Journal of African Economies, 2023, 32, ii164–ii182
https://doi.org/10.1093/jae/ejac047
Supplement Paper
Progress on Poverty in Africa: The
Importance of Growth and Inequality
Augustin Kwasi Fosu*a,b,c
aInstitute of Statistical, Social, and Economic Research (ISSER), University of Ghana, Legon, Ghana
bCollege of Business and Economics, University of Johannesburg, Johannesburg, South Africa
cFaculty of Economic and Management Sciences, University of Pretoria, Pretoria, South Africa
*Corresponding author: Augustin Kwasi Fosu. E-mail: afosu@ug.edu.gh; afosu@isser.edu.gh
Abstract
Employing World Bank data, this paper, first, historically examines Africa’s record on poverty incidence,
spread and severity, as compared with other regions of the world, at the US$1.90 and US$3.20 per day
(2011 PPP) poverty standards. Second, it evaluates country-specific progress on growth, poverty and
inequality, and compares the ‘poverty transformation efficiency vector’ (PTEV) among African countries.
Third, the study analyses the relative roles of income growth and inequality changes in explaining
African countries’ poverty records, through a decomposition of poverty changes using ‘optimal’ income
and inequality elasticity estimates from the ‘identity’ model. The study finds that following the dismal
record on poverty during the 1980s, progress on poverty has been appreciable since Africa’s growth
resurgence starting in the mid-1990s, and that this progress was driven mainly by income growth,
consistent with the global evidence. Nonetheless, inequality often played a complementary role in
most of the countries and, in a small number of cases, it was the primary driver of changes in poverty.
Thus, the present study sheds light on country-specific differences in the relative roles of growth and
inequality in poverty reduction on the continent, based on both qualitative and quantitative evidence.
The study should, therefore, provide a useful compass to those who seek to understand country-
specific situations within the African context.
Keywords: growth, income, inequality, poverty, Africa, poverty transformation efficiency vector (PTEV)
JEL classification: D31, I32, O11, O49, O55
1. Introduction
Reducing poverty has been at the heart of global public policy; it was indeed theMillennium
Development Goal 1 (MDG1) and is currently designated as the Social Development Indi-
cators Goal 1 (SDG1). In general, African countries have registered appreciable reductions
in poverty since the mid-1990s, contrary to the ‘lost decade’ of the 1980s. This progress
coincides with the region’s growth resurgence.
The extant literature suggests that countries’ initial characteristics, growth and redistribu-
tive policies are the underlining factors behind poverty reduction. Focusing on sub-Saharan
Africa (SSA), Fosu (2015) argues that Africa’s1 relatively recent progress on poverty may
1 I employ ‘SSA’ and ‘Africa’ interchangeably herein.
© The Author(s) 2023. Published by Oxford University Press on behalf of the Centre for the Study of African
Economies.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction
in any medium, provided the original work is properly cited.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii165
be attributable primarily to its growth resurgence. However, that study also emphasises
the importance of inequality. Thus, it would be desirable to identify those countries where
inequality and/or growth constitute significant constraints on poverty reduction.
This paper attempts to answer the following research questions:
• How has Africa generally performed historically on poverty reduction, absolutely and
relatively to other regions of the world?
• What is the distribution of African countries on poverty-reduction performance, espe-
cially since the early-mid-1990s when Africa has generally experienced growth resur-
gence?
• How has poverty responded to African economic growth generally and how does this
responsiveness differ across countries?
• Specifically, what role has inequality played, relative to income growth, in the poverty-
reduction process, generally and by country?
• For each country, how might the results provide a compass for crafting appropriate
policies to most effectively reduce poverty?
To effectively address the above research questions, this paper proceeds as follows. First, a
brief review of the literature is presented. Second, poverty trends for Africa, comparatively
with other global regions, are provided. Third, the study sheds light on the progress on
poverty for African countries, depending on data availability, separately and comparatively
among them, since about the mid-1990s when Africa has experienced growth resurgence
generally.
Fourth, the paper presents qualitative evidence on the growth-poverty-inequality nexus
in African countries, using the ‘poverty transformation efficiency vector’ first developed
in Fosu (2017a) and applied to a global sample. This presumably innovative approach
should help us discern qualitatively the cross-country relationships between growth and
changes in poverty, with inequality serving as an intermediating factor. Specifically, the
methodology is expected to provide a better understanding of countries’ performance in
transforming per capita GDP growth into income growth at the household level, and then
from income growth into poverty reduction, conditional on changes in income distribution.
Also reflected, therefore, is the importance of the relationship between per capita GDP
growth and income growth, in order to elicit the extent to which the former may be
translated into poverty reduction, for a given income distribution.
Fifth, the poverty-growth-inequality ‘identity’ model is presented and estimated, in order
to provide a quantitative relationship between the progress on poverty, on the one hand,
and income growth and the changes in inequality, on the other. Based on the estimates,
income and inequality elasticities are generated for each country, and are then employed
to decompose changes in poverty into the respective contributions by income growth and
changes in inequality. The quantitative results should complement the qualitative ones in
providing guidance for country-specific policy with respect to the relevant levers for efficient
poverty reduction: income growth versus income distribution.
2. Brief literature review and theoretical considerations
Traditionally, neoclassical economics has considered Income growth as the primary route for
poverty reduction (Deininger and Squire, 1998; Dollar and Kraay, 2002; Dollar et al., 2016).
The importance of income distribution for poverty reduction has also received increasing
attention, that is, in addition to such factors as the level of development and other initial
characteristics (see Ravallion, 2012; Dabla-Norris et al., 2015). In this vein, there has been
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii166 Augustin Kwasi Fosu
a special focus on African countries more recently (Fosu, 2008, 2009, 2010a, 2010b, 2010c,
2015, 2018; Thorbecke, 2013).
Income inequality may influence poverty reduction via two main channels: the ‘growth
channel’ (Alesina and Rodrik, 1994; Ravallion, 1997) and the ‘growth elasticity’ channel
(Ravallion, 1997; Easterly, 2000; Adams, 2004). The ‘growth channel’ is rather contro-
versial. On the one hand, a school of thought, consistent with the Kuznets’ hypothesis, is
that high levels of inequality would be associated with greater economic growth, at least
at the early stages of development (Kuznets, 1955). On the other hand, others stipulate
that income inequality hinders economic growth by fuelling social discontent or weakening
existing institutions (see, Thorbecke and Charumilind, 2002).
I focus here on the ‘growth elasticity channel’, that is, given its level, how is economic
growth translated into poverty reduction? Many studies, both cross-country and country-
specific, find that a high level of inequality tends to result in low growth elasticity of poverty,
thus limiting the extent to which income growth is transformed into poverty reduction
(Bourguignon, 2003; Kalwij and Verschoor, 2007). Similarly, for African countries, the
translation of income into poverty reduction tends to be hampered by higher levels of
inequality (Fosu, 2008, 2010a, 2010b, 2010c). Furthermore, based on the ‘identity’ model
of Bourguignon (2003) and Epaulard (2003), Fosu (2015, 2017a, 2017b, 2018) find that
both income and inequality elasticities of poverty decline with initial inequality. Hence,
for purposes of poverty reduction, not only might higher levels of inequality hamper the
effectiveness of income growth, but they could also attenuate the ability of improvements
in income distribution to reduce poverty.
The direct effect of a redistribution of resources from the rich to the poor segment of
society should, therefore, reduce poverty, even with negligible income growth (Bourguignon,
2003). In very poor countries, however, a reduction in inequality could actually perversely
lead to a rise in poverty (Fosu, 2008, 2010a, 2010b, 2010c, 2009, 2015, 2017a, 2017b).
Nonetheless, these and other studies find that on average, attenuating inequality would
help to lower poverty directly, or indirectly via increasing the poverty elasticity with respect
to growth.
In sum, Fosu (2015, p. 56) writes: ‘On average, income growth has constituted the main
engine for poverty reduction in SSA. In several African countries, however, inequality has
been crucial. This role has had two parts: (1) declining inequality tended to decrease poverty
(though not necessarily in very low-income countries), and (2) lower initial inequality raised
the rate at which growth was transformed into poverty reduction. The former role may be
impeded, however, by low incomes as well as by high levels of initial inequality. High levels
of inequality have indeed constrained poverty reduction in many African countries. Thus,
it appears that according greater attention to reducing inequality constitutes a desirable
objective, especially in certain African countries.’
Nor is inequality the only variable that intermediates the relationships between income
growth and changes in income, on the one hand, and poverty reduction, on the other.
Indeed, based on analysis-of-covariance models (Fosu, 2010a, 2010b, 2010c), a lower level
of income is found to decrease the rate at which greater income distribution reduces poverty,
while a higher level of inequality tends to attenuate the ability of income increases to reduce
poverty. Similarly, on the basis of the ‘identity’ poverty equation (Fosu, 2009, 2015, 2017a,
2017b, 2018), it is observed, similarly to high initial inequality, that a low-income level, as
apparent in many African countries, would likely decrease the poverty-reduction potency
of both growth and improvements in income distribution.
Yet, growth itself may raise the level of income and, hence, the respective abilities of
income increases and/or decreases in inequality to further reduce poverty. In effect, countries
that succeed in initially improving both income and income distribution should have an
easier time reducing poverty even further.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii167
3. Progress on poverty
3.1. Progress on poverty: regional trends
Using poverty data from the World Bank PovcalNet, Appendix Figures A1-A6 present the
1981–2018 trends in the progress on poverty across the developing world generally and for
the sub-regions: East Asia and Pacific (EAP), Middle East and North Africa (MENA), Latin
America and Caribbean (LAC), South Asia (SA) and sub-Saharan Africa (SSA). Employed
are the three Foster-Green-Thorbecke (FGT) measures: the headcount ratio, poverty gap
(spread) and squared poverty gap (severity) at the US$1.90 per-day and US$3.20 per-day
poverty standards.2
These appendix graphs show considerable declines in poverty for the developing world as
a whole. The progress is observed for all the three FGT measures. Furthermore, the regional
disaggregation reveals, among the initially high poverty regions, that while EAP and SA have
enjoyed substantial and consistent declines in all the FGT measures since 1981, poverty
increased in SSA during the 1980s and until the mid-1990s, corresponding to the period
of dismal economic growth performance for the region. However, SSA’s growth resurgence
since then has been accompanied by appreciable poverty reduction.
Nonetheless, despite the apparent decline in poverty in SSA during the latter period, the
pace has been slow compared to other regions, leading to a widening gap between the region
and the rest of the developing world (DW). Indeed, for the $1.90 standard, the SSA/DW
poverty incidence ratio increased from 1.47 in 1993 to 3.63 in 2015. Similar observations
are made for the $3.20 standard, and for the poverty gap and squared poverty gap as well.
3.2. Progress on poverty and inequality: trends among African countries
Beyond the average global and regional pictures, poverty progress across African countries
is quite heterogeneous. Tables 1A and B provide evidence for a sample of African countries
with the available data. Presented are (logarithmic) changes in per capita GDP, inequality
and the three FGT poverty measures—poverty incidence or headcount (P0), poverty gap
(P1) and squared poverty gap (P2)—from the early-mid-1990s to the latest year for
which data were available (2000s). The GDP data come from the World Bank’s online
World Development Indicators (WDI), while the rest are derived from the World Bank
PovcalNet. Because the poverty and related data are based on household surveys that are
generally conducted at different country years, the statistics in the tables are annualised for
comparability across countries.
3.3. The poverty transformation efficiency vector
For better comparability, and following Fosu (2017a), Tables 1A and B are translated
into quintile ranks in order to provide a more compact qualitative representation of the
growth-inequality-poverty relationship. The resulting vector, the poverty transformation
efficiency vector (PTEV), is presented in Tables 2A and B for the US$1.90 and US$3.20,
respectively, for each country: with the first five coordinates representing the quintile ranks
for per capita GDP growth, income growth, as well as changes in poverty incidence, poverty
gap, and squared poverty gap, respectively, where the highest rank 5 represents the worst
performance while rank 1 is assigned to the best performers. The sixth coordinate following
the semi-colon is the quintile rank for inequality.
With a vector of (3, 5, 5, 5, 4; 4) from Table 2A, Zambia for instance is among the
worst performers on nearly all the variables, except on per capita GDP growth, on which
it performed moderately well. Hence, Zambia’s outturn on per capita GDP growth was not
transformed into progress on poverty, regardless of the FGT measure used, mainly because
it considerably overstated the country’s actual income growth at the household level. A
2 These standards are based on 2011 PPP international dollars.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii168 Augustin Kwasi Fosu
Table 1A. Annualised Growths (%) of Poverty Measures (Poverty line: US$1.90 a day in 2011 PPP), Per Capita
GDP, Income and Inequality (Gini index), Early-mid-1990s—‘present’∗
Country Period Sub- Per Income Head- Poverty Squared Inequal-
region capita growth count gap Pov. ity
GDP poverty (P1) gap (Gini)
growth (P0) growth (P2) growth
growth growth
Algeria 1995–2011 NA 2.10 1.61 −15.63 −12.32 −7.50 −1.52
Botswana 1993–2016 SA 2.58 1.28 −3.69 −5.23 −6.18 −0.58
Burkina Faso 1994–2014 WA 2.85 3.65 −3.25 −7.43 −10.59 −1.56
Burundi 1992–2014 EA −1.30 1.11 −0.55 −0.88 −1.07 0.69
Cameroon 1996–2014 CA 1.51 3.08 −3.77 −3.89 −3.81 0.27
CAR 1992–2008 CA −0.33 4.68 −1.57 −3.73 −5.20 −0.56
Cote d’Ivoire 1993–2015 WA 0.31 −0.44 1.02 1.33 1.62 0.23
Egypt 1991–2018 NA 2.27 0.47 −3.01 −2.57 −2.32 −0.05
Eswatini 1995–2016 SA 2.26 6.26 −4.85 −7.80 −9.89 −0.48
Ethiopia 1995–2016 EA 5.02 2.12 −3.93 −5.80 −6.83 −1.19
Gambia 1998–2015 WA −0.14 5.61 −11.14 −16.03 −19.51 −1.74
Ghana 1992–2017 WA 2.89 3.86 −5.33 −5.30 −4.99 0.50
Guinea 1991–2012 WA 1.31 6.18 −4.45 −8.42 −11.23 −1.56
Guinea-Bissau 1993–2010 WA −0.48 0.59 0.16 0.41 0.60 0.88
Kenya 1992–2016 EA 0.94 −2.14 0.67 −0.06 −0.64 −1.45
Lesotho 1994–2017 SA 2.86 0.29 −2.64 −4.83 −6.44 −1.51
Madagascar 1993–2012 EA −0.27 −1.27 0.54 1.23 1.59 −0.32
Malawi 1998–2016 EA 1.40 −3.32 0.56 0.95 1.20 −2.09
Mali 1994–2010 WA 2.13 4.17 −3.32 −7.55 −10.66 −2.66
Mauritania 1993–2014 WA 0.52 2.16 −9.19 −10.78 −11.92 −2.04
Morocco 1991–2014 NA 2.58 1.23 −4.55 −3.64 −2.56 0.04
Mozambique 1996–2014 SA 5.17 3.28 −1.40 −3.00 −4.06 0.05
Niger 1993–2014 WA 0.29 2.41 −2.59 −4.36 −5.61 −0.24
Nigeria 1992–2019 WA 1.78 0.95 −1.53 −3.14 −4.47 −0.92
Senegal 1991–2011 WA 0.81 1.97 −2.86 −5.10 −6.70 −1.47
South Africa 1993–2015 SA 1.34 2.12 −2.39 −2.59 −2.21 0.28
Tanzania 1992–2018 EA 2.39 1.89 −1.47 −2.55 −3.40 0.53
Tunisia 1990–2015 NA 2.81 2.38 −14.79 −17.70 −18.35 −0.80
Uganda 1992–2017 EA 3.18 1.75 −1.77 −2.78 −3.58 0.13
Zambia 1991–2015 SA 1.90 −0.12 0.26 −0.61 −1.41 −0.24
Mean 1.69 1.93 −3.55 −4.81 −5.54 −0.65
Median 1.84 1.93 −2.75 −3.81 −4.73 −0.52
Max 5.17 6.26 1.02 1.33 1.62 0.88
(Mozam- (Eswa- (Cote (Cote (Cote (Guinea-
bique) tini) d’Ivoire) d’Ivoire) d’Ivoire) Bissau)
Min −1.30 −3.32 −15.63 −17.70 −19.51 −2.66
(Burundi) (Malawi) (Alge- (Tunisia) (Gam- (Mali)
ria) bia)
Notes: The annualised growth rates are obtained by taking the (logarithmic) difference in the latest-year and
the beginning-year values, dividing by the number of years between the periods, and multiplying by 100. The
data on per capita GDP are obtained from the World Bank Development Indicators (WDI), World Bank
(2020a). The data on income, headcount ratio, poverty gap, squared poverty gap and Gini index are obtained
from Povca∗lNet, World Bank (2020b). For all the poverty measures, the poverty line is US$1.90 per day in2011 PPP. The latest year for which data were available at the time of the study.
similar observation is made of Lesotho’s (1,5,3,3,2;2) PTEV, which shows that per capita
GDP growth grossly overstated income growth, thus leading generally to moderate progress
on poverty, despite the relatively stellar performance on inequality.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii169
Table 1B. Annualised Growths (%) of Poverty Measures (Poverty line: US$3.20 a day in 2011 PPP), Per Capita
GDP, Income and Inequality (Gini index), Early-mid-1990s—‘present’
Country Period Sub- Per Income Head- Poverty Squared Inequal-
region capita growth count gap Pov. ity
GDP poverty (P1) gap (Gini)
growth (P0) growth (P2) growth
growth growth
Algeria 1995–2011 NA 2.10 1.61 −11.36 −13.37 −13.35 −1.52
Botswana 1993–2016 SA 2.58 1.28 −1.77 −3.06 −4.01 −0.58
Burkina Faso 1994–2014 WA 2.85 3.65 −0.97 −3.54 −5.57 −1.56
Burundi 1992–2014 EA −1.30 1.11 −0.27 −0.57 −0.75 0.69
Cameroon 1996–2014 CA 1.51 3.08 −2.63 −3.36 −3.63 0.27
CAR 1992–2008 CA −0.33 4.68 −0.68 −2.21 −3.33 −0.56
Cote d’Ivoire 1993–2015 WA 0.31 −0.44 0.63 0.92 1.12 0.23
Egypt 1991–2018 NA 2.27 0.47 −1.24 −1.90 −2.13 −0.05
Eswatini 1995–2016 SA 2.26 6.26 −2.67 −5.00 −6.68 −0.48
Ethiopia 1995–2016 EA 5.02 2.12 −1.21 −3.26 −4.49 −1.19
Gambia 1998–2015 WA −0.14 5.61 −4.72 −9.20 −12.35 −1.74
Ghana 1992–2017 WA 2.89 3.86 −3.82 −4.64 −4.95 0.50
Guinea 1991–2012 WA 1.31 6.18 −1.56 −4.60 −6.81 −1.56
Guinea-Bissau 1993–2010 WA −0.48 0.59 −0.12 0.15 0.32 0.88
Kenya 1992–2016 EA 0.94 −2.14 0.85 0.59 0.28 −1.45
Lesotho 1994–2017 SA 2.86 0.29 −1.10 −2.73 −3.97 −1.51
Madagascar 1993–2012 EA −0.27 −1.27 0.13 0.63 0.98 −0.32
Malawi 1998–2016 EA 1.40 −3.32 0.28 0.58 0.78 −2.09
Mali 1994–2010 WA 2.13 4.17 −1.00 −3.87 −6.03 −2.66
Mauritania 1993–2014 WA 0.52 2.16 −5.06 −7.38 −8.87 −2.04
Morocco 1991–2014 NA 2.58 1.23 −4.04 −4.43 −4.42 0.04
Mozambique 1996–2014 SA 5.17 3.28 −0.57 −1.75 −2.57 0.05
Niger 1993–2014 WA 0.29 2.41 −0.93 −2.41 −3.40 −0.24
Nigeria 1992–2019 WA 1.78 0.95 −0.40 −1.51 −2.41 −0.92
Senegal 1991–2011 WA 0.81 1.97 −1.06 −2.85 −4.13 −1.47
South Africa 1993–2015 SA 1.34 2.12 −1.33 −1.98 −2.22 0.28
Tanzania 1992–2018 EA 2.39 1.89 −0.67 −1.42 −2.01 0.53
Tunisia 1990–2015 NA 2.81 2.38 −9.09 −11.26 −13.14 −0.80
Uganda 1992–2017 EA 3.18 1.75 −0.90 −1.67 −2.24 0.13
Zambia 1991–2015 SA 1.90 −0.12 0.30 −0.10 −0.53 −0.24
Mean 1.69 1.93 −1.90 −3.17 −4.02 −0.65
Median 1.84 1.93 −1.03 −2.57 −3.52 −0.52
Max 5.17 6.26 0.85 0.92 1.12 0.88
(Mozam- (Eswa- (Kenya) (Cote (Cote (Guinea-
bique) tini) d’Ivoire) d’Ivoire) Bissau)
Min −1.30 −3.32 −11.36 −13.37 −13.35 −2.66
(Burundi) (Malawi) (Alge- (Alge- (Alge- (Mali)
ria) ria) ria)
Notes: See Table 1A.
With respect to the role of inequality, many of the countries experienced limited progress
on poverty, thanks to the relatively poor performance on income distribution during the
periods of observation. For example, with a PTEV of (2,3,4,4,4;5), the relatively good
performance on per capita GDP and income growth in Tanzania was only transformed into
at most average progress on poverty, due to the dismal performance on income distribution.
Similarly, exhibiting a PTEV of (3,2,2,3,3;5), Cameroon performed well on income growth,
which was then translated into strong progress on poverty incidence; however, the country’s
poor performance on inequality has led to relatively weak progress on the poverty gap and
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii170 Augustin Kwasi Fosu
Table 2A. Poverty Transformation Efficiency Vector (Poverty line: US$1.90 a day in 2011 PPP)
Country Efficiency vector Country Efficiency vector
Algeria (3,3,1,1,2;2) Lesotho (1,5,3,3,2;2)
Botswana (2,4,2,2,2;3) Madagascar (5,5,5,5,5;3)
Burkina Faso (1,2,3,2,1;1) Malawi (3,5,5,5,5;1)
Burundi (5,4,4,4,5;5) Mali (3,1,2,2,1;1)
Cameroon (3,2,2,3,3;5) Mauritania (4,2,1,1,1;1)
CAR (5,1,4,3,3;3) Morocco (2,4,2,3,4;4)
Cote d’Ivoire (4,5,5,5,5;4) Mozambique (1,2,4,4,3;4)
Egypt (2,4,3,4,4;4) Niger (5,2,3,3,3;3)
Eswatini (2,1,1,1,2;3) Nigeria (3,4,4,3,3;2)
Ethiopia (1,3,2,2,2;2) Senegal (4,3,3,2,2;2)
Gambia (5,1,1,1,1;1) South Africa (4,3,3,4,4;5)
Ghana (1,1,1,2,3;5) Tanzania (2,3,4,4,4;5)
Guinea (4,1,2,1,1;1) Tunisia (2,2,1,1,1;3)
Guinea-Bissau (5,4,5,5,5;5) Uganda (1,3,4,4,4;4)
Kenya (4,5,5,5,5;2) Zambia (3,5,5,5,4;4)
Notes: Each vector has as coordinates the quintile ranks using data from Table 1A: per capita GDP growth,
income growth, P0 growth, P1 growth, P2 growth, and inequality (Gini) growth, respectively. Note that a lower
rank indicates a more favourable outcome.
Table 2B. Poverty Transformation Efficiency Vector (Poverty line: US$3.20 a day in 2011 PPP)
Country Efficiency vector Country Efficiency vector
Algeria (3,3,1,1,1;2) Lesotho (1,5,3,3,3;2)
Botswana (2,4,2,3,3;3) Madagascar (5,5,5,5,5;3)
Burkina Faso (1,2,3,2,2;1) Malawi (3,5,5,5,5;1)
Burundi (5,4,4,4,4;5) Mali (3,1,3,2,2;1)
Cameroon (3,2,2,2,3;5) Mauritania (4,2,1,1,1;1)
CAR (5,1,4,3,3;3) Morocco (2,4,1,2,2;4)
Cote d’Ivoire (4,5,5,5,5;4) Mozambique (1,2,4,4,3;4)
Egypt (2,4,2,4,4;4) Niger (5,2,3,3,3;3)
Eswatini (2,1,2,1,1;3) Nigeria (3,4,4,4,4;2)
Ethiopia (1,3,3,2,2;2) Senegal (4,3,3,3,2;2)
Gambia (5,1,1,1,1;1) South Africa (4,3,2,3,4;5)
Ghana (1,1,1,1,2;5) Tanzania (2,3,4,4,4;5)
Guinea (4,1,2,2,1;1) Tunisia (2,2,1,1,1;3)
Guinea-Bissau (5,4,5,5,5;5) Uganda (1,3,4,4,4;4)
Kenya (4,5,5,5,5;2) Zambia (3,5,5,5,5;4)
Notes: See Table 2A.
squared poverty gap. A similar observation is made of Ghana, which displays a PTEV of
(1,1,1,2,3;5).
In contrast, strong records on income distribution actually supported progress on
poverty in several countries, including Burkina Faso, Gambia, Guinea, Malawi, Mali and
Mauritania. For example, with respective PTEVs of (5,1,1,1,1;1) and (4,1,2,1,1;1), even
though Gambia and Guinea performed poorly on per capita GDP growth, their income
growth was stellar and was indeed transformed into strong poverty reduction, thanks to
the high complementarity from improvements in income distribution.
With respect to the bottom performers on poverty reduction, the transformation vectors
presented in Tables 2A and B show that the weak progress of Burundi, Cote d’Ivoire
and Guinea-Bissau on poverty alleviation is a combination of the lack of income growth,
mainly reflecting the countries’ poor performance on per capita GDP, as well as the
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii171
worsening income distribution. In this regard, Cote d’Ivoire, for example, exhibits a vector
of (4,5,5,5,5;4) at both US$1.90 and US$3.20 poverty standards. Indeed, based on PTEV,
one could judge for each country, the ‘pro-poor’ nature of economic growth, as well as the
implications of growth for the progress on poverty, as mediated by changes in inequality.
4. Estimating model
This section provides a quantitative analysis of the importance of growth and inequality
for poverty reduction in Africa, based on the decomposition of poverty changes into the
contributions by income growth and by changes in income distribution (see Fosu, 2015,
2017b, 2018). First, the following ‘identity’ model is estimated:3
p = b1 + b2y+ b3yGI + b4y (Z/Y)+ b g+ b gGI5 6 + b7g (Z/Y)+ b8GI + b9Z/Y (1)
where p is the growth of the poverty rate, y is income growth, g is growth in the Gini
coefficient as a measure of the level of inequality,GI is the initial Gini coefficient (expressed
in natural logarithm),Z/Y is the ratio of the poverty lineZ to income Y (expressed in natural
logarithm), and bj (j = 1, 2, . . . ,9) are the respective coefficients to be estimated.
The sign of b2 is anticipated to be negative, since an increase in income growth would
decrease poverty growth, ceteris paribus.4 In contrast, b3 is expected to be positive; a higher
level of initial inequality would reduce the rate at which growth acceleration is translated
into poverty reduction. The sign of b4 should also be positive, consistent with the notion,
based on the lognormal income distribution, that a larger income (relative to the poverty
line) would have associated with it a higher income elasticity.5
The sign of b5 is expected to be positive, as a worsening income distribution should
increase poverty, ceteris paribus. In contrast, the sign of b6 is likely to be negative for a
diminishing poverty-increasing effect of rising inequality. The sign of b7 would be negative
as well, for in a relatively low-income economy (high Z/Y) improving income distribution
(lowering g) might worsen poverty by raising the likelihood that more people would fall into
poverty. Finally, b8 and b9 are likely to be positive: rising initial inequality or increasing
poverty line relative to income should, ceteris paribus, exacerbate poverty, respectively;
however, these coefficients do not affect the income or inequality elasticity of poverty.
From Equation (1), the respective income and inequality elasticities are obtainable as:
Ey = b2 + b3GI + b4Z/Y (2)
Eg = b5 + b I6G + b7Z/Y (3)
Therefore, given the above expected signs of the regression coefficients, Ey and Eg are
generally anticipated to be negative and positive, respectively, so that raising income growth
should reduce the growth of poverty, while worsening inequality changes would exacerbate
poverty increases. It is conceivable, though, that perverse signs of the elasticities could
occur. For example, in a highly unequal (high GI) and low-income (high Z/Y) economy,
the magnitude of the combined positive-signed b3 and b4 could actually overwhelm the
magnitude of the negative-signed b2, thus rendering Ey positive. Similarly, in such an
3 This ‘identity’ model, first derived by Bourguignon (2003), is based on an approximation to an assumed
lognormal income distribution, and allows one to explain the heterogeneity of the nexus across countries and
time periods. For details of the application of the Bourguignon model see, for example, Fosu (2009, 2011) and
Kalwij and Verschoor (2007); see also Epaulard (2003) for a version of this model.
4 For details on the expected signs of the coefficients see, for instance, Bourguignon (2003) and Epaulard
(2003).
5 I ignore the sign and adopt the convention of referring to the income elasticity by its magnitude.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii172 Augustin Kwasi Fosu
economy, Eg could be negative. These two elasticities are critical for determining what
happens to poverty reduction over time in a given economy.
Thus, poverty would decline faster as: (1) income growth is higher, (2) the decline in
inequality is larger, (3) initial inequality is smaller or (4) as income relative to the poverty
line is higher. Furthermore, the income growth and inequality-lowering effects on poverty
would be, respectively, larger as: (a) initial inequality is lower and (b) income relative to the
poverty line is higher. These last two effects, therefore, work via the income and inequality
elasticities of poverty. That is, both elasticities would decrease with initial inequality but
increase with income relative to the poverty line. Hence, initial inequality plays only a part
of this growth-to-poverty transformation process; the level of income also matters.
5. Data and estimation
The above model is estimated using a panel of at least 49 African countries over the period
1985-present, conditional on data availability at the time of the analysis. The World Bank
(2020a) PovcalNet database provides the main data source for the present analysis. The
panel derived from this database is extremely unbalanced, as it is based on available country
surveys, which may be conducted in different years across countries. Thus, considerable
adjustments are required in order to obtain reasonably reliable regional estimates for the
time series (Chen and Ravallion, 2008). The data are therefore revised by the World Bank as
necessary, with implications for regional comparability over time. For example, Fosu (2015)
finds less-than-stellar performance for SSA compared with South Asia since the mid-1990s,
when 2014 rather than 2009 PovcalNet data are used. This revelation suggests that as the
data are revised and improved, updated studies are called for.
The estimation is conducted with the fixed effects (FE), random effects (RE) and two-
step system GMM (SGMM) techniques. Estimates of Ey and Eg are then obtained using
Equations (2) and (3), respectively. Following Fosu (2018), the ‘optimal’ elasticity estimates
selected for the final poverty decomposition are those that minimise the prediction error, r,
in the decomposition formula:
p = yEy + gEg + r; (4)
where, as already defined, p, y, and g are growth rates of poverty, income and inequality
(the Gini coefficient), respectively, and Ey and Eg are the respective income and inequality
elasticities; r is the residual term.
Using the two selection criteria proposed by Fosu (2018), namely, the root mean squared
(RMSE) and mean absolute error (MAE), the estimation procedure with the best predictive
power is identified as that with the minimum RMSE and MAE. Ideally, there would be
no conflict between these two criteria, but where there is, a judgment call must be made.
The results for both criteria are provided in Tables 3A and B for the US$1.90 and US$3.20
poverty lines, respectively.
Focusing, first, on the US$1.90 poverty levels, RE clearly dominates; it is ‘optimal’ for P0,
P1, and P2 as it minimises both RMSE and MAE. Incidentally, SGMM predicts quite poorly
in each case, consistent with the observation in Fosu (2018), except for the squared poverty
gap where the results are mixed. In the case of the US$3.20 standard, RE is ‘optimal’ for P1
and P2, but FE dominates in P0. The values of both RMSE and MAE, however, differ quite
marginally between FE and RE for P0 and P1. Thus, both techniques are equally reliable in
predicting the headcount ratio and the poverty gap. And, once again, SGMM predicts quite
poorly, despite its ability to account for potential endogeneity (Fosu, 2018).6
6 Note that the SGMM results suggest that the usual condition of the appropriateness of instruments is
satisfied (see Fosu, 2018).
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii173
Table 3A. Root RMSE and MAE, Poverty line: US$1.90 a day in 2011 PPP
Headcount ratio (P0) Poverty gap (P1) Squared poverty gap (P2)
RMSE MAE RMSE MAE RMSE MAE
Fixed effects 4.04 2.87 4.77 3.27 6.54 4.09
Random effects 3.82 2.74 4.43 3.05 5.95 3.61
Two-step system GMM 5.29 3.84 6.33 4.74 6.33 6.79
Notes: RMSE and MAE are computed based on Equation (4) of the text, using the observed and predicted
values of the poverty growth rate for all sample countries.
Table 3B. RMSE and MAE, Poverty line: US$3.20 a day in 2011 PPP
Headcount ratio (P0) Poverty gap (P1) Squared poverty gap (P2)
RMSE MAE RMSE MAE RMSE MAE
Fixed effects 2.13 1.44 2.99 1.86 3.55 2.22
Random effects 2.14 1.45 2.95 1.83 3.38 2.10
Two-step system GMM 2.33 1.61 3.57 2.28 4.72 3.29
Notes: See Table 3A.
5.1. Income and inequality elasticities
Figure 1, based on estimates from the ‘optimal’ RE model, illustrates for P0 at the US$1.90
standard a negative relationship between the income elasticity Ey (in absolute value) and
initial inequality. It shows a distribution about ‘Africa’, which is based on Equation (2), with
the (Africa) sample mean income relative to the poverty line employed, while the country
estimates are derived using the country values. Note that the Central African Republic
(CAR), for instance, has the lowest Ey estimate, while Algeria (ALG) enjoys the highest.
CAR’s lowest elasticity status derives from having the highest initial inequality combined
with the lowest income; Algeria (ALG) enjoys the highest income with relatively low initial
inequality, however. Meanwhile, South Africa’s Ey estimate is large despite its high initial
inequality, thanks to its large income.
Figure 2 similarly depicts a negative relationship between the inequality elasticity Eg.
and initial inequality for the US$1.90 standard, also based on Equation (3), using the RE
‘optimal’ estimates. It is noteworthy that Eg appears to be less sensitive to initial inequality
than Ey was, suggesting that disparities of inequality across African countries matter less
for the effect of income distribution to influence the pattern of poverty progress, than for
the impact of income growth.
With respect to differences across countries, Tunisia (TUN) enjoys the highest Eg
value, thanks mainly to the country’s very high income and moderate initial inequality.
South Africa also enjoys a very large Eg estimate due entirely to the country’s very high
income, even though its initial inequality is quite large. Again, CAR exhibits the lowest Eg
estimate due to the country’s extremely low income and very high initial inequality levels.
Interestingly, note that several countries register negative Eg values; these are countries
with very low incomes. As argued in Fosu (2010c, 2015, 2017a, 2017b), for instance, such
negative values are indeed permissible, thanks to the likelihood that in such low countries,
reducing inequality that might likely entail transferring income from the near-poor to the
poor might actually render greater poverty prevalence overall.
5.2. Contributions of growth and inequality changes to poverty reduction
Assessed next are the relative roles of income and inequality in the performance of African
countries on poverty since the early-mid-1990s to the present when the continent has
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii174 Augustin Kwasi Fosu
4
3.5 ALG MAR
EGY TUN
3
GHA
2.5 CIV
TZA SACMR MRT BWA
2 NERUGA GNB GMBBDI ETH NGA
1.5 GIN
KEN
MDG BFA SWZMLI SEN LSOZMB Africa
1 MOZ
CAR MWI
0.5
0
30 35 40 45 50 55 60 65 70
Inial Gini index
Figure 1. Income Elasticity (absolute valued) versus Initial Inequality—Based on the Optimally Selected RE
Results [US$1.90 a day in 2011 PPP, Headcount Ratio]. Notes: The income elasticities are obtained based on
Equation (2). The straight line is the Africa line. The countries above the line have higher levels of income
relative to the average, while those below the line have relatively low incomes.
7.50
6.50
ALG TUN SA
5.50 MAR BWA
4.50 EGY
GHA
3.50 MRT
CIV CMR
2.50 GMB KEN SWZ LSO Africa
1.50 UGA GNB
E ZMB
0.50 TZA N
THGA SEN
NER GINBFA MLI MWI
-0.50 30 35 MDG MOZBDI 40 45 50 55 60 CAR 65 70
-1.50
-2.50
Inial Gini index
Figure 2. Inequality Elasticity versus Initial Inequality—Based on the Optimally Selected RE Results [US$1.90
a day in 2011 PPP, Headcount Ratio]. Notes: The inequality elasticities are obtained based on Equation (3).
See Figure 1 for further details.
generally experienced growth resurgence. Based on Equation (4), changes in poverty during
the period are decomposed into the relative contribution of income growth and changes in
inequality. The results are reported in Tables 4A-C for the three measures of poverty at the
US$1.90 poverty standard, and in Tables 5A-C for the US$3.20 poverty standard.
On average, growth has been critical for poverty reduction in Africa, according to
Tables 4A-C and Tables 5A-C. The decomposition results show greater relative contri-
butions by income growth to poverty changes compared to inequality. For instance, the
decomposition of the headcount ratio at the US$1.90 poverty line (see Table 4A) shows that
although the contributions of growth and inequality have been largely complementary, of
Inequality elascity Income elasticity (absolute value)
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii175
Table 4A. Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth—Based
on the Optimally Selected RE Results [US$1.90 a day in 2011 PPP, Headcount Ratio], Early-mid-1990s—
‘present’
Countries experiencing poverty reduction
A B A+B
Country Period Sub- Headcount Ey∗dlnY Eg∗dlnG Predicted
region poverty (P0) headcount
growth poverty (P0)
growth
Algeria 1995–2011 NA −15.63 −5.66 −8.92 −14.58
Botswana 1993–2016 SA −3.69 −2.97 −3.20 −6.17
Burkina Faso 1994–2014 WA −3.25 −5.26 −0.19 −5.46
Burundi 1992–2014 EA −0.55 −2.2 −0.52 −2.71
Cameroon 1996–2014 CA −3.77 −7.14 0.79 −6.35
CAR 1992–2008 CA −1.57 −3.44 0.39 −3.05
Egypt 1991–2018 NA −3.01 −1.59 −0.24 −1.83
Eswatini 1995–2016 SA −4.85 −9.26 −1.03 −10.28
Ethiopia 1995–2016 EA −3.93 −3.84 −1.14 −4.98
Gambia 1998–2015 WA −11.14 −11.12 −4.06 −15.19
Ghana 1992–2017 WA −5.33 −10.95 1.91 −9.04
Guinea 1991–2012 WA −4.45 −9.51 −0.46 −9.97
Lesotho 1994–2017 SA −2.64 −0.4 −3.19 −3.59
Mali 1994–2010 WA −3.32 −5.51 −0.07 −5.58
Mauritania 1993–2014 WA −9.19 −4.77 −7.13 −11.9
Morocco 1991–2014 NA −4.55 −4.11 0.23 −3.88
Mozambique 1996–2014 SA −1.40 −3.52 −0.02 −3.54
Niger 1993–2014 WA −2.59 −4.90 −0.03 −4.94
Nigeria 1992–2019 WA −1.53 −1.65 −0.68 −2.33
Senegal 1991–2011 WA −2.86 −2.79 −1.47 −4.26
South Africa 1993–2015 SA −2.39 −5.26 1.65 −3.61
Tanzania 1992–2018 EA −1.47 −4.14 0.29 −3.85
Tunisia 1990–2015 NA −14.79 −7.89 −4.89 −12.78
Uganda 1992–2017 EA −1.77 −3.56 0.16 −3.40
Mean −4.57 −5.06 −1.33 −6.39
Countries experiencing poverty increases
A B A+B
Country Period Sub- Headcount Ey∗dlnY Eg∗dlnG Predicted
region poverty (P0) headcount
growth poverty (P0)
growth
Cote d’Ivoire 1993–2015 WA 1.02 1.13 0.71 1.84
Guinea-Bissau 1993–2010 WA 0.16 −1.11 0.94 −0.17
Kenya 1992–2016 EA 0.67 3.59 −3.62 −0.03
Madagascar 1993–2012 EA 0.54 1.83 0.12 1.95
Malawi 1998–2016 EA 0.56 2.64 −0.24 2.4
Zambia 1991–2015 SA 0.26 0.14 −0.26 −0.12
Mean 0.54 1.37 −0.39 0.98
Notes: A: Predicted poverty growth due to income growth, B: pred∗icted poverty growth due to changes ininequality; A+B: total predicted poverty growth due to A and B. Most recent year for which data are
available.
the twenty-four African countries that experienced poverty reduction, twenty have reduced
poverty owing primarily to their positive performance on income growth. In the four
remaining countries, namely, Algeria, Botswana, Lesotho and Mauritania, however, the
fall in poverty was driven mainly by the reduction in inequality. Overall, these results are
consistent across the FGT poverty measures.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii176 Augustin Kwasi Fosu
Table 4B. Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth—Based
on the Optimally Selected RE Results [US$1.90 a day in 2011 PPP, Poverty Gap], Early-mid-1990s—‘present’
Countries experiencing poverty reduction
A B A+B
Country Period Sub- Pov. gap Ey∗dlnY Eg∗dlnG Predicted
region (P1) growth Pov. gap
(P1) growth
Algeria 1995–2011 NA −12.32 −6.23 −11.54 −17.78
Botswana 1993–2016 SA −5.23 −3.31 −4.07 −7.38
Burkina Faso 1994–2014 WA −7.43 −6.66 −1.41 −8.07
Burundi 1992–2014 EA −0.88 −2.71 0.01 −2.69
Cameroon 1996–2014 CA −3.89 −8.24 1.12 −7.12
CAR 1992–2008 CA −3.73 −5.09 0.07 −5.01
Egypt 1991–2018 NA −2.57 −1.78 −0.32 −2.10
Eswatini 1995–2016 SA −7.80 −11.23 −1.50 −12.73
Ethiopia 1995–2016 EA −5.80 −4.65 −2.24 −6.89
Gambia, The 1998–2015 WA −16.03 −13.09 −5.97 −19.06
Ghana 1992–2017 WA −5.30 −12.37 2.60 −9.77
Guinea 1991–2012 WA −8.42 −11.89 −1.72 −13.61
Kenya 1992–2016 EA −0.06 4.27 −5.17 −0.90
Lesotho 1994–2017 SA −4.83 −0.49 −4.67 −5.15
Mali 1994–2010 WA −7.55 −7.08 −2.06 −9.14
Mauritania 1993–2014 WA −10.78 −5.48 −9.69 −15.18
Morocco 1991–2014 NA −3.64 −4.52 0.30 −4.22
Mozambique 1996–2014 SA −3.00 −4.74 0.01 −4.73
Niger 1993–2014 WA −4.36 −5.94 −0.25 −6.19
Nigeria 1992–2019 WA −3.14 −2.01 −1.50 −3.52
Senegal 1991–2011 WA −5.10 −3.48 −2.74 −6.22
South Africa 1993–2015 SA −2.59 −5.82 2.08 −3.74
Tanzania 1992–2018 EA −2.55 −4.95 0.79 −4.16
Tunisia 1990–2015 NA −17.70 −8.67 −6.27 −14.94
Uganda 1992–2017 EA −2.78 −4.25 0.29 −3.96
Zambia 1991–2015 SA −0.61 0.18 −0.47 −0.29
Mean −5.70 −5.39 −2.09 −7.48
Countries experiencing poverty increases
A B A+B
Country Period Sub- Pov. Gap Ey∗dlnY Eg∗dlnG Predicted
region (P1) growth Pov. Gap
(P1) growth
Cote d’Ivoire 1993–2015 WA 1.33 1.30 1.00 2.30
Guinea-Bissau 1993–2010 WA 0.41 −1.34 1.78 0.44
Madagascar 1993–2012 EA 1.23 2.34 −0.11 2.23
Malawi 1998–2016 EA 0.95 3.73 −1.64 2.09
Mean 0.98 1.51 0.26 1.77
Notes: See Table 4A.
As suggested earlier by the qualitative analysis, among the top performers on poverty
reduction, Algeria, Botswana, Eswatini, Ethiopia, Gambia, Ghana, Guinea, Mali, Maurita-
nia and Tunisia have achieved considerable progress on all the poverty measures as a result
of the joint contributions of growth and inequality reductions. Meanwhile, in countries
like Cameroon and Ghana, greater progress might have been achieved if inequality had not
increased.
In countries where poverty increased, the findings are quite mixed. In Cote d’Ivoire and
Madagascar, for example, the increase in poverty can be attributed to the complementary
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii177
Table 4C.Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth— Based
on the Optimally Selected RE Results [US$1.90 a day in 2011 PPP, Squared Poverty Gap], Early-mid-1990s
– ‘present’
Countries experiencing poverty reduction
A B A+B
Country Period Sub- Squared Ey∗dlnY Eg∗dlnG Predicted
region Pov. gap squared
(P2) growth Pov. gap
(P2) growth
Algeria 1995–2011 NA −7.50 −6.25 −14.22 −20.47
Botswana 1993–2016 SA −6.18 −3.56 −4.94 −8.50
Burkina Faso 1994–2014 WA −10.59 −7.65 −2.17 −9.81
Burundi 1992–2014 EA −1.07 −2.91 0.31 −2.60
Cameroon 1996–2014 CA −3.81 −8.75 1.41 −7.34
CAR 1992–2008 CA −5.20 −6.83 −0.06 −6.88
Egypt 1991–2018 NA −2.32 −1.78 −0.41 −2.19
Eswatini 1995–2016 SA −9.89 −13.02 −1.90 −14.91
Ethiopia 1995–2016 EA −6.83 −5.13 −3.05 −8.18
Gambia, The 1998–2015 WA −19.51 −14.27 −7.59 −21.87
Ghana 1992–2017 WA −4.99 −12.73 3.26 −9.48
Guinea 1991–2012 WA −11.23 −13.48 −2.55 −16.03
Kenya 1992–2016 EA −0.64 4.83 −6.48 −1.65
Lesotho 1994–2017 SA −6.44 −0.57 −5.87 −6.44
Mali 1994–2010 WA −10.66 −8.28 −3.26 −11.54
Mauritania 1993–2014 WA −11.92 −5.89 −12.05 −17.94
Morocco 1991–2014 NA −2.56 −4.57 0.36 −4.20
Mozambique 1996–2014 SA −4.06 −5.79 0.02 −5.77
Niger 1993–2014 WA −5.61 −6.38 −0.39 −6.77
Nigeria 1992–2019 WA −4.47 −2.23 −2.09 −4.32
Senegal 1991–2011 WA −6.70 −4.04 −3.66 −7.70
South Africa 1993–2015 SA −2.21 −6.18 2.52 −3.66
Tanzania 1992–2018 EA −3.40 −5.26 1.15 −4.11
Tunisia 1990–2015 NA −18.35 −8.78 −7.68 −16.46
Uganda 1992–2017 EA −3.58 −4.58 0.39 −4.20
Zambia 1991–2015 SA −1.41 0.22 −0.61 −0.40
Mean −6.58 −5.92 −2.68 −8.59
Countries experiencing poverty increases
A B A+B
Country Period Sub- Squared Ey∗dlnY Eg∗dlnG Predicted
region Pov. gap squared
(P2) growth Pov. gap
(P2) growth
Cote d’Ivoire 1993–2015 WA 1.62 1.35 1.27 2.62
Guinea-Bissau 1993–2010 WA 0.60 −1.47 2.41 0.94
Madagascar 1993–2012 EA 1.59 2.68 −0.24 2.44
Malawi 1998–2016 EA 1.20 4.96 −2.42 2.54
Mean 1.25 1.88 0.26 2.14
Notes: See Table 4A.
effect of a fall in household income and an increase in inequality over the period. In Kenya,
Malawi and Zambia, the rise in poverty is solely driven by the reduction in income, while
in Guinea-Bissau, an increase in inequality has been the primary determinant of worsening
poverty.
The dominant role of growth in African countries’ progress on poverty from the early-
mid-1990s to the 2000s is also confirmed at the US$3.20 poverty standard, and for
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii178 Augustin Kwasi Fosu
Table 5A.Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth— Based
on the Optimally Selected FE Results [US3.20 a day in 2011 PPP, Headcount Ratio], Early-mid-1990s –
‘present’
Countries experiencing poverty reduction
A B A+B
Country Sub-region Period Headcount Ey∗dlnY Eg∗dlnG Predicted
poverty (P0) headcount
growth poverty (P0)
growth
Algeria 1995–2011 NA −11.36 −2.97 −2.29 −5.26
Botswana 1993–2016 SA −1.77 −1.58 −0.95 −2.53
Burkina Faso 1994–2014 WA −0.97 −2.59 −0.03 −2.62
Burundi 1992–2014 EA −0.27 −1.07 −0.26 −1.34
Cameroon 1996–2014 CA −2.63 −3.68 0.21 −3.47
CAR 1992–2008 CA −0.68 −1.60 0.06 −1.53
Egypt 1991–2018 NA −1.24 −0.82 −0.06 −0.88
Eswatini 1995–2016 SA −2.67 −4.76 −0.33 −5.09
Ethiopia 1995–2016 EA −1.21 −1.92 −0.26 −2.19
Gambia, The 1998–2015 WA −4.72 −5.70 −1.12 −6.83
Ghana 1992–2017 WA −3.82 −5.67 0.48 −5.19
Guinea 1991–2012 WA −1.56 −4.70 −0.08 −4.79
Guinea-Bissau 1993–2010 WA −0.12 −0.56 0.21 −0.35
Lesotho 1994–2017 SA −1.10 −0.20 −1.05 −1.25
Mali 1994–2010 WA −1.00 −2.70 −0.03 −2.73
Mauritania 1993–2014 WA −5.06 −2.48 −2.01 −4.49
Morocco 1991–2014 NA −4.04 −2.16 0.06 −2.10
Mozambique 1996–2014 SA −0.57 −1.69 0.00 −1.69
Niger 1993–2014 WA −0.93 −2.43 0.02 −2.40
Nigeria 1992–2019 WA −0.40 −0.82 −0.15 −0.97
Senegal 1991–2011 WA −1.06 −1.40 −0.46 −1.86
South Africa 1993–2015 SA −1.33 −2.81 0.48 −2.32
Tanzania 1992–2018 EA −0.67 −2.06 0.00 −2.06
Tunisia 1990–2015 NA −9.09 −4.09 −1.26 −5.36
Uganda 1992–2017 EA −0.90 −1.79 0.03 −1.76
Mean −2.37 −2.49 −0.35 −2.84
Countries experiencing poverty increases
A B A+B
Country Sub-region Period Headcount Ey∗dlnY Eg∗dlnG Predicted
poverty (P0) headcount
growth poverty (P0)
growth
Cote d’Ivoire 1993–2015 WA 0.63 0.58 0.18 0.76
Kenya 1992–2016 EA 0.85 1.85 −1.11 0.75
Madagascar 1993–2012 EA 0.13 0.89 0.05 0.94
Malawi 1998–2016 EA 0.28 1.28 −0.31 0.97
Zambia 1991–2015 SA 0.30 0.07 −0.09 −0.02
Mean 0.44 0.93 −0.26 0.68
Notes: See Table 4A.
the poverty gap and squared poverty gap. Yet, the contribution of inequality cannot be
overlooked, since in most of the countries, it played a complementary role in determining
the rate of poverty changes; indeed, in a small number of countries, its role was actually
dominant.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii179
Table 5B.Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth— Based
on the Optimally Selected RE Results [US$3.20 a day in 2011 PPP, Poverty Gap], Early-mid-1990s – ‘present’
Countries experiencing poverty reduction
A B A+B
Country Period Sub- Pov. gap Ey∗dlnY Eg∗dlnG Predicted
region (P1) growth Pov. gap
(P1) growth
Algeria 1995–2011 NA −13.37 −4.12 −5.26 −9.38
Botswana 1993–2016 SA −3.06 −2.27 −2.03 −4.30
Burkina Faso 1994–2014 WA −3.54 −4.35 −0.76 −5.11
Burundi 1992–2014 EA −0.57 −1.71 −0.10 −1.81
Cameroon 1996–2014 CA −3.36 −5.45 0.53 −4.92
CAR 1992–2008 CA −2.21 −3.40 −0.08 −3.48
Egypt 1991–2018 NA −1.90 −1.16 −0.14 −1.30
Eswatini 1995–2016 SA −5.00 −7.63 −0.80 −8.43
Ethiopia 1995–2016 EA −3.26 −3.04 −1.08 −4.12
Gambia, The 1998–2015 WA −9.20 −8.69 −2.93 −11.62
Ghana 1992–2017 WA −4.64 −8.15 1.19 −6.96
Guinea 1991–2012 WA −4.60 −7.76 −0.88 −8.64
Lesotho 1994–2017 SA −2.73 −0.33 −2.53 −2.87
Mali 1994–2010 WA −3.87 −4.64 −1.21 −5.85
Mauritania 1993–2014 WA −7.38 −3.67 −4.74 −8.41
Morocco 1991–2014 NA −4.43 −3.01 0.14 −2.87
Mozambique 1996–2014 SA −1.75 −3.11 0.01 −3.10
Niger 1993–2014 WA −2.41 −3.81 −0.09 −3.90
Nigeria 1992–2019 WA −1.51 −1.32 −0.73 −2.04
Senegal 1991–2011 WA −2.85 −2.32 −1.48 −3.80
South Africa 1993–2015 SA −1.98 −3.99 1.03 −2.95
Tanzania 1992–2018 EA −1.42 −3.18 0.31 −2.87
Tunisia 1990–2015 NA −11.26 −5.70 −2.84 −8.54
Uganda 1992–2017 EA −1.67 −2.77 0.13 −2.64
Zambia 1991–2015 SA −0.10 0.12 −0.27 −0.15
Mean −3.92 −3.82 −0.98 −4.80
Countries experiencing poverty increases
A B A+B
Country Period Sub- Pov. gap Ey∗dlnY Eg∗dlnG Predicted
region (P1) growth Pov. gap
(P1) growth
Cote d’Ivoire 1993–2015 WA 0.92 0.85 0.46 1.31
Guinea-Bissau 1993–2010 WA 0.15 −0.88 0.84 −0.03
Kenya 1992–2016 EA 0.59 2.89 −2.67 0.22
Madagascar 1993–2012 EA 0.63 1.52 −0.06 1.46
Malawi 1998–2016 EA 0.58 2.55 −1.28 1.27
Mean 0.57 1.39 −0.54 0.85
Notes: See Table 4A.
6. Conclusion and implications
UsingWorld Bank data, this paper, first, examined Africa’s historical record, compared with
the other regions of the world, on poverty incidence, spread and severity since the early-
mid 1990s at the US$1.90 and US$3.20 per day poverty standards. Second, it evaluated
country-specific progress on growth, poverty and inequality, and compared the ‘poverty
transformation efficiency vector’ (PTEV) amongAfrican countries.This arguably innovative
approach allowed comparisons across African countries even prior to the application of
regression methods. Third, the study analysed the relative roles of income growth and
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii180 Augustin Kwasi Fosu
Table 5C.Decomposition of Poverty Growth into the Contributions of Inequality and Income Growth— Based
on the Optimally Selected RE Results [US$3.20 a day in 2011 PPP, Squared Poverty Gap], Early-mid-1990s
– ‘present’
Countries experiencing poverty reduction
A B A+B
Country Period Sub- Squared Ey∗dlnY Eg∗dlnG Predicted
region Pov. gap squared
(P2) growth Pov. gap
(P2) growth
Algeria 1995–2011 NA −13.35 −4.77 −7.56 −12.33
Botswana 1993–2016 SA −4.01 −2.65 −2.82 −5.47
Burkina Faso 1994–2014 WA −5.57 −5.36 −1.26 −6.62
Burundi 1992–2014 EA −0.75 −2.09 0.04 −2.05
Cameroon 1996–2014 CA −3.63 −6.45 0.77 −5.68
CAR 1992–2008 CA −3.33 −4.43 −0.15 −4.58
Egypt 1991–2018 NA −2.13 −1.35 −0.21 −1.56
Eswatini 1995–2016 SA −6.68 −9.23 −1.13 −10.36
Ethiopia 1995–2016 EA −4.49 −3.68 −1.67 −5.35
Gambia, The 1998–2015 WA −12.35 −10.38 −4.25 −14.63
Ghana 1992–2017 WA −4.95 −9.55 1.74 −7.81
Guinea 1991–2012 WA −6.81 −9.51 −1.44 −10.95
Lesotho 1994–2017 SA −3.97 −0.40 −3.55 −3.95
Mali 1994–2010 WA −6.03 −5.75 −1.99 −7.74
Mauritania 1993–2014 WA −8.87 −4.34 −6.75 −11.08
Morocco 1991–2014 NA −4.42 −3.48 0.20 −3.28
Mozambique 1996–2014 SA −2.57 −3.93 0.02 −3.91
Niger 1993–2014 WA −3.40 −4.61 −0.18 −4.79
Nigeria 1992–2019 WA −2.41 −1.60 −1.15 −2.74
Senegal 1991–2011 WA −4.13 −2.84 −2.18 −5.02
South Africa 1993–2015 SA −2.22 −4.63 1.43 −3.20
Tanzania 1992–2018 EA −2.01 −3.82 0.55 −3.28
Tunisia 1990–2015 NA −13.14 −6.61 −4.04 −10.65
Uganda 1992–2017 EA −2.24 −3.32 0.20 −3.12
Zambia 1991–2015 SA −0.53 0.15 −0.38 −0.23
Mean −4.96 −4.59 −1.43 −6.02
Countries experiencing poverty increases
A B A+B
Country Period Sub- Squared Ey∗dlnY Eg∗dlnG Predicted
region Pov. gap squared
(P2) Pov. gap
Growth (P2) growth
Cote d’Ivoire 1993–2015 WA 1.12 1.01 0.68 1.68
Guinea-Bissau 1993–2010 WA 0.32 −1.06 1.3 0.25
Kenya 1992–2016 EA 0.28 3.47 −3.78 −0.31
Madagascar 1993–2012 EA 0.98 1.87 −0.13 1.74
Malawi 1998–2016 EA 0.78 3.27 −1.84 1.42
Mean 0.70 1.71 −0.75 0.96
Notes: See Table 4A.
inequality changes in explaining African countries’ poverty record, through a decomposition
of poverty changes, using ‘optimal’ income and inequality elasticity estimates from the
‘identity’ model.
The study finds that while the African poverty record was dismal in the 1980s, poverty
reduction on the continent has been appreciable since the continent’s growth resurgence
starting in the 1990s, and that this progress was driven mainly by income growth, consistent
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
Progress on Poverty in Africa: The Importance of Growth and Inequality ii181
with the global evidence.Nonetheless, inequality often played a complementary role in most
of the countries and, in a small number of cases, it was the primary driver of changes in
poverty. Thus, the present study sheds light on country-specific differences in the relative
roles of growth and inequality in poverty reduction on the continent, based on both
qualitative and quantitative evidence. The study should, therefore, provide a useful compass
to those who seek to understand country-specific situations within the African context.
Based on the present study, one could elicit the relative roles of growth and income
distribution in generating a particular country’s poverty progress outcomes. On the one
hand, where the poverty record is weak, one might craft income growth and/or income
distribution policies required to turn things around. On the other hand, where the poverty
record is favourable, growth and inequality-related factors responsible for the outcome may
be elicited. Such an analysis would not only provide a guidepost for what a country’s policy
makers might wish to continue with, but could also serve as a positive lesson for other
African countries.
Acknowledgements
I am grateful for useful comments from participants in the AERC Collaborative Research
Workshop held in Nairobi during November 2019, to an anonymous reviewer, and to
Dede Gafa for valuable research assistance. I also thank the AERC, as well as the National
Research Foundation (NRF) of South Africa, for grant support.
Supplementary material
Supplementary material is available at Journal of African Economies online.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding
author
References
Adams R. H. (2004) ‘Economic Growth, Inequality and Poverty: Estimating the Growth Elasticity of
Poverty’,World Development, 32(12): 1989–2014.
Alesina A., Rodrik D. (1994) ‘Distributive Politics and Economic Growth’, The Quarterly Journal of
Economics, 109(2): 465–90.
Bourguignon F. (2003) ‘The Growth Elasticity of Poverty reduction: Explaining Heterogeneity across
Countries and Time Periods’, in T.S. Eicher and S.J. Turnovsky (eds), Inequality and Growth: Theory
and Policy Implications, vol. 1. Boston, MA: MIT Press.
Chen S., Ravallion M. (2008) The Developing world is Poorer than We Thought, But No Less Successful
in the Fight against Poverty. Poverty Research Paper No. 4703. Washington, DC: World Bank.
Dabla-Norris E., Kochhar K., Ricka F., Suphaphiphat N., Tsounta E. (2015) Causes and Consequences of
Income Inequality: A Global–> Perspective, IMF Discussion Note, DN/15/13. Washington, DC: IMF.
Deininger K., Squire L. (1998) ‘New Ways of Looking at Old Issues: Inequality and Growth’, Journal of
Development Economics, 57(2): 259–87.
Dollar D., Kraay A. (2002) ‘Growth Is Good for the Poor’, Journal of Economic Growth, 3: 195–225.
Dollar D., Kleineberg T., Kraay A. (2016) ‘Growth Still Is Good for the Poor’,European Economic Review,
81: 68–85.
Easterly W. (2000) The Effect of IMF and World Bank Programs on Poverty. Washington, DC: World
Bank, mimeo.
Epaulard A. (2003)Macroeconomic Performance and Poverty Reduction, IMFWorking Paper No. 03/72.
Washington, DC: IMF.
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023
ii182 Augustin Kwasi Fosu
Fosu A.K. (2008) ‘Inequality and the Growth–Poverty Nexus: Specification Empirics Using African Data’,
Applied Economics Letters, 15(7–9): 563–6.
Fosu A. K. (2009) ‘Inequality and the Impact of Growth on Poverty: Comparative Evidence for Sub-
Saharan Africa’, Journal of Development Studies, 45(5): 726–45.
Fosu A. K. (2010a) ‘The Effect of Income Distribution on the Ability of Growth to Reduce Poverty:
Evidence from Rural and Urban African Economies’, American Journal of Economics and Sociology,
69(3): 1034–53.
Fosu A. K. (2010b) ‘Does Inequality Constrain Poverty Reduction Programs? Evidence from Africa’,
Journal of Policy Modeling, 32(6): 818–27.
Fosu A. K. (2010c) ‘Inequality, Income, and Poverty: Comparative Global Evidence’, Social Science
Quarterly, 91(5): 1432–46.
Fosu A. K. (2015) ‘Growth, Inequality and Poverty in Sub-Saharan Africa: Recent Progress in a Global
Context’,Oxford Development Studies, 43(1): 44–59.
Fosu A. K. (2017a) ‘Growth, Inequality, and Poverty Reduction in Developing Countries: Recent Global
Evidence’, Research in Economics, 71(2): 306–36.
Fosu A.K. (2017b) ‘Growth, Inequality, and Poverty Reduction: Africa in a Global Setting’, inM.Nissanke
andM.Ndulo (eds), Poverty Reduction in the Course of African Development, vol. 2017.Oxford, UK:
Oxford University Press, pp. 57–76.
Fosu A. K. (2018) ‘The Recent Growth Resurgence in Africa and Poverty Reduction: The Context and
Evidence’, Journal of African Economies, 27(1): 92–107.
Kalwij A. Verschoor A. (2007) ‘Not by Growth Alone: The Role of the Distribution of Income in Regional
Diversity in Poverty Production’, European Economic Review, 51(4): 805–829.
Kuznets S. (1955) ‘Economic Growth and Income Inequality’, The American Economic Review, 45(1):
1–28.
Ravallion M. (1997) ‘Can High Inequality Developing Countries Escape Absolute Poverty?’ Economics
Letters, 56(1): 51–7.
Ravallion M. (2012) ‘Why Don’t we See Poverty Convergence?’ American Economic Review, 102(1):
504–23.
Thorbecke E. (2013) ‘The Interrelationship Linking Growth, Inequality and Poverty in Sub-Saharan
Africa’, Journal of African Economies, 22(suppl 1): i15–48.
Thorbecke E., Charumilind C. (2002) ‘Economic Inequality and its Socioeconomic Impact’, World
Development, 30(9): 1477–95.
World Bank (2020a) POVCALNET 2020. http://iresearch.worldbank.org/PovcalNet/index.htm.
World Bank (2020b)World Development Indicators Online 2020. http://data.worldbank.org/data-catalo
g/world-development-indicators
Downloaded from https://academic.oup.com/jae/article/32/Supplement_2/ii164/7118985 by University of Ghana, Legon user on 03 July 2023