africa in Accra 2014. 11 Growth in labor productivity can be decomposed as: (1) with y denoting overall labor productivity and y i labor productivity in sector i= T,M (traditional and modern sector respectively), π i the share of sector i in the economy and S i , the share of employment. Increasing labor productivity within existing economic activities is captured by the first two components of the equation. The third component captures productivity change by moving labor into higher productivity sectors. 12 Sectoral employment shares are often based on outdated records/censuses and occupational status of the household head. Yet households typically allocate time across different activities making it hard to calculate the sectoral returns. In addition, labor productivity are partial measures of productivity and are also affected by the amount of human and physical capital allocated. Systematic differences in human and physical capital applied across sectors may artificially raise the gap.
When accounting for sectoral differences in human capital and hours worked, the gap declines by about 50 percent to a factor 2.2 (or 1.9 for the median). 14 As expected- While the urban self-employed enjoy the largest labor productivity on average, this is mostly driven by high earnings at the top end of the distribution. This also holds for the larger average returns among the rural self-employed, even though the distribution is less skewed than among the urban self-employed.
Net earnings among the bottom 5 percentile are even negative for both groups. Labor productivity among farmers is less dispersed, though net earnings among the 75 percentile household are still more than 4 times higher than those among the 25 percentile household. Also, the median (50 th percentile) farmer earns amounts per hour worked similar to the median rural self-employed, and more than the median urban self-employed.
Three insights emerge. First, despite substantial labor productivity gaps on average between agriculture and nonagriculture, low-and high-productivity sectors do not nicely coincide with agriculture and nonagriculture respectively. This was recognized early on by Lewis, and is also increasingly recognized within the literature (Gollin 2014). Yet it remains largely ignored within the macro-policy dialogues.
14 Many households or household heads who have agriculture as their primary activity, also work outside agriculture. This leads to an overestimation of the number of agricultural workers and an underestimation of their labor productivity when expressed in terms of persons as opposed to hours.
Source: Gollin, Lagakos, and Waugh, 2014.  Finally, the move from low-productivity jobs in agriculture to high-productivity jobs in the city, which is often implicitly assumed in the analysis of structural transformation, is far from obvious.
These differences in labor productivity ignore the contribution of physical capital. The larger (gross) labor productivity among rural household enterprises observed in Ethiopia, in rural transport enterprises, bars, and restaurants, all which are more capital intensive, does thus not surprise ( figure 32). Yet, it is especially these larger capital requirements (including human capital) that often make the more productive jobs less accessible to the poor (Barrett, Reardon, and Webb 2001). As a result, income diversification among the poor (and structural transformation) is often in the less remunerative nonfarm sector, consistent with the limited decline in poverty reduction despite substantial growth.

African migrants are mainly in search of better public amenities, not better wages
Clearly, obtaining reliable estimates of marginal labor productivity gaps that account for differences in human and physical capital, as well as hours worked, remains challenging in practice. Taking a more indirect approach, the empirical evidence on labor productivity gaps presented above would suggest substantial migratory pressures. Surprisingly, de Brauw and Mueller (2014)  In a novel recent study, Dustmann and Okatenko (2014) examine individuals' intentions to move away from their area of current residence over the next 12 months using data from the Gallup World Poll 2005-6. 16 Overall, 19 percent of respondents in Asia and Latin America and 29 percent in Sub-Saharan 15 De Brauw and Mueller define the rural-urban migration rate as the difference between rural and urban population growth and find it to be 1.07 percent on average (population weighted) for Sub-Saharan Africa during 1990-2000, with few countries experiencing rural-urban migration rates exceeding 2 percent. 16 While this does not represent actual migration, intentions often represent the best point estimates of respondents' future behavior (Manski, 1990).

Rural income diversification holds additional promise, especially for poverty reduction
So, what has been the contribution of structural transformation in Sub-Saharan Africa's recent history?
Decomposing growth in labor productivity across 16 Sub-Saharan African countries during 1995-2009, using the national accounts and sectoral employment data from population censuses, McMillan and Harttgen (2014) find that labor reallocation across sectors accounted on average for about half of overall labor productivity growth (recorded at 21.8 percent) ( figure 33). This is substantial.
17 An index of contentment was constructed using polychoric principal component analysis applied to respondents' opinions on the availability of public services such as health care, schools, and air quality in the residence area.  There was also substantial heterogeneity across countries. The contribution was small, but positive, for example, in higher-income and well-diversified economies such as Mauritius (2000-07), where agriculture and manufacturing were already much smaller and where the highly productive services sector has managed to absorb labor. In resource-rich Nigeria (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009), however, structural change contributed a lot more. Further inspection shows that this largely resulted from big shifts of labor out of agriculture (and to a lesser extent, services) into manufacturing, since the productivity differences between the three sectors were rather small, probably due to the high degree of informality in all three sectors.
This contrasts with the experience of resource-poor Malawi (1998Malawi ( -2005, which displayed little structural change and even a decline in labor productivity within its sectors. Finally, structural change contributed substantially to labor productivity growth in resource-poor, but emerging, Uganda (1999Uganda ( -2009, with the contribution of structural change resulting both from large labor shifts out of agriculture and large initial differences in labor productivity across sectors (see below).
But what about the contribution of structural transformation to poverty reduction? Analyzing this requires micro household panel data. Consistent with the notion of high productivity growth potential outside agriculture, Christiaensen and Kaminski (2014)  Yet, 70 percent of the 4-percentage-point decline in the poverty headcount resulted from raising agricultural incomes among people staying in agriculture. The other one-third came from rural nonfarm diversification, which also contributed disproportionately to consumption growth-it accounted for one-fifth of overall consumption growth, even though the number of households that diversified out of agriculture into rural nonfarm activities made up only 13 percent of the population. Put differently, fostering nonagriculture (both in rural and urban areas) appears disproportionally good for growth, and fostering agricultural productivity appears disproportionally good for poverty reduction. Structural transformation into the rural economy (rural nonfarm income diversification) benefited both poverty and growth.
In addition to re-emphasizing the continuing role of within-sector productivity increases for poverty reduction (agriculture) and growth (nonagriculture), the findings also call attention to the spatial aspects of the structural transformation for poverty reduction. Structural transformation can contribute to poverty reduction, but especially if the non-farm jobs are nearby, in the rural economy. The importance of rural development (especially in agriculture, but also in the rural nonfarm sector) is also borne out by other cross-country micro-econometric evidence (Imai, Gaiha and Garbero, 2014) and is consistent with the insights about people's migratory intentions discussed above (Dustmann and Okatenko, 2014).

INCREASING AGRICULTURAL PRODUCTIVITY IS AN IMPORTANT DRIVER OF STRUCTURAL TRANSFORMATION
Many of the results about the contributions to sectoral growth and structural transformation presented so far hold only in an accounting sense; they abstract from the "knock on effects of sectoral growth in one sector, on structural transformation or the other sector, and do not tell how to bring about the structural transformation. Yet the origins for structural transformation may lie both with short-and longrun policies to foster sectoral growth as well as with the removal of barriers to labor movement. And the latter may also benefit sectoral growth. Increasing land tenure security can for example help release labor for the non-agricultural sector while also fostering investment in land productivity increasing measures such as agro-forestry and erosion control. Given the complex and dynamic nature of these processes, solid evidence on the actual drivers of structural transformation is hard to come by. One crude, initial attempt by Gollin, Lagakos and Waugh (2014) explores how their adjusted measure of the (average) agricultural productivity gap relates to measures of labor mobility (such as restrictions on domestic movement 19 and ethnic fractionalization), measures of institutional quality (Kaufmann, Kraay, and Mastruzzi's rule of law index) and geographic features (such as terrain ruggedness, a proxy for physical productivity in agriculture, 20 the fraction of land categorized as fertile and tropic, and the share of fuels and minerals in merchandise exports).
When looking at each of the factors separately, the results accord with expectations. 21 When looking at all of them simultaneously, only the coefficient on ruggedness is statistically significant and largely unchanged from when looking at the different factors independently. The findings underscore the critical role of agricultural productivity in fostering the structural transformation. In analyzing the factors affecting the share of the labor force in agriculture in Sub-Saharan Africa, McMillan and Harttgen (2014) also find a negative correlation with agricultural productivity, providing further support for the proposition that the road out of agriculture very much runs through it. They also find small declines in the agricultural labor force with rural schooling and with population growth, hinting at the emergence of land scarcity.

WAYS FORWARD
Investments and policies to foster growth in the rural economy (that is, close to where the people work and are) emerge as critical for accelerating poverty reduction in Sub-Saharan Africa and fostering inclusive growth. This will also help move people out of agriculture over time. It is difficult to imagine how large amounts of labor can be productively released with cereal yields still standing at historical lows of 1.4 ton/ha. Even China and Vietnam already enjoyed cereal yields well above 2 ton/ha when their institutional reforms unleashed a boost in agricultural productivity and poverty reduction (Ravallion and Chen, 2007).
Africa's rising urbanization and solid economic growth are also generating substantial domestic demand, both for staples and for more protein-rich foods (meat, dairy) and higher value products (vegetables, fruits). As of now, much of these stand to be met by imports, 22 but could be produced locally, generating important import substitution and employment opportunities (for example, rice in West Africa and poultry around urban centers). Designing institutional arrangements and policies that help poorer smallholders benefit maximally from these opportunities, directly through production as well as indirectly through the labor market, is the key challenge moving forward.
19 Taken from Cingranelli and Richards (2010), who rely on U.S. State Department country human rights reports to categorize a country as having restrictions on internal mobility. 20 The ruggedness index measures the average variability in elevation within a country. It can be seen as a measure of physical productivity in agriculture, with more ruggedness associated with lower productivity (Nunn and Puga, 2012). 21 Countries with restrictions on domestic movement have larger adjusted agricultural productivity gaps, as do countries with more ethnic fractionalization. Similarly, countries that lack rule of law have larger gaps, with a 1 standard deviation decrease in the rule of law associated with a 16 percent higher residual gap. Finally, countries with lower agricultural productivity also have higher gaps, with a 1 standard deviation increase in ruggedness associated with an 18 percent higher gap. A one standard deviation in export dependence on fuel or minerals also increases the gap by 18 percent. 22 The latest agricultural Outlook from OECD-FAO predicts that net food imports for Sub-Saharan Africa will rise dramatically over the next 10 years (by an additional 6.8 million tonnes of wheat to 20.8 million tonnes net import in 2023 and an additional 6.2 million tonnes of rice to 17 million. But Sub-Saharan Africa is also projected to become a net importer of coarse grains (4.8 million tonnes in 2023) (maize/ millet/sorghum) and there are increasing imports of poultry (1.8 million tonnes in 2023) and pork (0.7 million tonnes). With this, Sub-Saharan Africa stands to become the world's major rice importer. Large importers include, for example, Nigeria (5 million tonnes net rice imports by 2023), South Africa (1.7 million tonnes), but also Ghana, Tanzania and Mozambique with between 0.5 and 1 million tonnes each. This is projected to happen despite increasing production (by 35 to 40 percent over the next decade for rice and meat; 28 percent for coarse grains, and 20 percent for wheat).
But boosting agricultural productivity alone will not suffice. Investments in rural public goods (for example, education, health, rural roads, electricity, and ICT) and services (including in small towns) will be equally important to boost the rural economy and facilitate the structural transformation through rural income diversification, while also equipping the next generation for migration to the cities. While Africa's urbanization rate is still relatively low compared to that in other continents, Africa has been urbanizing at a rate similar to that of Asia during 1960-2010. Yet, its urban population has been growing much faster (that is, the rate at which its urban population expands), exceeding Asia's urban growth by 1.5 to 2 percentage points, due to much higher urban fertility. To put this in perspective, when the population in urban centers grows at 3. Finally, while manufacturing may not provide a panacea, Sub-Saharan Africa should also expand its manufacturing base, especially by boosting its fundamentals (business climate, macroeconomic stability, lower transport cost, cheaper and more reliable power, and a more educated labor force ), which will benefit all sectors. When support is targeted, it should not crowd out investments in the building blocks for inclusive growth.