Resources Policy 83 (2023) 103702

Contents lists available at ScienceDirect

Resources Policy
journal homepage: www.elsevier.com/locate/resourpol

Natural resources, child mortality and governance quality in

African countries
Sosson Tadadjeu a, Henri Njangang b, Simplice A. Asongu c, d, *, Brice Kamguia a
a
Dschang School of Economics and Management, University of Dschang, Cameroon
b
Faculty of Economics and Management (LAREFA), University of Dschang, Cameroon
c
School of Economics, University of Johannesburg, Johannesburg, South Africa
d
“New Uzbekistan” University, 54, Mustaqillik street, Mirzo Ulugbek district, Tashkent, Uzbekistan

A R T I C L E I N F O A B S T R A C T

JEL classification: This paper contributes to the resource curse literature by investigating the effect of natural resources on under-
J13 five mortality in a sample of 50 African countries over the period 1996 to 2018. Additionally, we examine the
O55 extent to which governance shapes the relationship between natural resources and under-five mortality. The
Q33
empirical analysis is based on the system generalized method of moments and the results show that natural
Q34
Q38
resources have increased under-five mortality. Resource rents also have detrimental effects on child mortality by
age, gender, and the three major causes of infant mortality from infectious diseases. However, an extended
Keywords:
Natural resources
analysis of different types of natural resources suggests that point resources (such as oil, natural gas, and mineral
Child mortality rents) increase under-five mortality in contrast to diffuse resources (such as forest rent). We also find that
Governance governance mitigates the unfavorable effect of natural resources on child mortality. Corresponding governance
Africa policy thresholds that should be attained in order to reverse the unfavorable effects of natural resources on child
mortality are provided. We thus suggest an increase in the funds allocated to the health sector from resource
rents and encourage efforts to improve governance standards in sampled countries.

1. Introduction
In recent decades, the improvement of child health has been and
continues to be on the agenda of development partners and governments
around the world. The United Nations, through the Millennium Devel­
opment Goals (MDGs), urged international agencies and national gov­
ernments to achieve Goal 4, which aimed to reduce child mortality by
two-thirds between 1990 and 2015. The Sustainable Development
Goals (2016–2030) also reaffirm the importance of improving child
survival rates (Goal 3). Given these efforts, the under-five mortality rate
fell from 9.8 million in 2000 to 5.3 million in 2018, a drop of over 45%
(UN IGME, 2019). However, this apparent success hides huge disparities
by region and income level. According to data from IHME (2019), in
2017, 93% of child deaths occurred in low and middle-income countries,
and the majority of these deaths were recorded in Africa. While devel­
oping countries like Peru have been able to reduce their child mortality
rate by more than half between 2000 and 2017 (from 38.6 deaths per
1000 in 2000 to 15.5 deaths in 2017), several African countries still have
death rates per 1000 greater than 100. According to the World Bank
(2020), the average under-five mortality rate in Sub-Saharan Africa in
2018 was 78 per 1,000, compared to a global average of 39 per 1000.
Furthermore, if current fertility and mortality trends continue, Africa
will account for more than 60% of all child deaths by 2030, up from 50%
in 2013 (Liu et al., 2015).
Given the high child mortality rates in some parts of the world,
particularly in Africa, and its potential political and socio-economic
implications, it is highly important to understand its determinants.
Recent studies have made major inroads into documenting the de­
terminants of child mortality, highlighting the roles of factors such as
education (Grépin and Bharadwaj, 2015; Andriano and Monden, 2019),
health expenditure (Anyanwu and Erhijakpor, 2009; Bernet et al.,
2018), health aid (Mishra and Newhouse, 2009; Pickbourn and Ndiku­
mana, 2019), trade openness (Panda, 2020), income inequality (Wil­
kinson and Pickett, 2006), institutional quality (Wigley and
Akkoyunlu-Wigley, 2017; Achim et al., 2019), conflicts (Gates et al.,
2012), globalization (Welander et al., 2015), women parliamentarians

* Corresponding author. School of Economics, University of Johannesburg, Johannesburg, South Africa.

E-mail addresses: stadadjeu@yahoo.fr (S. Tadadjeu), ndieupahenri@gmail.com (H. Njangang), asongusimplice@yahoo.com, asongus@afridev.org (S.A. Asongu),
bricekamguiedj@yahoo.fr (B. Kamguia).

https://doi.org/10.1016/j.resourpol.2023.103702
Received 18 June 2022; Received in revised form 3 May 2023; Accepted 8 May 2023
0301-4207/© 2023 Elsevier Ltd. All rights reserved.
S. Tadadjeu et al.
(Macmillan et al., 2018), remittances (Azizi, 2018), and recently eco­
nomic complexity (Vu, 2020). The present study considers that an
important factor of child mortality that has not been sufficiently
analyzed is natural resources. As Gylfason (2001) suggests, if oil reve­
nues are well managed, they can educate, heal, and create employment
for populations. However, some argue that oil rent engenders authori­
tarianism, increases the risk and duration of conflict, increases income
inequality, and leads to low institutional quality (Van der Ploeg, 2011;
Ross, 2015). As an illustration, in 2019, Nigeria and Angola were the two
largest oil producers in Africa, but they were also the countries with
some of the highest under-five mortality rates in 2018, with 120 and 77
per 1000 live births, respectively. This suggests that natural resources
may paradoxically be a curse rather than a blessing.
Since the influential works of Sachs and Warner (1995, 1999) sup­
porting the negative effect of natural resources on economic growth,
empirical and theoretical studies have analyzed the relationship be­
tween natural resources and economic growth with rather mixed evi­
dence. Several studies provide evidence of a negative relationship
between natural resources and economic growth (Papyrakis and Ger­
lagh, 2004, 2007; Satti et al., 2014; Tiba and Frikha, 2020). However,
other studies provide evidence rejecting the resource curse hypothesis
(Brunnschweiler and Bulte, 2008; Adika, 2020). Through a
meta-analysis of 1419 estimates, Dauvin and Guerreiro (2017) show that
while there is a “soft curse” in developing countries, natural resources do
not harm growth in developed countries. The way natural resources are
measured, as well as their “appropriability,” explains part of the het­
erogeneity of the results found in the literature. In recent decades,
studies have extended the effect of natural resources to other aspects of
economic development, such as health outcomes (De Soysa and Gizelis,
2013; El Anshasy and Katsaiti, 2015; Edwards, 2016; Kim and Lin, 2017;
Chang and Wei, 2019; Madreimov and Li, 2019; Chang, 2020; Oduyemi
et al., 2021). However, few studies have focused on the effect of natural
resources on child mortality, particularly in Africa. Empirical literature
on the effect of natural resources on child mortality is still at the nascent
stage and the results remain inconclusive. Cotet and Tsui (2013) show
that oil reduces infant mortality in 150 countries. Wigley (2017) finds
that petroleum-poor countries outperform petroleum-rich countries in
reducing under-five mortality in a panel of 167 countries. Bellinger and
Fails (2020) analyze the effect of oil wealth on child mortality rates in
non-democratic countries and identify some specific conditions under
which oil can be detrimental to child mortality. What about Africa?
Apart from the literature on natural resources and health, the role of
governance has received very little attention. Mehlum et al. (2006)
suggest that aggregate income is reduced by natural resources when
institutions can easily be captured by a few elites, while when in­
stitutions cannot easily be captured, more resources engender higher
income levels. Several studies share this view that natural resources
interact with the quality of institutions, which determines whether
resource revenues become a blessing or a curse (Robinson et al., 2006;
Epo and Faha, 2020; Njangang et al., 2022). The effect of natural re­
sources on the economy depends critically on institutions, as these
determine the extent to which political incentives are embedded in
policy outcomes (Robinson et al., 2006). Countries with institutions that
promote state accountability and competence will benefit from natural
resources. Countries without such institutions may suffer from a
resource curse. Considering this argument, this study seeks to under­
stand if governance quality plays any significant role in the relationship
between natural resources and child mortality in Africa.
This paper departs from the previous studies and attempts to fill the
gap in the existing literature on at least four points. First, we contribute
to the growing literature on the resource curse within one specific re­
gion. We focus on Africa, where a particularly interesting case is given
that Africa is a resource-rich continent but, paradoxically, a part of the
world where children’s health problems are much more acute. This
study therefore aims to fill this gap by providing one of the first
empirical studies on the relationship between natural resources and
2
Resources Policy 83 (2023) 103702
child mortality in Africa. Second, we analyze the effect of resource rents
on child mortality by the age of the child, gender, as well as the three
main causes of child mortality. Despite progress in the fight against
childhood diseases, infectious diseases1 remain a major cause of death
among children under five, particularly in sub-Saharan Africa and South
Asia. Pneumonia (15%), diarrhea (8%) and malaria (5%) are identified
as the leading causes of death among children under five worldwide (UN
IGME, 2019). Using these original indicators makes it possible to iden­
tify the age, gender and cause of death for which the effects of natural
resources are most detrimental. To our knowledge, no study has tested
the effect of natural resources on these three dimensions of child mor­
tality. Third, several studies have shown that different types of natural
resources can have different effects on economic growth (Bulte et al.,
2005; Yilanci et al., 2021), education expenditures (Cockx and
Francken, 2016), and access to drinking water and sanitation (Tadadjeu
et al., 2020). Their argument implies that different types of natural re­
sources may also have different effects on child mortality. Despite these
arguments, existing studies have focused on the effect of oil on infant
mortality, which gives only a partial view of the resource curse. We fill
this gap by analyzing the effects of five different types of natural re­
sources on under-five mortality. Fourth, we examine the moderating
role of institutional quality in the relationship between natural resources
and child mortality. We assume that better governance promotes
transparency in the collection of natural resource revenues and im­
proves the quality of investments, particularly in the health sector. To
the best of our knowledge, this paper is the first to empirically examine
the role of governance in the relationship between natural resources and
child mortality. Besides using the six governance indicators provided by
the World Governance Indicators (WGI) of the World Bank, we construct
four other indicators through a Principal Component Analysis (PCA),
one representing global governance, and the other three representing
political governance, institutional governance, and economic gover­
nance (Asongu and Nwachukwu, 2016, 2017). To increase the relevance
of policy implications, we use both individual indicators and composite
indicators of governance variables.
To sum up, using the two-step system Generalized Method of Mo­
ments (GMM), the following results are established. We find evidence of
a positive and significant relationship between natural resources and
under-five mortality. Natural resources also have a positive effect on
child mortality by age, gender, and the major causes of infant mortality
from infectious diseases (i.e. pneumonia, diarrhea, and malaria). Look­
ing at the types of natural resources, we find that point resources have a
positive effect on under-five mortality, as opposed to diffuse resources.
Finally, we show that the quality of governance enables African gov­
ernments to partially break the resource curse. Corresponding gover­
nance policy thresholds that should be attained in order to reverse the
positive effects of resources on child mortality are provided.
The rest of this paper is organized as follows: Section 2 describes the
theoretical predictions through which natural resources affect child
mortality. Section 3 discloses the data and methodology, while Section 4
presents the empirical results. Section 5 concludes with implications and
future research directions.
2. Natural resources and child mortality: transmission channels
Several studies have provided a review of the literature on trans­
mission channels explaining the negative correlation between natural
resources and development (see Frankel, 2010; Badeeb et al., 2017).
Based on this theoretical framework, several channels can be conjec­
tured to explain the link between natural resources and child mortality,
including institutions, internal conflict, inequality, and public health
1
Infectious diseases account for 53% of under-five deaths worldwide in 2018
(UN IGME, 2019).
S. Tadadjeu et al.
expenditures.2
2.1. Institutional channel
Commodities can be a blessing or a curse for resource-rich countries,
some of which suffer severely from corruption, while others have been
managed to prosper with a relatively low level of corruption (Neudorfer,
2018). Regarding the type of natural resources, economists have iden­
tified oil as a substantial cause of corruption (Okada and Samreth,
2017). Arezki and Brückner (2011) establish that a growth in oil rent
significantly increases corruption and deteriorates political rights.
Similarly, Okada and Samreth (2017) find that while more oil rents
increase corruption, non-oil resources do not. In the same vein, several
studies have analyzed the relationship between resource wealth and
democracy, especially oil wealth. Most of these studies validate the
premise that higher levels of oil wealth are associated with more stable
autocratic governments, which are less likely to change to liberal dem­
ocratic governments (Anyanwu and Erhijakpor, 2014; Ross, 2015; Ber­
gougui and &Murshed, 2020). Government dependence on oil reduces
democracy and accountability by weakening the links between the state
and society, facilitating government investment in patronage and
repression, and fostering political corruption (Ross, 2015). Regarding
the effects of institutional quality on health, there is a consensus sug­
gesting that corruption reduces health outcomes, while democracy re­
duces child mortality (Dincer and Teoman, 2019; Wigley and
Akkoyunlu-Wigley, 2017). Li et al. (2018) argue that extortion and
bribery in the public and private sectors enable the production and sale
of counterfeit medicines, which lead to patient death and morbidity,
pathogen resistance and reduced drug effectiveness. Authors such as
Besley and Kudamatsu (2006) establish that democracy contributes to
improved health, as measured by life expectancy and infant mortality.
Thus, resource-dependent governments may be characterised by high
levels of corruption and weak democracy, which reduce the provision of
care for better child health.
2.2. Civil conflicts channel
Concerning civil conflicts, the appropriation and exploitation of
natural resources has frequently been mentioned as a cause of civil
conflicts (Collier and Hoeffler, 2004; Welsch, 2008). In fact, competition
for rent can exacerbate existing tensions between ethnic groups or fac­
tions in power, leading to armed conflict. Experts have cited greed and
grievances as reasons for how natural resources can lead to conflict.
Whereas the grievance hypothesis regards civil wars as originating from
poverty, the greed-based explanation (Collier and Hoeffler, 2004) em­
phasizes voracity as a cause of conflict (Welsch, 2008). In
conflict-affected countries, the damage to the health system may sub­
sequently manifest itself in the destruction of health services, the lack of
medical personnel, and the massive displacement of populations,
exposing them to multiple risk factors (Gates et al., 2012). These risk
factors include the increasing prevalence of infectious diseases,
including diarrheal diseases and respiratory infections, which are among
the leading causes of child mortality. One can therefore postulate that
natural resources, through their effects on the risk and duration of civil
conflicts, can lead to higher child mortality rates.
2.3. Income inequality channel
There is an emerging strand of literature suggesting that natural
resources are associated with increased income inequality (Fum and
Hodler, 2010; Buccellato and Mickiewicz, 2009; Carmignani, 2013).
Natural resources are often unequally distributed within countries, be­
tween ethnic groups, or between regions. Basedau and Lay (2009) show
2 3
It should be noted that some of these channels are linked.
3
Resources Policy 83 (2023) 103702
that elites may distribute rents selectively and create clientelism net­
works, from which those who largely benefit are leaders of politically
important organizations. Through this channel, personal ties motivate
access to and distribution of resource revenues, which end-up with a
relatively small part of the population. Regarding the effect of income
inequality on health, the health economic literature provides evidence
suggesting that income inequality is associated with poorer health (see
Wilkinson and Pickett (2006) for a meta-analysis). Therefore, if resource
endowment is an additional factor leading to income inequality, other
things being equal, we should see evidence that natural resources are
associated with higher under-five mortality rates.
2.4. Health expenditure channel
Another channel through which natural resources can affect child
mortality is health expenditure. An important measure of the level of
investment in health is health expenditure in a country. Accordingly,
government expenditure, such as health expenditure, can enable the
administration and provision of health care services and hence play the
role of a relevant policy tool for the government (Anyanwu and Erhi­
jakpor, 2009). Thus, public health expenditure is recognized as having
an important contribution to improving health outcomes (see Bernet
et al., 2018; Chireshe and Ocran, 2020). Theoretically, natural resources
can be an effective way for resource-rich countries to increase health
budgets and therefore positively affect health outcomes. However,
emerging literature shows that natural resources are negatively associ­
ated with health expenditure (Cockx and Francken, 2014; Turan and
Yanıkkaya, 2020). Cockx and Francken (2014)found a significant in­
verse relationship between natural resources and public health expen­
diture. Consequently, through its negative effects on health expenditure,
natural resources will lead to a decrease in health care provision and an
increase in child mortality.
3. Data and methodology
3.1. Data
Our sample covers 50 African countries over the period 1996–2018
with data from various sources: World Bank (2020): World Development
Indicators (WDI), World Bank (2020): World Governance Indicators
(WGI), World Health Organisation (WHO, 2020): Global Health Obser­
vatory data repository, and Alesina et al. (2003). The periodicity under
investigation is chosen according to data availability constraints.
3.1.1. Dependent variable
The principal dependent variable is the under-five mortality rate
(IMR5) which represents the probability, per 1000 live births, that a
newborn baby will die prior to reaching the age of five, if subject of age-
specific mortality rates of the specified year. For robustness, we also use
the under-five mortality rate by gender, distinguishing between the fe­
male mortality rate and the male mortality rate aged 0–5 years.
Although under-five mortality is the most widely used variable for
measuring child mortality, it is worth noting that this variable does not
capture all dimensions of child mortality. As alternative measures in the
robustness analyses, we use the infant mortality rate, which represents
the mortality of children aged 0–12 months (IMR1), and the neonatal
mortality rate, which denotes the mortality of children aged 0–28 days
(IMRNEO). We also use indicators representing the three main causes of
child mortality in developing countries, namely, mortality from pneu­
monia, mortality from diarrhea, and mortality due to malaria. These
indicators reflect the mortality per 1000 live births of children aged
0–59 months due to pneumonia, diarrhea, or malaria. Data for the latter
measures are available in the WHO (2020) .3
Data for these indicators is available for the period 2000–2017.
S. Tadadjeu et al.
3.1.2. Independent variable
Consistent with recent work by Tadadjeu et al. (2021) and Ncho­
foung et al. (2021), we use total natural resource rents as a percentage of
GDP as a measure of natural resource dependence (TNR). Indeed, it is
one of the most widely used measures in the literature on the resource
curse, given its availability. Fig. 1 shows a positive correlation between
natural resources and under-five mortality, which is consistent with the
resource curse hypothesis. In other words, resource-rich countries have
on average higher under-five mortality rates compared resource-poor
countries.
3.1.3. Governance variables
We analyze the role of governance in the relationship between nat­
ural resources and under-five mortality using six unbundled governance
indicators; namely: the rule of law, control of corruption, regulatory
quality, government effectiveness, voice and accountability, and polit­
ical stability, plus four measures that are bundled through PCA, notably
political governance, economic governance, institutional governance,
and general governance.
According to Asongu and Nwachukwu (2016, 2017), this study em­
ploys PCA in order to derive composite governance indicators. Building
on the attendant literature, the PCA is widely used a statistical method
that is employed to reduce a set of highly correlated variables into a
smaller set of uncorrelated variables, which are known as principal
components (PCs). “Eco. gov (economic governance) is the formulation and
implementation of policies that deliver public commodities. Pol. Gov (political
governance) is defined as the election and replacement of political leaders.
Inst. gov (institutional governance) is the respect of the State and citizens of
institutions that govern interactions between them(Asongu and Nwa­
chukwu, 2017 p.258)”. Table 1 presents the results of the PCA and the
construction of each governance indicator. Table 2 displays the sum­
mary statistics.
3.1.4. Control variables
To ensure that the results are not biased by omission of variables and
in line with the recent literature on child mortality (Wigley, 2017;
Pickbourn and Ndikumana, 2019), this paper includes several control
variables. They comprise the log of GDP per capita, female primary
school enrollment, urban population growth, and the prevalence of fe­
male HIV/AIDS. The negative relationship between income per capita
and child mortality is well documented in the health economic literature
(Khanam et al., 2009). Higher incomes may be correlated with healthier
environments (including housing) and more nutritious diets. While
several studies agree that women’s education, particularly through
better knowledge of medical practices in child health, reduces child
mortality (Grépin and Bharadwaj, 2015), the prevalence of HIV/AIDS
Fig. 1. Relationship between natural resources and under-five mortality.
4
Resources Policy 83 (2023) 103702
among women is rather a factor that is detrimental to child survival
(Adetunji, 2000). However, the effect of urbanization, as measured by
the proportion of people living in urban areas, is associated with poor
health outcomes (see for example, Goryakin et al., 2017; Brueckner,
2019).
For robustness purposes, we also use additional control variables
such as female parliamentarians (%), trade openness (% GDP), official
development assistance (% GNI), CO2 emissions, and ethnic fraction­
alization. Consistent with recent work by Macmillan et al. (2018),
Pickbourn and Ndikumana (2019), and Panda (2020), we expect a
negative effect of female parliamentarians, trade openness, and foreign
aid on child mortality. In contrast, Churchill et al. (2017) and Bou­
choucha (2021) show that ethno-linguistic fractionalization and carbon
emissions have a negative effect on health. Therefore, we expect a
positive effect of CO2 emissions and ethnic fractionalization on infant
mortality.
3.2. Methodology
In an effort to analyze the relationship between natural resources,
governance, and under-five mortality, this paper has two main objec­
tives. First, we examine the direct effect of natural resources on child
mortality in African countries. We hypothesize that natural resources
are positively related to child mortality. Therefore, we investigate the
following dynamic model in Equation (1)
IMR5i,t = α + β1 IMR5i,t− 1 + β2 TNRi,t + γXi,t + μi + νt + εit (1)
Where IMR5i,t , is our dependent variable measured by the under-five
mortality rate of country i in year t. IMR5i,t− 1 represents suitable lags
of the dependent variables. TNRi,t is the total natural resource rents as
percentage of GDP. Xi,t is the vector of control variables and εit denotes
the error term. μi and vt are the country and time specific effects
respectively.
The second objective of this paper is to investigate the role of
governance in the relationship between natural resources and child
mortality. Therefore, we formulate the following interactive model in
Equation (2) as follows:
IMR5i,t = α + β1 IMR5i,t− 1 + β2 TNRi,t + β3 GOVi,t + β4 (TNR × GOV)i,t
(2)
+ γXi,t + μi + νt + εit
Where GOVi,t represents the different governance indicators,
(TNR × GOV)i,t is the interaction between natural resources and the
different governance indicators. We conclude a mitigating effect if β2 >
0 and β4 < 0, with both estimated coefficients being statistically sig­
nificant. The rest of the parameters have the same significance as in
equation (1).
The introduction of the lagged dependent variable as an explanatory
variable leads to a dynamic panel bias (Nickell, 1981). We address this
bias and endogeneity concerns by using the Generalized Method of
Moments (GMM) (Arellano and Bover, 1995; Blundell and Bond, 1998).
The GMM not only addresses the endogeneity problem resulting from
possible reverse causality, measurement errors or omitted variables; it
also takes into account unobserved country-specific effects. In addition,
the GMM solves the autocorrelation problem caused by the inclusion of
the lagged value of the dependent variable as a regressor by differenti­
ating the lagged value of the dependent variable.
By considering the first difference on both sides of equation (2), we
eliminate the fixed effects and the bias related to unobserved hetero­
geneity over time.
ΔIMR5i,t = α + β1 ΔIMR5i,t− 1 + β2 ΔTNRi,t + β3 ΔGOVi,t
(3)
+ β4 Δ(TNR × GOV)i,t + γΔXi,t + Δεit
Equation (3) specifies the GMM in difference introduced by Arellano
and Bond (1991). This method consists of estimating equation (3) using
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Table 1
Principal component analysis (PCA) for governance.
Principal components Eigen value Component matrix (loadings) Proportion Cumulative proportion

VA PS GE RQ CC RL

First PC (G. Gov) 4.695 0.38 0.356 0.425 0.418 0.414 0.444 0.782 0.782
Second PC 0.494 0.1390 0.857 − 0.356 − 0.321 − 0.108 − 0.062 0.082 0.865
First PC (Pol. Gov) 1.607 0.707 0.707 0.803 0.803
Second PC 0.392 − 0.707 0.707 0.196 1.000
First PC (Eco. Gov) 1.870 0.707 0.707 0.935 0.935
Second PC 0.129 0.707 − 0.707 0.064 1.000
First PC (Inst. Gov) 1.865 0.707 0.707 0.933 0.933
Second PC 0.134 0.707 − 0.707 0.067 1.000

P.C: Principal Component. CC: Control of Corruption. GE: Government Effectiveness. PS: Political Stability. R.Q: Regulation Quality. RL: Rule of Law. VA: Voice and
Accountability. G. Gov (General Governance): First PC of VA, PS, RQ, GE, RL and CC. Pol. gov (Political Governance): First PC of VA and PS. Eco. gov (Economic
Governance): First PC of RQ and GE. Inst. gov (Institutional Governance): First PC of RL and CC.

equation (4).
Table 2 [ ] [ ] [ ]
Descriptive statistics. IMR5i,t IMR5i,t− j Xi,t
=θ + + εit (4)
Variables Obs Mean Std. Min Max
ΔIMR5i,t ΔIMR5i,t− j ΔXi,t

Child Mortality Blundell and Bover (1998) show that the system-based GMM gen­
IMR5 1150 97.491 48.361 14.5 265.8 erates more efficient estimates than the difference-based GMM. How­
Female IMR5 1150 91.441 46.480 13 261.5
ever, there is still unobserved heterogeneity in the model. In order to
Male IMR5 1150 103.229 50.223 15.9 269.8
IMR1 1150 62.650 27.231 12.5 157.5 eliminate the remaining unobserved heterogeneity, the model is
IMRNEO 1150 30.749 10.634 5.6 59.2 assumed to have the orthogonality conditions shown in equation (5).
Mortality_pneumonia 900 12.222 7.127 0.7 38.9 This involves the instrumentation of differential equations with delayed
Mortality_diarrhea 900 9.281 6.494 0 39.5 levels and level equations with delayed differences.
Mortality_malaria 900 11.149 13.88 0 71.1
[ ] [ ]
Natural resources E ΔXi,t− 1 (αi εit ) = E ΔIMR5i,t− 1 (αi εit ) = 0 (5)
TNR 1130 12.065 11.901 0.001 84.228
Coal rent 1130 0.108 0.528 0 7.869 For all these reasons, the two-stage GMM is the best estimator for the
Forest rent 1134 5.634 6.182 0 40.426
characteristics of our models.4 In addition, as suggested by Roodman
Mineral rent 1134 1.816 4.504 0 46.624
Gasrent 1104 0.235 0.788 0 7.071 (2009a), we pay close attention to the choice of the number of in­
Oil rent 1130 4.255 10.643 0 78.541 struments in order to reduce the risk of “instrument over identification”
Control variables in the model. The instruments for the first-difference regression are
Income(ln) 1112 7.107 0.988 5.233 9.9297 lagged endogenous level explanatory variables, and the instruments for
Female education 909 93.03 26.266 21.441 151.314
the level regression are lagged endogenous difference explanatory var­
Urban population 1120 40.012 17.104 7.412 89.37
HIV female 1127 3.209 4.837 0.1 24.2 iables. We use lags of order 2 to 4 for these estimates.
Women in parliaments 1029 15.760 11.07 0 63.75 Two main tests are associated with the GMM, namely the model
CO2 emissions 1043 0.944 1.70523 0.016 11.203 over-identification test (Hansen-test), through which we check the val­
Trade openness 1059 70.11 36.845 17.858 347.99
idity of the instruments used, in the sense that they must be correlated
Ethnic fractionalization 1127 0.632 0.244 0 0.930
Foreign aid 1118 8.523 9.170 − 0.250 92.141
with the instrumented variables and not with the error term (Hansen,
Governance variables 1982). We also perform the error autocorrelation test of Arellano and
Control of Corrupt. 1000 − 0.613 0.586 − 1.826 1.216 Bond (1991) (AR(1) and AR(2)). For the first-order autocorrelation test,
Gov. Effectiveness 999 − 0.717 0.588 − 1.884 1.05 the null hypothesis is that there is no first-order correlation of the errors
Political stability 1000 − 0.532 0.855 2.444 1.219
in the first-difference equation. For the second-order autocorrelation
−
Regulatory quality 1000 − 0.655 0.586 − 2.297 1.127
Rule of Law 1000 − 0.667 0.614 − 2.129 1.077 test, the null hypothesis is the presence of second-order correlation of
Voice and Account. 1000 − 0.613 0.717 − 2.226 1.007 the errors in the first difference equation. Thus, we should reject the null
G. Gov 999 0.0000 2.166 − 5.417 5.784 hypothesis of no first order serial correlation (AR(1) test) but not reject
Pol. Gov 1000 0.0000 1.267 3.009 2.883
−
the null hypothesis of no second order serial correlation.
Eco. Gov 999 0.0000 1.367 − 3.374 4.264
Inst. Gov 1000 0.0000 1.365 − 3.020 4.085
4. Results
IMR5, IMR1, IMRNEO and TNR denote under-five mortality, infant mortality,
neonatal mortality and natural resource rent, respectively.G. Gov (General
4.1. Baseline results
Governance), Pol. gov (Political Governance), Eco. gov (Economic Governance),
Inst. gov (Institutional Governance).
Table 3 presents the baseline results. The consistency and coherence
of the two-step GMM estimators depend on the validity of the assump­
the lagged values of the explanatory variables as instruments. Further­
tion of no second order serial correlation of the error terms and the
more, it should be noted that the instruments must be uncorrelated with
validity of the instruments (Hansen test). The results of the diagnostic
the first difference of the error term.
tests show that our models are well specified. The Hansen test does not
However, there are two potential problems with specifying the GMM
in difference. First, first-differencing could amplify the impact of mea­
surement errors on the dependent variables (Griliches and Hausman,
4
1986). Second, the instruments chosen may be weak for first-difference We run our regressions with the Stata 14 software, in which the system
equations (Arellano and Bover, 1995). To alleviate these problems, GMM is pre-programmed (“xtabond2” and “twostep” commands). Also, in
Blundell and Bond (1998) propose using first difference variables as writing the commands related to our estimates, we rely on the recommenda­
tions of Roodman (2009a, 2009b) and Newey and Windmeijer (2009),
instruments for level equations in a system of equations represented by
including the application of the Windmeijer (2005) correction.

5
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Table 3
Baseline results.
Dependent variable: IMR5

(1) (2) (3) (4) (5)

Lagged Mortality_5 0.947*** 0.956*** 0.909*** 0.917*** 0.903***

(0.003) (0.004) (0.012) (0.003) (0.007)
TNR 0.014*** 0.009** 0.038*** 0.041*** 0.070***
(0.003) (0.004) (0.013) (0.009) (0.015)
Income(ln) − 1.800** − 0.364 0.280 − 0.437
(0.740) (0.508) (0.298) (0.555)
Female education − 0.137*** − 0.136*** − 0.161***
(0.026) (0.007) (0.014)
Urban pop. Growth − 0.046*** 0.008
(0.014) (0.029)
HIV female 0.379***
(0.078)
Constant 1.496*** 13.526** 20.251*** 16.786*** 21.905***
(0.233) (5.521) (5.792) (2.204) (3.761)
Observations 1028 1008 838 838 823
Number of countries 50 50 48 48 47
Number of instruments 26 24 23 39 33
AR(1) 0.023 0.032 0.024 0.045 0.041
AR(2) 0.279 0.260 0.296 0.296 0.269
Hansen OIR 0.167 0.238 0.254 0.330 0.419

Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels, respectively. Robust standard errors reported in parenthesis. The coefficients are based on
the two-step GMM system estimation, using the finite sample correction of Windmeijer (2005). All explanatory variables are treated as potentially endogenous. The
lags of the explanatory variables are taken as an instrument for the difference equation, while the first differences of the explanatory variables are taken as an in­
strument for the level equation. IMR5 and TNR denote under-five mortalityand natural resource rent, respectively.

reject the validity of the instruments. Furthermore, the test results
validate the absence of second order serial correlation. Too many in­
struments can seriously weaken and bias Hansen’s test of identification
restrictions and the rule of thumb is therefore that the number of in­
struments should be less than the number of countries (Roodman,
2009b). The two-step system GMM presented in Table 2 generated a
maximum of 39 instruments, which is less than the number of countries,
and the regression results are therefore free of instrument proliferation.
Column (1) of Table 3 presents the results of the relationship be­
tween natural resources and under-five mortality without control vari­
ables. The coefficient associated with resource rent has a positive and
statistically significant sign. This suggests that natural resources are on
average associated with higher under-five mortality rates. When we
control for per capita income in column (2), the sign of the coefficient
associated with natural resources remains positive and statistically sig­
nificant. Furthermore, consistent with Khanam et al. (2009), we show
that income reduces child mortality. We introduce female primary
school enrollment in column (3). Once again, the effect of natural re­
sources remains positive and significant, confirming the resource curse
thesis. We also find that female education reduces child mortality
consistent with Grépin and Bharadwaj (2015). Indeed, more educated
women have better knowledge about preventive and curative measures
to ensure the health of their children. In column (4), we include the
growth rate of the urban population. We again find that natural re­
sources are positively associated with under-five mortality. However,
we find that urbanization increases child mortality. Indeed, the pressure
on health facilities in urban areas, which is characteristic of many
developing regions, leads to increased mortality rates (Arthur and Oai­
khenan, 2017). In column (5), when we consider HIV/AIDS prevalence
among women, the effect of natural resource rent remains positive and
significant. Thus, natural resources, through their negative effects on
institutional quality, democracy, and health expenditure, lead to higher
under-five mortality rates. Similarly, natural resource dependence being
a source of wealth for existing generations increases inequality in the
short and long terms (Sebri and Dachraoui, 2021), which also leads to
higher infant mortality rates. These results agree with those of Wigley
(2017), who shows that oil wealth is positively associated with
under-five mortality. We also show that HIV-positive women are more
likely to infect and cause the deaths of their children.
4.2. Robustness checks
We apply three approaches to test the robustness of the previous
results. First, we introduce five additional control variables, including
women parliamentarians, CO2 emissions, trade openness, ethnic frac­
tionalization, and foreign aid. We summarize the estimation results in
Table 4. From column (1) to column (5), when we introduce each con­
trol variable individually, the coefficient associated with natural re­
sources remains positive and statistically significant. In column (6),
when we introduce the five additional control variables into the same
regression, the coefficient associated with resource rent remains positive
and statistically significant. We can conclude that our result is robust to
the inclusion of additional control variables. Regarding additional
control variables, in column (1), we find that female parliamentarians
reduce child mortality. This result is similar to the work of Macmillan
et al. (2018) showing that female parliamentarians are negatively
associated with child mortality in developing countries. Similarly, we
find that trade openness and foreign aid reduce child mortality. This
result is similar to the findings of Mishra and Newhouse (2009), sug­
gesting that aid allocated to health reduces child mortality in recipient
countries. This result is also similar to those of Jawadi et al. (2018), who
show that trade openness has a positive effect on health in the MENA
region, as it reduces infant mortality rates and increases life expectancy
for men and women.
Second, we estimate our model using alternative dependent vari­
ables. We summarize the results in Table 5. The first-two columns
summarize the results of estimating the effect of natural resources on
under-five mortality by gender. We find that natural resource rents have
a positive and significant effect on child mortality by gender, with a
larger effect on male mortality. Then, in columns (3) and (4), we esti­
mate the effect of resource rents on child mortality by age by dis­
tinguishing between neonatal mortality (column 3) and infant mortality
(column 4). Once again, we find that resource rents have a positive effect
on child mortality by age, with the effect being larger for children aged
0–1 year. Finally, we are interested in the effect of natural resources on
child mortality from infectious diseases such as pneumonia, diarrhea,
and malaria. We find that natural resources also have a positive and
significant effect on deaths from infectious diseases. Specifically, the
effect is more detrimental on mortality due to pneumonia, mortality due

6
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Table 4
Robustness to additional control variables.
Dependent variable: IMR5

(1) (2) (3) (4) (5) (6)

Lagged IMR5 0.896*** 0.892*** 0.898*** 0.902*** 0.910*** 0.945***

(0.006) (0.013) (0.007) (0.008) (0.004) (0.009)
TNR 0.075*** 0.049*** 0.046*** 0.054*** 0.051*** 0.063***
(0.012) (0.018) (0.017) (0.021) (0.010) (0.016)
Baseline control Yes Yes Yes Yes Yes Yes
Women in parliaments − 0.049** − 0.044***
(0.023) (0.017)
CO2 emissions v 0.508** 0.050
(0.245) (0.264)
Trade openness − 0.016* − 0.015
(0.009) (0.016)
Ethnic fractionalization 2.060 − 3.276
(1.359) (2.502)
Foreign aid − 0.139*** − 0.130***
(0.014) (0.025)
Constant 28.182*** 35.151*** 29.284*** 22.284*** 28.439*** 28.258***
(2.825) (7.802) (3.652) (5.544) (2.333) (5.803)
Observations 779 734 794 794 807 679
Number of countries 47 47 47 47 47 47
Number of instruments 38 25 31 31 45 35
AR(1) 0.031 0.035 0.046 0.044 0.031 0.002
AR(2) 0.240 0.354 0.260 0.286 0.277 0.311
Hansen OIR 0.549 0.320 0.480 0.171 0.811 0.760

Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels respectively. Robust standard errors reported in parenthesis. IMR5 and TNR denote under-
five mortality and natural resource rent, respectively.

Table 5
Robustness to alternative dependent variables.
Dependent variables: Mortality by gender Mortality by age Mortality by causes

Female IMR5 Male IMR5 IMRNEO IMR1 Mortality_ pneumonia Mortality_ Mortality_

diarrhea malaria

(1) (2) (3) (4) (5) (6) (7)

Lagged dependent variables 0.900*** 0.899*** 0.971*** 0.865*** 0.800*** 0.839*** 0.797***
(0.008) (0.007) (0.023) (0.043) (0.007) (0.003) (0.093)
TNR 0.025** 0.028*** 0.018** 0.062** 0.014*** 0.005** 0.003**
(0.011) (0.010) (0.007) (0.028) (0.002) (0.002) (0.001)
Baseline control Yes Yes Yes Yes Yes Yes Yes
Constant 32.758*** 35.979*** 3.347* 23.482** 9.954*** 4.838*** 1.818
(3.309) (3.339) (2.017) (10.518) (0.871) (0.398) (1.325)
Observations 823 823 758 823 609 567 567
Number of countries 47 47 47 47 47 46 46
Number of instruments 33 43 40 39 35 40 39
AR(1) 0.046 0.041 0.021 0.013 0.008 0.018 0.039
AR(2) 0.282 0.319 0.123 0.641 0.702 0.509 0.768
Hansen OIR 0.576 0.588 0.196 0.308 0.395 0.523 0.407

Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels respectively. Robust standard errors reported in parenthesis. IMR1, IMRNEO and TNR
denote infant mortality, neonatal mortality and natural resource rent, respectively.

to diarrhea, and mortality due to malaria.
In our empirical estimation, we have so far used annual data, and our
regression results may be affected by business cycle effects. Common
practice is to take 3- or 5-year averages to filter out business cycle in­
fluences. As discussed in Islam and McGillivray (2020), the longer the
period over which the averages are taken, the less the estimates are
affected by business cycles and transitory dynamics. This leads us to
conduct a third robustness test using data averaged over three years. The
advantage of using 3-year averages of data is that it limits the prolifer­
ation of instruments and mitigates short-term disturbances that can be
significant (Njangang et al., 2022).5 The results are presented in Table 6,
where we have estimated the effect of natural resources on under-five
5
Averages are taken over 7-year periods (1996–1998, 1999–2001,
2002–2004, 2005–2007, 2008–2010, 2011–2013, and 2014–2016), then we
take a final 2-year period (2017–2018).
mortality, child mortality by age, gender, and causes. We find that the
coefficient associated with total natural resource rent remains positive
and statistically significant in all models. Our results are not affected by
the short-run business cycle and thus remain robust to a change in the
data structure.
4.3. Do different types of natural resources have different effects?
Several authors argue that the type of resources matters in the val­
idity of the resource curse hypothesis (Isham et al., 2005; Bulte et al.,
2005). The literature distinguishes between point resources and diffuse
resources based on the concentration of production patterns and in­
come. Point resources are extracted from a narrow economic and
geographic base and entail oil and minerals (Bulte et al., 2005). The fact
that these resources are concentrated spatially is an indication that they
can be controlled and protected at a relatively low cost. The diffuse

7
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Table 6
Robustness to alternative data structure.
IMR5 Female IMR5 Male IMR5 IMRNEO IMR1 Mortality_ pneumonia Mortality_ Mortality_

diarrhea malaria

(1) (2) (3) (4) (5) (6) (7) (8)

Lagged dependent variables 0.900*** 0.895*** 0.905*** 0.977*** 0.929*** 0.712*** 0.656*** 0.569***
(0.007) (0.008) (0.007) (0.006) (0.008) (0.037) (0.032) (0.012)
TNR 0.047*** 0.052*** 0.039** 0.013*** 0.019* 0.031** 0.027*** 0.008***
(0.016) (0.017) (0.016) (0.005) (0.011) (0.012) (0.010) (0.001)
Baseline control Yes Yes Yes Yes Yes Yes Yes Yes
Constant − 4.692* − 4.440* − 5.360** − 3.348*** − 5.442*** 7.972** − 0.903 1.952***
(2.404) (2.571) (2.417) (0.678) (1.575) (3.404) (1.454) (0.136)
Observations 233 233 233 233 233 59 90 168
Number of countries 47 47 47 47 47 17 18 47
Number of instruments 37 37 37 37 37 26 32 42
AR(1) 0.059 0.062 0.060 0.001 0.007 0.030 0.048 0.093
AR(2) 0.118 0.112 0.135 0.502 0.391 0.289 0.622 0.244
Hansen OIR 0.248 0.262 0.235 0.403 0.222 0.915 0.989 0.276

Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels respectively. Robust standard errors reported in parenthesis. IMR5, IMR1, IMRNEO and
TNR denote under-five mortality, infant mortality, neonatal mortality and natural resource rent, respectively.

resource are rather scattered in space and have a greater understanding between natural resources and under-five mortality. In this perspective,
of agricultural or forest products. first, we assume that individual governance quality moderates the
Table 7 summarizes the results of estimating the effect of different relationship between natural resources and under-five mortality. Sec­
resource rents on under-five mortality. We find that point resources ond, we assume that, composite governance also mitigates the positive
(such as minerals, natural gas, and oil) have a positive and significant effect of total resource rents on under-five mortality. We put the first
effect on under-five mortality. Specifically, oil rent has the largest effect, claim to the test by estimating models that include each governance
followed by natural gas rent. However, we find that forest rent has a indicator proposed by WGI, as well as a term that represents its inter­
weak negative effect on under-five mortality. According to the litera­ action with total rents.
ture, these results suggest that dependence on point resources, through We summarize our estimation results in Table 8. We find that the
their deleterious effects on the quality of institutions and the risk of coefficients of the interaction terms between natural resources and four
conflicts, leads to poor economic and social performance, whereas indicators of the quality of governance are negative and significant.
dependence on diffuse resources does not. Specifically, we find that control of corruption, government effective­
ness, political stability, and voice and accountability moderate the
positive effect of natural resources on under-five mortality. Thus, the
4.4. Can governance lift the curse? effect of natural resources on child mortality is conditioned by the
quality of governance. In other words, better control of corruption,
We analyze the extent to which governance shapes the relationship improved government effectiveness, coupled with political stability and
accountability, mitigate, the resource curse for child health. Consistent
Table 7 with contemporary threshold literature based on interactive regressions
Effects of different resource types on under five mortality. (Tchamyou and Asongu, 2017; Asongu et al., 2018; Tchamyou, 2019;
Dependent variable: IMR5
Njangang et al., 2022), we compute the corresponding governance
policy thresholds that should be attained in order to reverse the positive
(1) (2) (3) (4) (5)
effects of natural resources on child mortality. These thresholds are
Lagged IMR5 0.959*** 0.936*** 0.906*** 0.922*** 0.886*** obtained by dividing the unconditional effect of the total rents by the
(0.039) (0.009) (0.024) (0.021) (0.009)
absolute value of the associated conditional or interactive effect. Fig. 2
Coal rent − 0.215
(0.683) shows the marginal effects of resource rents on under-five mortality
Forest rent − 0.106** conditional on individual indicators of governance quality. The mar­
(0.049) ginal effect of natural resources is negative when corruption control,
Mineral rent 0.118* government effectiveness, political stability, and accountability are
(0.065)
above a certain threshold.
Gas rent 2.972**
(1.400) We test the second assertion by examining the role of aggregated
Oil rent 0.067*** governance indicators and distinguishing between general governance,
(0.020) political governance, economic governance, and institutional gover­
Baseline Yes Yes Yes Yes Yes
nance. This approach provides precise information on the extent to
control
Constant 2.827 16.528*** 23.696** 29.814*** 29.292*** which governance transforms natural resources to reduce child mor­
(15.601) (5.888) (11.633) (10.724) (3.658) tality. The results summarized in Table 9 show that the coefficient
Observations 794 828 827 778 823 associated with the interaction variable between natural resources and
Number of 47 47 47 47 47 general governance is negative and significant. This suggests that
countries
improved general governance partially mitigates the positive effect of
Number of 34 28 44 18 34
instruments natural resources on under-five mortality. Regarding the different forms
AR(1) 0.240 0.236 0.251 0.300 0.250 of governance, we find that political governance, economic and insti­
AR(2) 0.268 0.289 0.282 0.276 0.326 tutional governance attenuates the effect of resource rents on child
Hansen OIR 0.189 0.164 0.847 0.563 0.497
mortality. Consistent with the narrative in Table 8, the corresponding
Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels composite governance thresholds needed to overturn the effect of nat­
respectively. Robust standard errors reported in parenthesis. IMR5 denotes ural resources are also provided. Similarly, in Fig. 3, we see that the
under-five mortality.

8
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Table 8
The role of individual governance quality.
Dependent variable: IMR5

(1) (2) (3) (4) (5) (6)

Lagged IMR5 0.860*** 0.874*** 0.908*** 0.850*** 0.887*** 0.882***

(0.005) (0.007) (0.004) (0.003) (0.008) (0.007)
TNR 0.021*** 0.020** 0.031*** 0.035*** 0.083*** 0.015*
(0.007) (0.009) (0.007) (0.011) (0.026) (0.008)
Control of Corrupt. − 1.914***
(0.118)
TNR × Control of Corrupt. − 0.039***
(0.011)
Gov. Effectiveness − 0.724***
(0.236)
TNR × Gov. Effectiveness − 0.022***
(0.007)
Political Stability − 0.148
(0.215)
TNR × Political Stability − 0.041***
(0.011)
Regulatory Quality − 0.450***
(0.148)
TNR × Regulatory Quality − 0.039***
(0.012)
Rule of Law − 0.783
(0.505)
TNR × Rule of Law 0.026
(0.019)
Voice and Account. 0.150
(0.151)
TNR × Voice and Account. − 0.016**
(0.007)
Baseline control Yes Yes Yes Yes Yes Yes
Constant 36.073*** 31.174*** 12.009*** 35.775*** 31.728*** 14.480***
(2.172) (2.498) (2.075) (2.375) (3.600) (3.139)
Individual governance thresholds 0.538 0.909 0.756 0.897 na 0.937
Observations 594 594 675 594 594 594
Number of countries 47 47 47 47 47 47
Number of instruments 41 40 46 30 38 40
AR(1) 0.081 0.005 0.000 0.072 0.011 0.058
AR(2) 0.521 0.580 0.356 0.556 0.680 0.532
Hansen OIR 0.286 0.615 0.525 0.320 0.192 0.220

Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels respectively. Robust standard errors reported in parenthesis. na: not applicable because at
least one estimated coefficient needed for the computation of thresholds is not significant. IMR5 and TNR denote under-five mortalityand natural resource rent,
respectively.

marginal effect of natural resources is negative when overall gover­
nance, political governance, economic governance, and institutional
governance are above a certain threshold.
5. Concluding implications and future research directions
Abundant literature has shown a negative relationship between
natural resources and economic growth, known as the resource curse. In
recent years, several studies have extended the resource curse hypoth­
esis to quality of life, including health outcomes. To contribute to this
emerging literature, this paper has studied the effect of natural resources
on child mortality in 50 African countries for the period 1996–2018. The
results show that total resource rent has a positive effect on under-five
mortality. We also find that natural resource dependence has a posi­
tive effect on under-five mortality by age, gender, and the three major
causes of child mortality from infectious diseases, namely, mortality
from pneumonia, diarrhea, and malaria. Looking at the types of re­
sources, we find that the resource curse is attributed to point resources
(such as oil, natural gas, and mineral rents), as opposed to diffuse re­
sources (such as forest rents), which have a negative effect on child
mortality. Finally, we provide evidence that improving the quality of
governance, particularly political and institutional governance, miti­
gates the resource curse. Corresponding governance policy thresholds
that should be attained in order to reverse the unfavorable effects of
resources on child mortality are provided.
The adverse effect of natural resources on child survival is likely due
to the fact that the ruling elites have little incentive to invest in and take
effective actions to improve child health. However, this trend can be
broken if strong institutions ensure that political leaders are more
accountable to the people. Thus, there is an urgent need for African
countries to continue with efforts towards improving governance. Some
countries have already begun this process, but much remains to be done.
In addition, in order to achieve the SDG target of reducing under-five
mortality to less than 25 per 1000 births by 2030, it is important that
African leaders allocate a fraction of their resources to the health sector
in order to subsidize the treatment and prevention of common childhood
diseases.
The importance of political and institutional governance in moder­
ating the positive effects of natural resources on child mortality has two
main direct policy implications: (i) the procedures by which political
leaders are elected and replaced should be improved in sampled coun­
tries. Such is the definition of political governance as conceptualized,
measured, and applied in this study in the light of extant literature
(Asongu and Odhiambo, 2019; Tchamyou, 2021). (ii) By the same vein
(Asongu and Odhiambo, 2021), procedures by which the state and cit­
izens respect institutions that govern interactions between them (i.e.
institutional governance) should improve.
In light of the established importance of governance dynamics in
moderating the unfavorable incidence of natural resources on child
mortality, the corresponding governance policy thresholds that have

9
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Fig. 2. Average marginal effects of natural resources rent with individuals measures of governance.

been provided are indicative of the actionable measures that sampled

Table 9
countries should take in order to enhance suggested governance mea­
The role of composite indicators of governance quality.
sures to levels at which the natural resource curse with respect to child
Dependent variable: IMR5 mortality is no longer apparent. In other words, at the established
(1) (2) (3) (4) governance thresholds, governance completely dampens the effect of
Lagged IMR5 0.849*** 0.843*** 0.862*** 0.853***
natural resources in order to engender overall negative impacts on child
(0.004) (0.008) (0.008) (0.009) mortality dynamics.
TNR 0.033*** 0.022*** 0.033** 0.032** This study obviously leaves room for improving extant knowledge in
(0.006) (0.005) (0.014) (0.013) this area of economic development, especially as it relates to examining
G. Gov − 0.243***
whether these findings withstand empirical scrutiny within country-
(0.073)
TNR × G. Gov − 0.017*** specific remits. Accordingly, engaging country-specific studies is rele­
(0.004) vant for country-specific implications and, by extension, addresses a
Pol. Gov − 0.899*** shortcoming in the adopted estimation strategy which fails to account
(0.083) for country-specific effects in order to avoid concerns of endogeneity
TNR × Pol. Gov − 0.014*
(0.008)
pertaining to the correlation between the lagged outcome variable and
Eco. Gov − 0.077 the error term. In the same vein, it would also be very interesting if
(0.201) future research on the relationship between natural/mineral resources
TNR × Eco. Gov − 0.027*** and health (child and maternal) could be examined at the micro level.
(0.008)
Indeed, there are few studies that explore the resource curse at the micro
Inst. Gov − 0.548***
(0.140) level, particularly in Africa. This opens the door to examining whether
TNR × Inst. Gov − 0.025*** revenues from natural resource exploitation can improve access to
(0.009) health services and health outcomes in specific countries and to drawing
Baseline control Yes Yes Yes Yes policy implications.
Constant 35.494*** 34.271*** 40.958*** 37.700***
(1.233) (3.365) (4.898) (4.041)
Composite governance 1.941 1.571 1.222 1.28 Authors’ statement
thresholds
Observations 594 645 594 556
Number of countries 47 47 47 47
Sosson Tadadjeu: conceptualisation, data analysis, writing of paper
Number of instruments 40 40 39 42 and proofreading of final manuscript. Henri Njangang: conceptualisa­
AR(1) 0.064 0.024 0.092 0.005 tion, data analysis, writing of paper and proofreading of final manu­
AR(2) 0.493 0.627 0.612 0.627 script. Simplice A. Asongu: conceptualisation, data analysis, writing of
Hansen OIR 0.301 0.222 0.412 0.257
paper and proofreading of final manuscript. Brice Kamguia: con­
Notes: *, **, *** denote statistical significance at the 10%, 5% and 1% levels ceptualisation, data analysis, writing of paper and proofreading of final
respectively. Robust standard errors reported in parenthesis. na: not applicable manuscript.
because at least one estimated coefficient needed for the computation of
thresholds is not significant. IMR5 and TNR denote under-five mortalityand
natural resource rent, respectively. G. Gov (General Governance), Pol. gov Declaration of competing interest
(Political Governance), Eco. gov (Economic Governance), Inst. gov (Institutional
Governance).
The authors have no conflict of interest.

10
S. Tadadjeu et al. Resources Policy 83 (2023) 103702

Fig. 3. Average marginal effects of natural resources rent with composites measures of governance.

Data availability

Data will be made available on request.

Appendix
Table A1
List of the countries

Algeria Comoros Gabon Mali Senegal

Angola Congo, Dem, Rep, Gambia, The Mauritania Sierra Leone

Benin Congo, Rep, Ghana Mauritius South Africa
Botswana Cote d’Ivoire Guinea Morocco Sudan
Burkina Faso Djibouti Guinea-Bissau Mozambique Tanzania
Burundi Egypt, Arab Rep, Kenya Namibia Togo
Cabo Verde Equatorial Guinea Lesotho Niger Tunisia
Cameroon Eritrea Liberia Nigeria Uganda
Central African Republic Eswatini Madagascar Rwanda Zambia
Chad Ethiopia Malawi Sao Tome and Principe Zimbabwe

References Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of
error-components models. J. Econom. 68 (1), 29–51.
Arezki, R., Brückner, M., 2011. Oil rents, corruption, and state stability: evidence from
Achim, M.V., Văidean, V.L., Borlea, S.N., 2019. Corruption and health outcomes within
panel data regressions. Eur. Econ. Rev. 55 (7), 955–963.
an economic and cultural framework. Eur. J. Health Econ. 1–13.
Arthur, E., Oaikhenan, H.E., 2017. The effects of health expenditure on health outcomes
Adetunji, J., 2000. Trends in under-5 mortality rates and the HIV/AIDS epidemic. Bull.
in Sub-Saharan Africa (SSA). Afr. Dev. Rev. 29 (3), 524–536.
World Health Organ. 78, 1200–1206.
Asongu, S.A., Nwachukwu, J.C., 2016. The mobile phone in the diffusion of knowledge
Adika, G., 2020. Economic growth dynamics between resource-rich and resource-poor
for institutional quality in sub-Saharan Africa. World Dev. 86, 133–147.
countries in sub-Saharan Africa: the role of politics and institutions. Afr. Dev. Rev. 32
Asongu, S.A., Nwachukwu, J.C., 2017. The impact of terrorism on governance in African
(3), 303–315.
countries. World Dev. 99, 253–270.
Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R., 2003.
Asongu, S.A., Odhiambo, N.M., 2019. Governance and social media in African countries:
Fractionalization. J. Econ. Growth 8 (2), 155–194.
an empirical investigation. Telecommun. Pol. 43 (5), 411–425.
Andriano, L., Monden, C.W., 2019. The causal effect of maternal education on child
Asongu, S.A., Odhiambo, N.M., 2021. Enhancing governance for environmental
mortality: evidence from a quasi-experiment in Malawi and Uganda. Demography 56
sustainability in sub-Saharan Africa. Energy Explor. Exploit. 39 (1), 444–463.
(5), 1765–1790.
Asongu, S.A., Le Roux, S., Biekpe, N., 2018. Enhancing ICT for environmental
Anyanwu, J.C., Erhijakpor, A.E., 2009. Health expenditures and health outcomes in
sustainability in sub-Saharan Africa. Technol. Forecast. Soc. Change 127, 209–216.
Africa. Afr. Dev. Rev. 21 (2), 400–433.
Azizi, S., 2018. The impacts of workers’ remittances on human capital and labor supply
Anyanwu, J.C., Erhijakpor, A.E., 2014. Does oil wealth affect democracy in Africa? Afr.
in developing countries. Econ. Modell. 75, 377–396.
Dev. Rev. 26 (1), 15–37.
Badeeb, R.A., Lean, H.H., Clark, J., 2017. The evolution of the natural resource curse
Arellano, M., Bond, S., 1991. Some tests of specification for panel data: Monte Carlo
thesis: a critical literature survey. Resour. Pol. 51, 123–134.
evidence and an application to employment equations. Rev. Econ. Stud. 58 (2),
Basedau, M., Lay, J., 2009. Resource curse or rentier peace? The ambiguous effects of oil
277–297.
wealth and oil dependence on violent conflict. J. Peace Res. 46 (6), 757–776.

11
S. Tadadjeu et al.
Bellinger, N., Fails, M.D., 2020. When does oil harm child mortality? Polit. Res. Q.,
1065912920931201
Bergougui, B., Murshed, S.M., 2020. New evidence on the oil-democracy nexus utilising
the Varieties of Democracy data. Resour. Pol. 69, 101905.
Bernet, P.M., Gumus, G., Vishwasrao, S., 2018. Effectiveness of public health spending on
infant mortality in Florida, 2001–2014. Soc. Sci. Med. 211, 31–38.
Besley, T., Kudamatsu, M., 2006. Health and democracy. Am. Econ. Rev. 96 (2),
313–318.
Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel
data models. J. Econom. 87 (1), 115–143.
Bouchoucha, N., 2021. The effect of environmental degradation on health status: do
institutions matter? Journal of the Knowledge Economy 12 (4), 1618–1634.
Brueckner, M., 2019. Adult mortality and urbanization: examination of a weak
connection in sub-Saharan Africa. World Dev. 122, 184–198.
Brunnschweiler, C.N., Bulte, E.H., 2008. The resource curse revisited and revised: a tale
of paradoxes and red herrings. J. Environ. Econ. Manag. 55 (3), 248–264.
Buccellato, T., Mickiewicz, T., 2009. Oil and gas: a blessing for the few. Hydrocarbons
and inequality within regions in Russia. Eur. Asia Stud. 61 (3), 385–407.
Bulte, E.H., Damania, R., Deacon, R.T., 2005. Resource intensity, institutions, and
development. World Dev. 33 (7), 1029–1044.
Carmignani, F., 2013. Development outcomes, resource abundance, and the transmission
through inequality. Resour. Energy Econ. 35 (3), 412–428.
Chang, W.Y., 2020. Democracy, natural resources, and infectious diseases: the case of
malaria, 1990–2016. Stud. Comp. Int. Dev. 55 (3), 354–380.
Chang, W.Y., Wei, D., 2019. Natural resources and infectious diseases: the case of
malaria, 2000–2014. Soc. Sci. J. 56 (3), 324–336.
Chireshe, J., Ocran, M.K., 2020. Health care expenditure and health outcomes in sub-
Saharan African countries. Afr. Dev. Rev. 32 (3), 349–361.
Churchill, A.S., Ocloo, J.E., Siawor-Robertson, D., 2017. Ethnic diversity and health
outcomes. Soc. Indicat. Res. 134 (3), 1077–1112.
Cockx, L., Francken, N., 2014. Extending the concept of the resource curse: natural
resources and public spending on health. Ecol. Econ. 108, 136–149.
Cockx, L., Francken, N., 2016. Natural resources: a curse on education spending? Energy
Pol. 92, 394–408.
Collier, P., Hoeffler, A., 2004. Greed and grievance in civil war. Oxf. Econ. Pap. 56 (4),
563–595.
Cotet, A.M., Tsui, K.K., 2013. Oil, growth, and health: what does the cross-country
evidence really show? Scand. J. Econ. 115 (4), 1107–1137.
Dauvin, M., Guerreiro, D., 2017. The paradox of plenty: a meta-analysis. World Dev. 94,
212–231.
De Soysa, I., Gizelis, T.I., 2013. The natural resource curse and the spread of HIV/AIDS,
1990–2008. Soc. Sci. Med. 77, 90–96.
Dincer, O., Teoman, O., 2019. Does corruption kill? Evidence from half a century infant
mortality data. Soc. Sci. Med. 232, 332–339.
Edwards, R.B., 2016. Mining away the Preston curve. World Dev. 78, 22–36.
El Anshasy, A.A., Katsaiti, M.S., 2015. Are natural resources bad for health? Health Place
32, 29–42.
Epo, B.N., Faha, D.R.N., 2020. Natural resources, institutional quality, and economic
growth: an African tale. Eur. J. Dev. Res. 32 (1), 99–128.
Frankel, J.A., 2010. Monetary Policy in Emerging Markets: a Survey (No. W16125).
National Bureau of Economic Research.
Fum, R.M., Hodler, R., 2010. Natural resources and income inequality: the role of ethnic
divisions. Econ. Lett. 107 (3), 360–363.
Gates, S., Hegre, H., Nygård, H.M., Strand, H., 2012. Development consequences of
armed conflict. World Dev. 40 (9), 1713–1722.
Goryakin, Y., Rocco, L., Suhrcke, M., 2017. The contribution of urbanization to non-
communicable diseases: evidence from 173 countries from 1980 to 2008. Econ.
Hum. Biol. 26, 151–163.
Grépin, K.A., Bharadwaj, P., 2015. Maternal education and child mortality in Zimbabwe.
J. Health Econ. 44, 97–117.
Griliches, Z., Hausman, J.A., 1986. Errors in variables in panel data. J. Econom. 31 (1),
93–118.
Gylfason, T., 2001. Natural resources, education, and economic development. Eur. Econ.
Rev. 45 (4–6), 847–859.
Hansen, L.P., 1982. Large sample properties of generalized method of moments
estimators. Econometrica 50 (4), 1029–1054.
IHME, 2019. Institute of Health Metrics and Evaluation. https://vizhub.healthdata.
org/child-mortality.
Isham, J., Woolcock, M., Pritchett, L., Busby, G., 2005. The varieties of resource
experience: natural resource export structures and the political economy of
economic growth. World Bank Econ. Rev. 19 (2), 141–174.
Islam, M.R., McGillivray, M., 2020. Wealth inequality, governance and economic
growth. Econ. Modell. 88, 1–13.
Jawadi, F., El Gouddi, S., Ftiti, Z., Kacem, A., 2018. Assessing the effect of trade openness
on health in the MENA region: a panel data analysis. Open Econ. Rev. 29, 469–479.
Khanam, R., Nghiem, H.S., Connelly, L.B., 2009. Child health and the income gradient:
evidence from Australia. J. Health Econ. 28 (4), 805–817.
12
Resources Policy 83 (2023) 103702
Kim, D.H., Lin, S.C., 2017. Human capital and natural resource dependence. Struct.
Change Econ. Dynam. 40, 92–102.
Li, Q., An, L., Xu, J., Baliamoune-Lutz, M., 2018. Corruption costs lives: evidence from a
cross-country study. Eur. J. Health Econ. 19 (1), 153–165.
Liu, L., Oza, S., Hogan, D., Perin, J., Rudan, I., Lawn, J.E., et al., 2015. Global, regional,
and national causes of child mortality in 2000–13, with projections to inform post-
2015 priorities: an updated systematic analysis. Lancet 385 (9966), 430–440.
Macmillan, R., Shofia, N., Sigle, W., 2018. Gender and the politics of death: female
representation, political and developmental context, and population health in a
cross-national panel. Demography 55 (5), 1905–1934.
Madreimov, T., Li, L., 2019. Natural-resource dependence and life expectancy: a
nonlinear relationship. Sustain. Dev. 27 (4), 681–691.
Mehlum, H., Moene, K., Torvik, R., 2006. Institutions and the resource curse. Econ. J.
116 (508), 1–20.
Mishra, P., Newhouse, D., 2009. Does health aid matter? J. Health Econ. 28 (4),
855–872.
Nchofoung, T.N., Achuo, E.D., Asongu, S.A., 2021. Resource rents and inclusive human
development in developing countries. Resour. Pol. 74, 102382.
Neudorfer, N.S., 2018. Commodities and corruption–How the middle class and
democratic institutions lead to less corruption in resource-rich countries. Resour.
Pol. 58, 175–191.
Newey, W.K., Windmeijer, F., 2009. Generalized method of moments with many weak
moment conditions. Econometrica 77 (3), 687–719.
Nickell, S., 1981. Biases in dynamic models with fixed effects. Econometrica. J. Econom.
Soc. 1417–1426.
Njangang, H., Asongu, S.A., Tadadjeu, S., Nounamo, Y., Kamguia, B., 2022. Governance
in mitigating the effect of oil wealth on wealth inequality: a cross-country analysis of
policy thresholds. Resour. Pol. 76, 102561.
Oduyemi, G.O., Owoeye, T., Adekoya, O.B., 2021. Health outcomes and the resource
curse paradox: the experience of African oil-rich countries. Resour. Pol. 73, 102201.
Okada, K., Samreth, S., 2017. Corruption and natural resource rents: evidence from
quantile regression. Appl. Econ. Lett. 24 (20), 1490–1493.
Panda, P., 2020. Does trade reduce infant mortality? Evidence from sub-Saharan Africa.
World Dev. 128, 104851.
Papyrakis, E., Gerlagh, R., 2004. The resource curse hypothesis and its transmission
channels. J. Comp. Econ. 32 (1), 181–193.
Papyrakis, E., Gerlagh, R., 2007. Resource abundance and economic growth in the
United States. Eur. Econ. Rev. 51 (4), 1011–1039.
Pickbourn, L., Ndikumana, L., 2019. Does health aid reduce infant and child mortality
from diarrhoea in sub-saharan Africa? J. Dev. Stud. 55 (10), 2212–2231.
Robinson, J.A., Torvik, R., Verdier, T., 2006. Political foundations of the resource curse.
J. Dev. Econ. 79 (2), 447–468.
Roodman, D., 2009a. A note on the theme of too many instruments. Oxf. Bull. Econ. Stat.
71 (1), 135–158.
Roodman, D., 2009b. How to do xtabond2: an introduction to difference and system
GMM. Stata. Stata Journal 9 (1), 86–136.
Ross, M.L., 2015. What have we learned about the resource curse? Annu. Rev. Polit. Sci.
18, 239–259.
Sachs, J.D., Warner, A.M., 1995. Natural Resource Abundance and Economic Growth (No.
W5398). National Bureau of Economic Research.
Sachs, J.D., Warner, A.M., 1999. The big push, natural resource booms and growth.
J. Dev. Econ. 59 (1), 43–76.
Satti, S.L., Farooq, A., Loganathan, N., Shahbaz, M., 2014. Empirical evidence on the
resource curse hypothesis in oil abundant economy. Econ. Modell. 42, 421–429.
Sebri, M., Dachraoui, H., 2021. Natural resources and income inequality: a meta-analytic
review. Resour. Pol. 74, 102315.
Tadadjeu, S., Njangang, H., Asongu, S., Nounamo, Y., 2021. Natural resources and
wealth inequality: a cross-country analysis. Journal of Economic and Administrative
Sciences. https://doi.org/10.1108/JEAS-05-2021-0099.
Tadadjeu, S., Njangang, H., Ningaye, P., Nourou, M., 2020. Linking natural resource
dependence and access to water and sanitation in African countries. Resour. Pol. 69,
101880.
Tchamyou, V.S., 2019. The role of information sharing in modulating the effect of
financial access on inequality. J. Afr. Bus. 20 (3), 317–338.
Tchamyou, V.S., 2021. Financial access, governance and the persistence of inequality in
Africa: mechanisms and policy instruments. J. Publ. Aff. https://doi.org/10.1002/
pa.2201.
Tchamyou, V.S., Asongu, S.A., 2017. Information sharing and financial sector
development in Africa. J. Afr. Bus. 18 (7), 24–49.
Tiba, S., Frikha, M., 2020. Africa is rich, Africans are poor! A blessing or curse: an
application of cointegration techniques. Journal of the knowledge economy 11 (1),
114–139.
Turan, T., Yanıkkaya, H., 2020. Natural resource rents and capital accumulation nexus:
do resource rents raise public human and physical capital expenditures? Environ.
Econ. Pol. Stud. 22 (3), 449–466.
UN IGME, 2019. United Nations Inter-agency Group for Child Mortality Estimation (Data
retrieved from UNICEF).
S. Tadadjeu et al.
Van der Ploeg, F., 2011. Natural resources: curse or blessing? J. Econ. Lit. 49 (2),
366–420.
Vu, T.V., 2020. Economic complexity and health outcomes: a global perspective. Soc. Sci.
Med. 265, 113480.
Welander, A., Lyttkens, C.H., Nilsson, T., 2015. Globalization, democracy, and child
health in developing countries. Soc. Sci. Med. 136, 52–63.
Welsch, H., 2008. Resource abundance and internal armed conflict: types of natural
resources and the incidence of ‘new wars. Ecol. Econ. 67 (3), 503–513.
Wigley, S., 2017. The resource curse and child mortality, 1961–2011. Soc. Sci. Med. 176,
142–148.
Wigley, S., Akkoyunlu-Wigley, A., 2017. The impact of democracy and media freedom on
under-5 mortality, 1961–2011. Soc. Sci. Med. 190, 237–246.
13
Resources Policy 83 (2023) 103702
Wilkinson, R.G., Pickett, K.E., 2006. Income inequality and population health: a review
and explanation of the evidence. Soc. Sci. Med. 62 (7), 1768–1784.
Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient two-
step GMM estimators. J. Econom. 126 (1), 25–51.
World Bank, 2020. World Development Indicators. https://databank.worldbank.org/
source/world-development-indicators.
World Health Organisation (WHO), 2020. Global Health Observatory (GHO) Data
Repository. https://apps.who.int/gho/data/node.main.COCD?lang=en.
Yilanci, V., Aslan, M., Ozgur, O., 2021. Disaggregated analysis of the curse of natural
resources in most natural resource-abundant countries. Resour. Pol. 71, 102017.