POVERTY & EQUITY NOTES M AY 2020 · NUMBER 20 How much will poverty rise in Sub-Saharan Africa in 2020? Jose Montes, Ani Silwal, David Newhouse, Frances Chen, Rachel Swindle, and Siwei Tian 1 The ongoing coronavirus pandemic is expected to drastically slow 2020 GDP per capita growth in Sub-Saharan Africa (SSA) by about 5 percentage points compared to pre-pandemic forecasts. This note presents results from an analysis of a comprehensive database of surveys from 45 of 48 SSA countries to examine the effects of the project fall in growth on poverty in the region. An additional 26 million people in SSA, and as much as 58 million, may fall into extreme poverty defined by the international poverty line of US$1.90 per day in 2011 PPP. The poverty rate for SSA will likely increase more than two percentage points, setting back poverty reduction in the region by about 5 years. The coronavirus global pandemic (COVID-19) has caused a sudden and sharp fall in global Methodology economic activity. The relatively low number of infections officially reported from SSA may be This note uses a collection of household surveys misleading because testing capacity is limited and (SSAPOV) 3 and GDP projections from the World the region may still be in the early stages of the Bank and the IMF’s World Economic Outlook pandemic. 2 A rapidly-expanding pandemic in (WEO) to examine the potential impact of the dense urban centers could easily overwhelm weak pandemic on poverty in SSA. The methodology health delivery systems. This will be compounded we use resembles that in Mahler et al. (2020). 4 The by a slowdown in global economic activity and the shock to GDP is defined as the difference in the additional effects of lockdowns imposed by many 2020 GDP growth rate between the most recent governments. World Bank projections and the Fall 2020 vintage of the WEO. This captures the projected growth Although still unfolding, we can begin to assess rates before and after the knowledge of the the effect of the pandemic on poverty in SSA. pandemic and corresponding economic This note estimates of the increase in poverty in contraction. SSA, given current knowledge. It also presents a profile of those vulnerable to falling into poverty as GDP per capita is expected to contract a result of the pandemic and poverty outcomes throughout SSA in 2020 (Figures 1a and 1b). Prior under selected economic scenarios. This to the pandemic, SSA’s GDP per capita was assessment may need to be revised regularly to forecasted to grow at 1.7 percent. As a result of the reflect the latest developments on the ground as pandemic and other associated shocks, the GDP the pandemic develops. per capita growth forecast is now expected to be 1 This note was written by the SSA Team for Statistical Development in the World Bank’s Poverty and Equity Global Practice. 2 As of April 28, 2020, only 1% of total official the total COVID 3 million cases globally were reported to be in Africa (Source: ft.com). 3 SSAPOV refers to a collection of harmonized nationally representative household surveys for Sub-Saharan Africa. 4 Mahler, D.G., C. Lakner, R.A. Castaneda Aguilar, and H. Wu. 2020. “The impact of COVID-19 (Coronavirus) on global poverty: Why Sub-Saharan Africa might be the region hardest hit.” World Bank Blog. April 20. https://blogs.worldbank.org/opendata/impact-covid- 19-coronavirus-global-poverty-why-sub-saharan-africa-might-be-region-hardest more than 5-7 percentage points lower, represents a rough estimate of the effect of the contracting 3.1 percent in the baseline scenario and crisis on poverty. 5.5 percent in the low scenario. The GDP per capita growth projection in the baseline scenario is lower This methodology makes strong assumptions, by more than 10 percentage points in 7 countries yet provides a useful rough estimate of poverty in the region: Zimbabwe, Botswana, Sao Tome and effects. The first strong assumption is that GDP Principe, Mauritius, and Seychelles (Figure 1b). growth is fully reflected in consumption; in other words, we ignore saving and borrowing. 7 The Figure 1a: Projected SSA negative GDP per capita second assumption is that the distributional effects growth, 2020 will be neutral. In reality, certain countries and 5 sectors are more vulnerable to economic 4 downturns and public health crises. For example, informal sector workers are likely to be laid off and Annual GDP growth per capita (%) 3 2 have less access to government programs. At the 1 same time, effects might be stronger in urban than 0 rural areas if social distancing is more economically -1 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 disruptive in cities. Further, developing areas like -2 SSA enter economic downturns with fewer -3 resources to weather the socioeconomic -4 Pre-pandemic With pandemic challenges accompanying a sudden contraction. -5 This methodology also cannot distinguish between Source: WEO and World Bank forecasts different causes of the 2020 downturn: while the main factor is likely to be the pandemic, an This sharp slowdown in growth will increase independent drop in oil prices also contributed to poverty. To assess poverty effects, we “line up” slowing growth. welfare, as proxied by per capita consumption, of the survey year welfare to 2020 under each projection. Survey data are available for 45 out of Figure 1b: Projected reduction in SSA GDP per 48 SSA countries. 5 The lining up procedure is the capita growth in 2020 same one used by the World Bank. 6 It is “distributionally neutral” because household welfare is adjusted in the same way for every household, by multiplying welfare by the growth rate in GDP per capita since the survey. In other words, we compute 2020 poverty rates for each country by scaling up welfare in the observed survey for each household according to growth in per capita GDP in each period. This was done both for the Fall 2019 vintage of the WEO and the most recent World Bank projections. The difference in lined-up poverty rates between the two vintages Sources: SSAPOV data, WEO and World Bank forecasts 5 The database includes the most recent household survey 6 The poverty rates for 2018 match the official World Bank used for measuring poverty in each country (dating from 2008 poverty rates published in the PovcalNet website. to 2018) and consists of responses from 2.4 million 7 Here we deviate slightly from Mahler et al. (2020) who, observations. Countries excluded from the database are based on historical data, assume that 85% of growth is passed Eritrea, Equatorial Guinea, and Somalia which do not have through to consumption. recent official surveys for measuring poverty. May 2020 · Number 20 2 If welfare changes are distributionally neutral, Implications for poverty in SSA in 2020 people self-employed outside of agriculture and living in urban areas would be disproportionately affected (Table 1). Even if the The 2020 downturn will likely increase the SSA fall in welfare were uniform across the population, poverty rate by at least 2 percentage points the impact on poverty would be different for (Figure 2). This would mean that an additional 26 different socio-demographic groups, because of million people in SSA would fall into extreme existing disparities in living standards. The poverty (defined as those living under US$1.90 per following table presents a profile of the existing day in 2011 PPP). This projected increase in the poor who would have been poor under the poverty rate would return SSA to 2015 poverty baseline forecast. The new poor are the people who levels, effectively wiping out 5 years of progress may fall into poverty as a result of the pandemic reducing impoverishment. Figure 3 presents the under the World Bank’s most recent growth increase in poverty rates for all countries in the forecasts. The non-poor represents the group region. Half of the new poor will live in five whose welfare will be greater than US$1.90 per day countries: Democratic Republic of Congo, Ethiopia, even under the latest World Bank forecasts. Kenya, Nigeria, and South Africa – with Nigeria contributing the most with 6.9 million new poor. Table 1: Profile of new poor in SSA, 2020 Sao Tome and Principe, Zimbabwe, Niger, Republic (poverty rate%) of Congo, Sierra Leone, and Botswana are expected How to read: e.g, 30.1% of new poor live in urban areas. to see the largest poverty rate increase. Existing Non- Socio-demographic group New poor poor poor Figure 2: Poverty rate for SSA forecast Lives in urban area 30.1 19.6 45.5 Self-employed outside Poverty rate at $1.90 line (%) 46.6 45.2 43.9 43.7 agriculture 29.8 21.0 29.4 Works in service/ sales 11.5 6.7 17.0 44.1 43.1 . 42.6 Is self-employed in 42.3 42.2 41.6 40.2 agriculture 53.7 62.7 42.6 39.4 2010 2011 2012 2013 2015 Year 2018 2020 2021 Some primary ed 63.3 57.7 70.8 Pre-Covid Post-Covid (baseline) Post-Covid (low) Child (younger than 18) 51.5 55.8 43.6 Source: SSAPOV/GMD database, PovcalNet, World Bank/IMF GDP forecasts Female 50.8 50.8 50.7 Sources: SSAPOV data, WEO and World Bank forecasts Lives in female-headed household 19.5 17.1 22.0 Figure 3: Forecast of SSA countries’ increase in poverty rates, 2020 (percentage points) The people who will fall into poverty as a result of the pandemic are very different than those who are already poor. They are more likely to be living in urban areas, have at least some primary education, be self-employed outside of agriculture, work in service or sales occupations, and live in a female-headed household. More than half the new poor are still expected to be children under 18 years. Sources: SSAPOV data, WEO and World Bank forecasts May 2020 · Number 20 3 mobility restrictions. The SSAPOV data can shed Scenario Analysis light on the effects of these sectoral shocks on poverty for 30 out of the 45 SSA countries, representing 77 percent of the 2020 population. The projected decline in GDP per capita growth could result in up to 58 million additional poor If incomes in the service sector were to drop by in SSA in 2020. Since the global pandemic is still 50 percent for 3 months in this sector, an unfolding and its full extent is yet unknown, we additional 18.4 million individuals would fall present several scenarios of reduction in household into poverty in the 30 countries (Table 3). welfare and increase in regional poverty. Compared We examine two sectoral scenarios: (a) a 50 percent to the baseline based on 2019 IMF forecasts, a 5, 7, drop in income for 3 months and (b) a 50 percent and 10 percent uniform drop in welfare would drop in income for 6 months. We adjust this shock result in additional 28, 40, and 58 million poor, to each household by the fraction of prime respectively. A larger shock in urban welfare working-age members (15-55 years) who work in a compared to rural welfare would also increase in given sector. 9 Nearly 1 of 4 prime-age employed the number of poor. For example, if the shock in adults works in the services sector. The last two urban areas was 10 percent and rural areas was 5 rows present scenario results where workers in percent, 6 million more people would fall into more than 1 sector lose income. poverty than in the scenario with a uniform 5 percent shock (36 vs. 28 million). 8 Table 3: Labor market scenarios of the impact of COVID-19 on poverty in SSA in 2020 Table 2: Growth scenarios for 2020 and SSA 50% drop in incomes 50% drop in incomes poverty effects Share for 3 months for 6 months prime- Poverty Add. age # poor Scenario rate for poor workers (mil) Additional Change in Additional Change SSA (pct) (mil) (%) Sector poor (mil) Gini poor (mil) in Gini Baseline (Oct-2019 WEO 40.2 448 forecasts) Services 23.4 18.4 -0.4 24.1 -0.8 5pct welfare drop 42.7 476 28 Industry 6.3 7.9 0.1 9.2 0.1 7pct welfare drop 43.7 488 40 Self- 10pct welfare drop 45.4 506 58 employed 15.9 14.8 0.1 18.9 0.2 Urban 10pct, rural 5pct 43.0 479 31 Services & welfare drop industry 29.8 20.2 -0.4 27.6 -0.9 Urban15 pct, rural 7pct Services, 44.2 493 45 welfare drop industry, & Note: The poverty line used to compute the poverty rate is self- $1.90 per day in 2011 PPP. employed 30.6 20.4 -0.4 27.9 -0.8 Note: The baseline scenario WEO October-2019 forecasts. Source: Authors’ analysis using SSAPOV database, World Bank, The pandemic has particularly hurt sectors such and IMF growth projections. as services and manufacturing where workers are in close proximity with other people. The incomes of self-employed workers outside of agriculture is also likely to fall because of the general slowdown in economic activity and 8 Other scenarios not examined here include a weaker shock 9 It would be more accurate to examine working hours and to GDP in SSA, a smaller “pass-through” from GDP growth to wages of all workers in the household. However, data on these household welfare, and differential impact on socio- variables are available for many fewer countries than data on demographic groups. sector of employment. May 2020 · Number 20 4 Table 4: 2020 forecast GDP per capita growth and poverty rates pre-COVID and post-COVID Pre-COVID Post-COVID Population in GDP per capita growth, GDP per capita growth, Poverty rate, Country Survey year Poverty rate, pct1 2020 (millions) pct (WEO Oct 2019) pct (World Bank) percent1 Angola 2018 32.9 -1.8 49.9 -7.7 53.5 Benin 2015 12.1 3.8 41.1 0.3 42.9 Botswana 2015 2.4 2.4 13.4 -10.4 18.2 Burkina Faso 2014 20.9 3.0 29.8 -0.9 33.1 Burundi 2013 11.9 -2.4 80.2 -1.8 79.4 Cabo Verde 2015 0.6 3.7 1.7 -6.6 2.6 Cameroon 2014 26.5 1.6 20.1 -2.6 21.8 Central African Republic 2008 4.8 2.9 69.7 -1.0 71.5 Chad 2011 16.4 2.9 40.0 -3.1 43.2 Comoros 2013 0.9 1.5 17.0 -4.0 18.1 Congo, Dem. Rep. 2012 89.6 0.9 69.8 -5.0 72.6 Congo, Rep. 2011 5.5 0.3 39.1 -8.6 45.1 Côte d'Ivoire 2015 26.4 4.6 17.6 0.1 19.6 Eswatini 2016 1.2 -0.6 27.4 -3.8 29.3 Ethiopia 2015 115.0 5.5 17.4 1.6 18.8 Gabon 2017 2.2 2.1 3.3 -4.5 3.6 Gambia, The 2015 2.4 3.3 7.0 -0.5 8.2 Ghana 2016 31.1 3.5 10.5 -0.5 11.5 Guinea 2012 13.1 3.4 19.2 -0.4 20.7 Guinea-Bissau 2010 2.0 2.6 59.1 -3.7 62.5 Kenya 2015 53.8 3.2 29.3 -1.2 31.7 Lesotho 2017 2.1 -0.8 26.5 -5.7 29.6 Liberia 2016 5.1 -1.0 43.9 -5.0 48.7 Madagascar 2012 27.7 2.5 74.0 -3.0 76.4 Malawi 2016 19.1 2.2 67.2 -0.8 68.9 Mali 2009 20.3 1.9 40.0 -2.0 42.7 Mauritania 2014 4.7 3.6 5.2 -4.1 6.5 Mauritius 2017 1.3 3.8 0.1 -6.8 0.2 Mozambique 2014 31.3 3.3 59.9 -1.4 62.0 Namibia 2015 2.5 -0.3 15.4 -6.6 17.2 Niger 2014 24.2 2.1 38.2 -2.7 42.1 Nigeria 2018 206.1 -0.1 51.0 -6.1 54.3 Rwanda 2016 13.0 5.6 44.7 0.4 46.7 Senegal 2011 16.7 3.8 24.5 -0.1 26.8 Seychelles 2013 0.1 2.5 0.7 -11.8 1.0 Sierra Leone 2018 8.0 2.4 36.6 -4.3 41.4 South Africa 2014 59.3 -0.4 19.9 -8.0 22.4 South Sudan 2015 11.2 5.0 82.0 1.7 82.0 Sudan 2014 43.8 -4.0 16.1 -6.4 17.2 São Tomé and Principe 2017 0.2 1.3 33.4 -11.7 41.3 Tanzania 2018 59.7 2.6 46.1 0.0 47.0 Togo 2015 8.3 2.7 43.0 -1.5 45.3 Uganda 2016 45.7 2.6 36.8 -0.2 39.1 Zambia 2015 18.4 -1.3 57.7 -3.7 58.5 Zimbabwe 2017 14.9 0.8 36.0 -11.7 43.8 1 The poverty line used for computing the poverty rate is $1.90 per capita per day in 2011 PPP. This note series is intended to summarize good ABOUT THE AUTHORS practices and key policy findings on Poverty-related Jose Montes is a Data Scientist in the World Bank’s Poverty and Equity Global Practice. Jmontes@worldbank.org. topics. The views expressed in the notes are those of David Newhouse is Senior Economist in the World Bank’s Poverty and Equity GP. dnewhouse@worldbank.org. the authors and do not necessarily reflect those of the World Bank, its board or its member countries. Copies Frances Chen, Ani Silwal, Rachel Swindle, and Siwei Tian are consultants in the World Bank’s Poverty and of these notes series are available on Equity GP. www.worldbank.org/poverty May 2020 · Number 20 5