Munyanyi

Essays on the Allocation and Efficacy of Foreign Aid at the
Sub-National Level
A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy
Musharavati Ephraim Munyanyi

MA. Economics – Istanbul Medeniyet University, Turkey
BCom. (Hons) Economics – Midlands State University, Zimbabwe
School of Economics, Finance and Marketing

College of Business and Law
RMIT University
April 2022
Declaration
I certify that except where due acknowledgement has been made, this research is that
of the author alone; the content of this research submission is the result of work which
has been carried out since the official commencement date of the approved research
program; any editorial work, paid or unpaid, carried out by a third party is
acknowledged; and, ethics procedures and guidelines have been followed.
In addition, I certify that this submission contains no material previously submitted for
award of any qualification at any other university or institution, unless approved for a
joint-award with another institution, and acknowledge that no part of this work will, in
the future, be used in a submission in my name, for any other qualification in any
university or other tertiary institution without the prior approval of the University, and
where applicable, any partner institution responsible for the joint-award of this degree.
I acknowledge that copyright of any published works contained within this thesis
resides with the copyright holder(s) of those works.
I give permission for the digital version of my research submission to be made

available on the web, via the University’s digital research repository, unless permission
has been granted by the University to restrict access for a period of time.
I acknowledge the support I have received for my research through the provision of an
Australian Government Research Training Program Scholarship.

06 April 2022
RMIT University
i
Acknowledgements
This thesis would not have been conclusive and informative without the assistance of
several people, to whom I would like to acknowledge and express my sincere
gratitude. I heartedly acknowledge and thank my thesis supervisors at RMIT
University, Professor Sefa Awaworyi Churchill and Professor Simon Feeny, for
providing invaluable guidance, council, immense knowledge, and motivation
throughout this research. I could not have imagined having better supervisors and
mentors for my PhD study.
I also wish to express my profound gratitude to RMIT University for providing me with
the financial support, which made this study possible. Additionally, I am indebted to
the administration team at RMIT University, which was always there whenever I
needed help and information. I also appreciate the feedback and comments I received
from panel members during my PhD candidature milestone reviews, which
significantly improved this work. This thesis also significantly benefited from the
extensive assistance of faculty members at Monash University, including Dr Klaus
Ackermann.
I wish to record my gratitude to my wife Margret and my family for their unstinted
support, encouragement and understanding while undertaking this PhD and writing
my thesis; I am truly grateful.
Last, I would like to give special thanks to God, for all the opportunities I received
during my candidature and for seeing me through all the difficulties; I will keep on
trusting you for my future.

April 2022
RMIT University
ii
Table of Contents
Declaration i
Acknowledgements ii
Table of Contents iii
List of Figures v
List of Tables vi
List of Abbreviations viii
Abstract 1
Chapter 1. Introduction to Thesis 3
1. Introduction 3
References 7
Chapter 2. Foreign Aid and Child Health Outcomes: Micro-Evidence from the Democratic
Republic of Congo 11
2.1 Introduction 11
2.2 A Brief Overview of Related Literature 16
2.3 Data and Variables 18
2.4 Methodology 20
2.5 Results 23
2.6 Conclusion 35
References 37
Chapter 2 – Appendices 45
Chapter 3. Foreign Aid and Energy Poverty: Sub-National Evidence from Senegal 54
3.1 Introduction 54
3.2 Related Literature 57
3.3 Channel Analysis: Why Should Foreign Aid Affect Energy Poverty? 59
3.4 Why Senegal? 62
3.6 Empirical Methodology 73
3.7 Results 76
3.8 Conclusion 87
References 90
Chapter 4. Temperature Shocks, Household Borrowing and the Mediating Role of Foreign Aid:
Evidence from Household Panel Data for Malawi 111
4.1 Introduction 111
4.2 Literature Review 115
4.4 Methodology 125
4.5 Results 125
4.6 Conclusion 136
iii
References 138
Chapter 5. Generosity or Exploitation? A Sub-National Analysis of Natural Resources and
Foreign Aid Allocation 148
5.1 Introduction 148
5.2 Why Should Natural Resources Influence Aid Allocation? 151
5.4 Methodology 157
5.5 Empirical Results 158
5.6 Conclusion 167
References 169
Chapter 5 – Appendix 5.A 176
Chapter 6. Conclusion to Thesis 177
6.1 Summary and Conclusions 177
6.2 Policy Implications and Recommendations 179
6.3 Limitations and Suggestions for Future Research 180
References 182
Appendix A 184
iv
List of Figures
Figure 2.1 – Distribution of DHS clusters and aid projects 53
Figure 3.1 – Distribution of DHS clusters with energy poverty 109
Figure 3.2 – Distribution of aid projects and DHS clusters 110
v
List of Tables
Table 2.1 – Main results 24
Table 2.2 – Different distance buffers 26
Table 2.3 – Alternative aid types 27
Table 2.4 – Foreign aid intensity 28
Table 2.5 – Foreign aid and conflict 29
Table 2.6 – Gender differentials in mean values of child health outcomes 32
Table 2.7 – Oaxaca-Blinder estimated child health coefficients, by gender 33
Table 2.8 – Oaxaca-Blinder child health outcomes decomposition 34
Table A2.1 – District-level aid projects in the DRC 45
Table A2.2 – Summary statistics 46
Table A2.3 – Effect of economic growth on total aid 47
Table A2.4 – Main results 48
Table A2.5 – Non-lagged aid and logit model 49
Table A2.6 – Individual vaccination indicators 50
Table A2.7 – Parallel trend analysis (balancing approach) 51
Table A2.8 – Parallel trend analysis (supplemental analysis) 52
Table 3.2 – Alternative distance buffers 78
Table 3.3 – Types of foreign aid 80
Table 3.4 – Alternative cut-offs for energy poverty index 81
Table 3.5 – Indicators of energy poverty 81
Table 3.6 – Foreign aid intensity 82
Table 3.7 – Exclusion of district fixed effects, and lagged and non-lagged foreign aid 83
Table 3.8 – Effects of aid on mechanisms 85
Table 3.9 – Effects of mechanisms 86
Table A3.1 – Description of aid projects in Senegal 102
Table A3.3 – Indicators and weights for MEPI 104
Table A3.4 – Effect of economic growth on foreign aid allocation 105
Table A3.5 – Main results 106
Table A3.6 – Aid and energy poverty over time 107
Table A3.7 – Parallel trend analysis 108
Table 4.2 – Alternative measures of temperature shocks 127
Table 4.3 – Alternative measures of credit access 128
Table 4.4 – Lagging temperature shocks 129
Table 4.5 – Controlling for rainfall 130
Table 4.6 – Access to formal and informal financial institutions or credit sources 131
Table 4.7 – Heterogenous effects of temperature shocks 132
vi
Table 4.8 – Correlation between temperature shocks and foreign aid 133
Table 4.9 – Foreign aid as a mediator between credit borrowing and temperature shocks 135
Table A4.2 – Full results with controls 147
Table 5.1 – Main Results 159
Table 5.2 – Aid intensity and natural resources 160
Table 5.3 – Aid and alternative indicators of vegetation 161
Table 5.4 – Aid and types of minerals 162
Table 5.5 – Types of aid and natural resources 162
Table 5.6 – Aid and distance to the capital city 164
Table 5.7 – Aid and district area 164
Table 5.8 – Aid and climate 165
Table 5.9 – Correlation between natural resources and mediators 166
Table 5.10 – Channel analysis 167
vii
List of Abbreviations
BCG Bacillus Calmette-Guerin
DAC Development Assistance Committee
DHS Demographic and Health Surveys
DID Difference-in-differences
DPT Diphtheria-pertussis-and-tetanus
DRC Democratic Republic of Congo
GDP Gross Domestic Product
GIE Geospatial Impact Evaluation
GPS Global Positioning System
IHPS Integrated Household Panel Survey
IMF International Monetary Fund
IPP Independent Power Producer
Km Kilometre
LIHC Low income-high cost
LPG Liquified Petroleum Gas
MPI Multidimensional Poverty Index
MEPI Multidimensional Energy Poverty Index
NGO Non-governmental organisation
NDVI Normalized Difference Vegetation Index
ODA Official Development Assistance
OECD Organisation for Economic Co-operation and Development
RCT Randomised Controlled Trial
UCDP Uppsala Conflict Data Program
UK United Kingdom
US United States
WSS Water supply and sanitation
viii
Abstract
This thesis is comprised of four related chapters on the allocation and efficacy of
foreign aid at the sub-national level. The first empirical chapter (Chapter 2) examines
the effect of aid on child health outcomes in the Democratic Republic of Congo (DRC).
It explores outcomes relating to three health categories: 1) prevalence of anaemia, 2)
vaccination rates and 3) mortality rates. Using data from AidData and Demographic
and Health Surveys (DHS), results from difference-in-differences (DID) estimations
show that aid increases vaccination coverage, reduces mortality rates, and lowers the
prevalence of anaemia among children living close to an aid project site. Further, the
chapter examines the role of aid in influencing the gender gap in child health
outcomes. The results show that aid narrows the gender gap in anaemia prevalence,
mortality, and vaccination rates against child-related diseases. Based on the findings,
the chapter suggests that widely spread aid projects that increase the chance of
household proximity can be useful in achieving desirable health outcomes.
The second empirical chapter (Chapter 3) contributes to the literature on foreign aid
effectiveness by examining the effect of foreign aid on energy poverty. Using five
rounds of DHS and sub-national aid data for Senegal, the chapter finds that aid lowers
the probability of energy poverty. The main results show that living within a 25-
kilometre radius of an aid project reduces the likelihood of being energy-poor by 3.3
percentage points. This finding is robust to a suite of sensitivity checks. The chapter
also examines four channels through which aid could influence energy poverty –
income poverty, education, health, and economic growth – and finds that income
poverty, education and economic growth are mechanisms through which aid transmits
to energy poverty.
The third empirical chapter (Chapter 4) examines the impact of temperature shocks
on household borrowing using three waves of Malawian household data from the
World Bank Microdata Library and weather data from the University of Delaware’s
online database. Applying a panel fixed effects model, the results show that
temperature shocks positively affect household borrowing. These results are robust to
several checks, including using alternative measures of temperature shocks and
household borrowing, applying additional lags on the temperature shocks variable,
and controlling for other factors, such as rainfall. Further, obtaining foreign aid data
1
from AidData, the chapter explores foreign aid as a potential sub-national transmission
channel through which temperature shocks may influence household borrowing.
Findings suggest that total foreign aid and sectorial-targeted foreign aid are channels
through which the impact of temperature shocks on household borrowing are
mediated. The chapter advocates for policies and strategies that enable households
to adapt and become resilient to temperature shocks.
The fourth empirical chapter (Chapter 5) contributes to the literature by examining the
effect of natural resources on foreign aid allocation. The chapter extracts district-level
data on natural resources and foreign aid for a panel of three countries for the period
2000 to 2014. Estimating a fixed-effects model, the results show that locations with
natural resources are positively associated with the receipt of foreign aid. The chapter
also investigates channels through which natural resources may influence foreign aid.
The results show that conflict and economic growth mediate the relationship between
natural resources and aid.
2
Chapter 1
Introduction to Thesis
1. Introduction
Huge volumes of foreign aid are annually disbursed to developing countries by the
donor community to help them address various development-related challenges,
including poverty, unemployment, poor health, civil conflicts, budget deficits and poor
growth, among others (see, e.g., OECD 2022a; World Bank 2020a). As of 2020, for
instance, the OECD’s Development Assistance Committee (DAC) disbursed about
US$161.2 billion in Official Development Assistance (ODA) to developing countries
(OECD 2021). Despite receiving such substantial aid amounts yearly, most of these
countries remain vulnerable to economic shocks and face many structural obstacles
to sustainable development (United Nations 2021b). Specifically, many developing
countries are characterised by poor economic performance, undermined health
systems, heightened poverty and inequality, and lower living standards, among others
(see, e.g., Abebe & Shushanik 2021; Ruch 2020; UNCTAD 2021), which begs the
common question in the literature of whether aid is effective at all.
A large body of literature thus examines the impact of foreign aid on development
outcomes at the macro level, and the findings from this literature remain mixed (see,
e.g., Banchani & Swiss 2019; Birchler & Michaelowa 2016; Munyanyi et al. 2020;
Thomson, M et al. 2017; Wayland 2019). On the one hand, some cross-country
studies in the literature posit that aid improves development outcomes, including
economic growth, education, health, access to water and sanitation, gender equality,
and poverty (see, e.g., Baliamoune‐Lutz 2016; Banchani & Swiss 2019; Galiani et al.
2017; Minasyan et al. 2017; Riddell & Niño-Zarazúa 2016). On the other hand, a
related strand of the literature, also employing cross-country analysis, argues that aid
lowers growth, increases poverty, widens the gender inequality gap, precipitates wars
and violence, creates aid-dependency and corruption, and undermines health
outcomes (see, e.g., Leeson 2008; Maïga 2014; Mishra & Newhouse 2009; Moyo
2009; Narang 2014; Thomson, M et al. 2017). Evidence suggests that examining the
effectiveness of aid using cross-country analysis results in these mixed or conflicting
views; hence the need to examine aid efficacy particularly at the sub-national or less
aggregated levels (Dreher & Lohmann 2015).
3
Sub-national level analysis has been argued as a better approach to evaluate the
efficacy of aid than cross-country analysis given that it 1) reduces the unobservable
country differences found in cross-country analyses that may lead to spurious relations
between aid and outcomes if not controlled for, 2) allows for the use of quasi-
experimental approaches in examining aid impacts and 3) enables an assessment of
aid effectiveness at the exact geographical location it is allocated (Marty et al. 2017;
Rajan & Subramanian 2008). However, there is a of lack robust research on aid
efficacy at the sub-national level, which hinders policy.
Given that much of what we know about aid impacts at the sub-national level has been
mostly focused on health outcomes (see, e.g., De & Becker 2015; Kotsadam et al.
2018; Marty et al. 2017; Odokonyero et al. 2018), this study extends the literature by
examining the effect of sub-national aid on other outcomes, providing new evidence
aimed at the development community (e.g., aid-recipient governments, donors, and
other key players). In attempting to fill this gap in the literature, Chapter 2 thus
examines the effect of foreign aid on energy poverty in Senegal, while Chapter 3
investigates aid impacts on child health outcomes in the Democratic Republic of
Congo (DRC). Senegal makes an interesting case to study aid impacts on energy
poverty given that the country is a major aid recipient (World Bank 2020a), relies on
traditional energy sources (e.g., firewood and charcoal) for cooking, and a large
percentage of its rural population still has no access to electricity (Ba 2018; USAID
2021b). Additionally, the country is politically stable (Vengroff & Creevey 2018), which
enables us to achieve the exact impacts of aid without weak institutional interference.
In the same context, like Senegal, DRC, receives huge amounts of foreign aid yearly
(World Bank 2020a). Moreover, the country is characterised serious health-related
issues, such as compromised health infrastructure, successive Ebola virus outbreaks,
increased malaria cases and deaths, heightened under five mortality rate and low
vaccination rates (see, e.g., Diese et al. 2018; Ducomble & Gignoux 2020; Kalisya et
al. 2015; Laokri et al. 2018; Lechthaler et al. 2019).
Although it is important to know about the impacts of aid at the sub-national level, it is
also crucial to understand how aid mediates relationships between outcomes. To shed
light on the mediating role of foreign aid, Chapter 4 explores how aid mediates the
relationship between temperature shocks and household borrowing in Malawi.
4
Temperature shocks may create conditions (lower economic growth, cause conflicts,
compromise health and raise poverty levels) (see, e.g., Acevedo et al. 2020; Al-
Bouwarthan et al. 2019; Azzarri & Signorelli 2020; Breckner & Sunde 2019) that can
either attract or discourage foreign aid, influencing household borrowing in the
process. We are motivated to place our study in Malawi by a number of attributes.
Specifically, Malawi is vulnerable to climate change and climate variability
(Warnatzsch & Reay 2019), often experiences droughts and floods (FAO 2021), has
a well transformed financial sector (World Bank 2021), and relies on foreign aid (Page
2019).
Given that developing countries depend on aid for development, it is equally important
to understand how donors choose to disburse aid in developing countries. Several
studies have sought to understand factors behind donor’s decision to allocate aid (see,
e.g., Alesina & Dollar 2000; Furuoka 2017; Kim & Oh 2012; Mahembe & Odhiambo
2019; Rahman & Giessen 2017). Similar to the aid efficacy literature, with the
exception of a few studies that conducted studies at the sub-national level (see, e.g.,
Anaxagorou et al. 2020; Bayale 2020; Bei 2019; Dipendra 2020; Yang et al. 2018),
most of the studies in this literature are at the macro-level.
The literature explores various determinants of aid allocation, including recipient’s

needs (e.g., better living standards, economic growth, health, employment and peace),
donor’s interests (e.g., new markets, trade benefits, influencing local politics), and
recipient’s merit (e.g., good policies and institutions) (see, e.g., Alesina & Dollar 2000;
Mahembe & Odhiambo 2019; Rahman & Giessen 2017). However, little is known
about how other outcomes, for example, natural resources influence aid allocation at
the sub-national level. Thus, Chapter 5 examines the impact of natural resources on
aid allocation using a panel of three African countries, including Nigeria, Somalia, and
Uganda.
Understanding aid impacts and allocation at the sub-national level provides a

theoretical framework against which governments can formulate fiscal policy
adjustments and draft supplementary budgets as well as fund programmes that
improve the welfare of the citizens. Moreover, such understanding would help donors
to evaluate the success of aid projects established primarily to improve household and
development outcomes.
5
Following this brief introductory chapter, Chapter 2 looks at the impact of sub-national
aid on child health outcomes in the Democratic Republic of Congo. Chapter 3
examines the effect of sub-national aid on energy poverty in Senegal. Chapter 4
explores the impact of temperature shocks on household borrowing, and the mediating
role of sub-national foreign aid between the two variables in Malawi. Chapter 5
investigates the effect of natural resources endowment on sub-national foreign aid
allocation, and Chapter 6 concludes the thesis.
6
References
Abebe, AS & Shushanik, H 2021, Six charts show the challenges faced by Sub-Saharan
Africa, International Monetary Fund (IMF), Washington, D.C.,
<https://www.imf.org/en/News/Articles/2021/04/12/na041521-six-charts-show-the-
challenges-faced-by-sub-saharan-africa>.
Acevedo, S, Mrkaic, M, Novta, N, Pugacheva, E & Topalova, P 2020, 'The Effects of Weather
Shocks on Economic Activity: What are the Channels of Impact?', Journal of Macroeconomics,
vol. 65, p. 103207.
Al-Bouwarthan, M, Quinn, MM, Kriebel, D & Wegman, DH 2019, 'Assessment of heat stress
exposure among construction workers in the hot desert climate of Saudi Arabia', Annals of
Work Exposures and Health, vol. 63, no. 5, pp. 505-520.
Alesina, A & Dollar, D 2000, 'Who gives foreign aid to whom and why?', Journal of Economic
Growth, vol. 5, no. 1, pp. 33-63.
Anaxagorou, C, Efthyvoulou, G & Sarantides, V 2020, 'Electoral motives and the subnational
allocation of foreign aid in sub-Saharan Africa', European Economic Review, vol. 127, p.
103430.
Azzarri, C & Signorelli, S 2020, 'Climate and poverty in Africa South of the Sahara', World
Development, vol. 125, p. 104691.
Ba, AS 2018, The energy policy of the Republic of Senegal, Université Paris Dauphine, Paris.
Baliamoune‐Lutz, M 2016, 'The Effectiveness of Foreign Aid to Women's Equality

Organisations in the MENA', Journal of International Development, vol. 28, no. 3, pp. 320-341.
Banchani, E & Swiss, L 2019, 'The impact of foreign aid on maternal mortality', Politics and
Governance, vol. 7, no. 1, pp. 53-67.
Bayale, N 2020, 'Empirical investigation into the determinants of foreign Aid in Sahel countries:
A panel bayesian model averaging approach', Defence and Peace Economics, vol., pp. 1-21.
Bei, LJ 2019, 'Where does the dragon’s gift go?: Subnational distribution of China’s aid to Sub-
Saharan Africa from 2007 to 2012', LSE Undergraduate Political Review, vol., pp. 27-59.
Birchler, K & Michaelowa, K 2016, 'Making aid work for education in developing countries: An
analysis of aid effectiveness for primary education coverage and quality', International Journal
of Educational Development, vol. 48, pp. 37-52.
Breckner, M & Sunde, U 2019, 'Temperature extremes, global warming, and armed conflict:
new insights from high resolution data', World Development, vol. 123, p. 104624.
De, R & Becker, C 2015, The foreign aid effectiveness debate: Evidence from Malawi,
AidData, USA.
Diese, M, Kalonji, A, Izale, B, Villeneuve, S, Kintaudi, NM, Clarysse, G, Ngongo, N & Ntambue,
AM 2018, 'Community-based maternal, newborn, and child health surveillance: perceptions
and attitudes of local stakeholders towards using mobile phone by village health volunteers in
7
the Kenge Health Zone, Democratic Republic of Congo', BMC Public Health, vol. 18, no. 1,
pp. 1-12.
Dipendra, K 2020, 'Which aid targets poor at the sub-national level?', World Development
Perspectives, vol. 17, p. 100177.
Dreher, A & Lohmann, S 2015, 'Aid and growth at the regional level', Oxford Review of
Economic Policy, vol. 31, no. 3-4, pp. 420-446.
Ducomble, T & Gignoux, E 2020, 'Learning from a massive epidemic: measles in DRC', The
Lancet Infectious Diseases, vol. 20, no. 5, p. 542.
FAO 2021, Economic and Policy Analysis of Climate Change, Food and Agriculture
Organization of the United Nations, Rome, <http://www.fao.org/in-
action/epic/countries/mwi/en/>.
Furuoka, F 2017, 'Determinants of China's and Japan's foreign aid allocations in Africa',
African Development Review, vol. 29, no. 3, pp. 376-388.
Galiani, S, Knack, S, Xu, LC & Zou, B 2017, 'The effect of aid on growth: Evidence from a
quasi-experiment', Journal of Economic Growth, vol. 22, no. 1, pp. 1-33.
Kalisya, LM, Salmon, M, Manwa, K, Muller, MM, Diango, K, Zaidi, R, Wendel, SK & Reynolds,
TA 2015, 'The state of emergency care in Democratic Republic of Congo', African Journal of
Emergency Medicine, vol. 5, no. 4, pp. 153-158.
Kim, EM & Oh, J 2012, 'Determinants of foreign aid: The case of South Korea', Journal of East
Asian Studies, vol. 12, no. 2, pp. 251-274.
Kotsadam, A, Østby, G, Rustad, SA, Tollefsen, AF & Urdal, H 2018, 'Development aid and
infant mortality. Micro-level evidence from Nigeria', World Development, vol. 105, pp. 59-69.
Laokri, S, Soelaeman, R & Hotchkiss, DR 2018, 'Assessing out-of-pocket expenditures for

primary health care: how responsive is the Democratic Republic of Congo health system to
providing financial risk protection?', BMC Health Services Research, vol. 18, no. 1, p. 451.
Lechthaler, F, Matthys, B, Lechthaler-Felber, G, Likwela, JL, Mavoko, HM, Rika, JM,

Mutombo, MM, Ruckstuhl, L, Barczyk, J & Shargie, E 2019, 'Trends in reported malaria cases
and the effects of malaria control in the Democratic Republic of the Congo', PloS One, vol. 14,
no. 7, p. e0219853.
Leeson, PT 2008, 'Escaping poverty: Foreign aid, private property, and economic
development', Journal of Private Enterprise, vol. 23, no. 2, pp. 39-64.
Mahembe, E & Odhiambo, NM 2019, 'Foreign aid, poverty and economic growth in developing
countries: A dynamic panel data causality analysis', Cogent Economics & Finance, vol. 7, no.
1, p. 1626321.
Maïga, EW 2014, Does foreign aid in education foster gender equality in developing
countries?, WIDER Working Paper.
Marty, R, Dolan, CB, Leu, M & Runfola, D 2017, 'Taking the health aid debate to the
subnational level: the impact and allocation of foreign health aid in Malawi', BMJ Global Health,
vol. 2, no. 1.
8
Minasyan, A, Nunnenkamp, P & Richert, K 2017, 'Does aid effectiveness depend on the quality
of donors?', World Development, vol. 100, pp. 16-30.
Mishra, P & Newhouse, D 2009, 'Does health aid matter?', Journal of Health Economics, vol.
28, no. 4, pp. 855-872.
Moyo, D 2009, Dead aid: Why aid is not working and how there is a better way for Africa,
Macmillan, UK.
Munyanyi, ME, Awaworyi Churchill, S & Skali, A 2020, 'Foreign Aid and Development Goals:
Revisiting the Evidence', in Moving from the Millennium to the Sustainable Development
Goals, Palgrave Macmillan, Singapore, pp. 181-197.
Narang, N 2014, 'Humanitarian assistance and the duration of peace after civil war', The
Journal of Politics, vol. 76, no. 2, pp. 446-460.
Odokonyero, T, Marty, R, Muhumuza, T, Ijjo, AT & Owot Moses, G 2018, 'The impact of aid
on health outcomes in Uganda', Health Economics, vol. 27, no. 4, pp. 733-745.
OECD 2021, Official Development Assistance (ODA), Organisation for Economic Cooperation
and Development (OECD), Paris, <https://www.oecd.org/dac/financing-sustainable-
development/development-finance-standards/official-development-assistance.htm>.
OECD 2022, OECD.Stat: Creditor Reporting System (CRS), Organisation for Economic
Cooperation and Development (OECD), Paris,
<https://stats.oecd.org/Index.aspx?DataSetCode=crs1>.
Page, S 2019, 'The Development Aid Situation in Malawi', in Development, Sexual Cultural
Practices and HIV/AIDS in Africa, Springer, pp. 43-60.
Rahman, MS & Giessen, L 2017, 'Formal and informal interests of donors to allocate aid:
Spending patterns of USAID, GIZ, and EU forest development policy in Bangladesh', World
Development, vol. 94, pp. 250-267.
Rajan, RG & Subramanian, A 2008, 'Aid and growth: What does the cross-country evidence
really show?', The Review of Economics and Statistics, vol. 90, no. 4, pp. 643-665.
Riddell, A & Niño-Zarazúa, M 2016, 'The effectiveness of foreign aid to education: What can
be learned?', International Journal of Educational Development, vol. 48, pp. 23-36.
Ruch, FU 2020, Policy Challenges for Emerging and Developing Economies: Lessons from
the Past Decade, World Bank, Washington DC.
Thomson, M, Kentikelenis, A & Stubbs, T 2017, 'Structural adjustment programmes adversely

affect vulnerable populations: a systematic-narrative review of their effect on child and
maternal health', Public Health Reviews, vol. 38, no. 1, p. 13.
UNCTAD 2021, Reaping the potential benefits of the African Continental Free Trade Area for
inclusive growth, United Nations Conference on Trade And Development (UNCTAD), Geneva.
United Nations 2021, Least Developed Countries (LDCs), United Nations: Department of
Economic and Social Affairs, New York, <https://www.un.org/development/desa/dpad/least-
developed-country-category.html>.
9
USAID 2021, Senegal Energy Sector Overview, United States Agency for International
Development, Washington DC, <https://www.usaid.gov/powerafrica/senegal>.
Vengroff, R & Creevey, L 2018, 'Senegal: the evolution of a quasi democracy', in Political
Reform in Francophone Africa, Routledge, pp. 204-222.
Warnatzsch, EA & Reay, DS 2019, 'Temperature and precipitation change in Malawi:

Evaluation of CORDEX-Africa climate simulations for climate change impact assessments and
adaptation planning', Science of the Total Environment, vol. 654, pp. 378-392.
Wayland, J 2019, 'Constraints on aid effectiveness in the water, sanitation, and hygiene
(WASH) sector: evidence from Malawi', African Geographical Review, vol. 38, no. 2, pp. 140-
156.
World Bank 2020, DataBank: World Development Indicators,

<https://databank.worldbank.org/reports.aspx?source=World-Development-Indicators#>.
World Bank 2021, Innovation and Digital Transformation in the Financial Sector in Malawi,
World Bank, Washington D.C, <https://www.worldbank.org/en/results/2021/01/14/innovation-
and-digital-transformation-in-the-financial-sector-in-malawi>.
Yang, H-m, Liu, P-l & Guo, Y 2018, 'Determinants of China’s development assistance for
health at the sub-national level of African countries (2006–2015)', Infectious Diseases of
Poverty, vol. 7, no. 1, pp. 1-9.
10
Chapter 2
Foreign Aid and Child Health Outcomes: Micro-Evidence from the Democratic
Republic of Congo
2.1 Introduction
Foreign aid has emerged as an important policy intervention for achieving

development. Despite its popularity, several aid-recipient nations still suffer from
economic hardships. This has spurred a large body of literature that examines the
impact of foreign aid on different development outcomes. Yet, the effectiveness of aid
on development remains unclear, as the evidence on the impact of aid remains mixed.
One strand of the literature, employing cross-country analysis, argues that foreign aid
improves development outcomes, including economic growth (see, e.g., Dollar &
Burnside 2000; Galiani et al. 2017; Gyimah‐Brempong et al. 2012), education (see,
e.g., Christensen et al. 2011; d’Aiglepierre & Wagner 2013), poverty (see, e.g., Alvi &
Senbeta 2012; Arndt et al. 2015; Hirano & Otsubo 2014), gender and income equality
(see, e.g., Baliamoune‐Lutz 2016; Pickbourn & Ndikumana 2016), and health (see,
e.g., Banchani & Swiss 2019; Bendavid & Bhattacharya 2014; Feeny & Ouattara
2013). In contrast, a related strand of the literature suggests that aid undermines
economic growth (see, e.g., Liew et al. 2012; Mallik 2008; Yiew & Lau 2018), widens
gender gaps (see, e.g., Maïga 2014), increases poverty in low-income countries (see,
e.g., Arvin & Barillas 2002), distorts markets and breeds corruption (see, e.g., Moyo
2009), and fails to improve health outcomes (see, e.g., Pandolfelli et al. 2014;
Thomson, M et al. 2017; Wilson 2011).
Until recently, the debate and findings on the effectiveness of aid have been primarily
based on cross-country evidence, and this has been cited as one of the main reasons
for the mixed findings in the literature (Dreher & Lohmann 2015). More importantly,
given that aid projects are often too localised and aid amounts are relatively small,
evidence drawn from macro-level estimates or cross-country comparisons are likely
to mask nuanced impacts of aid. Compared to cross-country macro analysis, sub-
national analysis has the advantage of allowing us to examine the efficacy of aid
projects at their exact geographical sites and how neighbouring households are
affected. Additionally, it enables us to estimate aid effects using quasi-experimental
11
techniques, such as the difference-in-differences (DID) technique (Marty et al. 2017).
However, we know very little about the effectiveness of aid at the sub-national level.
The only studies of which we are aware that use microdata at the sub-national level
have examined the effect of aid on infant mortality (Kotsadam et al. 2018); malaria
prevalence and health care quality (Marty et al. 2017); the productivity disease-burden
and sickness (Odokonyero et al. 2018); and school outcomes, disease severity, and
diarrhoea incidence (De & Becker 2015). We contribute to this literature by examining
the effects of aid on a wide range of child health outcomes at the sub-national level.
Although some evidence exists on the impacts of aid on health outcomes, several
questions remain unanswered. Much of what we know on the effectiveness of aid on
household- or individual-level development outcomes have focused on health
outcomes in general or adult populations (De & Becker 2015; Marty et al. 2017;
Odokonyero et al. 2018). De and Becker (2015) and Marty et al. (2017) focus on health
outcomes in the Malawian general population. De and Becker (2015) find evidence of
aid’s effectiveness in decreasing the prevalence of diarrhea and disease severity,
while Marty et al. (2017) find that health infrastructure and disease control aid are
associated with a decrease in the prevalence of malaria. Odokonyero et al. (2018)
focus on the adult population in Uganda and find that aid has an inverse relationship
with the number of productive days lost to illness. Although these studies provide very
useful evidence on the effectiveness of aid on health outcomes in general or adult
populations, they are less able to speak to how aid shapes health outcomes among
children. An exception is Kotsadam et al. (2018), who examine the effects of foreign
aid on infant mortality in Nigeria. Thus, there is a paucity of micro-level evidence on
the impact of aid on a wide range of child health outcomes, including child vaccination
coverage and prevalence of disease.
We extend and contribute to the growing literature on the impact of aid at the sub-
national level by examining the effect of foreign aid on under-five child mortality, the
prevalence of anaemia and vaccinations against five child health issues including
tuberculosis (Bacillus Calmette-Guerin (BCG) vaccine), polio, diphtheria-pertussis-
and-tetanus (DPT), measles, and vitamin A deficiency in the DRC. The focus on child
health outcomes is important for at least two reasons. First, the focus on children is
important given the role of children in economic development. Evidence shows that
children present important human capital benefits and ensure sustainability in the long
12
run (see, e.g., Campbell-Barr & Nygård 2014; Julian 2015; Tayefi Nasrabadi et al.
2021). For instance, improved child health is likely to promote socioeconomic position
in adulthood with significant implications for productivity and economic growth.
Children have been referred to as a ‘nation’s greatest resource’ in some contexts
(Allen & Martinez 2005). Understanding factors that influence child health outcomes
is therefore important, and can inform policy aimed at human capital accumulation.
The focus on children also has important implications for growth policies given that
human capital development in children is essential to the long-term economic
development of countries (see, e.g., Klobodu et al. 2018; Samson et al. 2016; Sharma
2018).
Second, children represent a highly vulnerable population, and thus child health
remains an important issue on the global development and policy agenda. Global
statistics show that the under-5 child mortality rate is still above the targeted rate of at
least 25 deaths per 1,000 live-births (Aheto 2019), as 1 child dies every second from
a vaccine-preventable disease (Papachrisanthou & Davis 2019). Yet, approximately
14 million infants have no access to vaccination services, with more mortality cases
observed in low income countries due to their poor health systems (UNICEF 2019)
and lack of financial resources to augment health sector budgets (Munyanyi et al.
2020). Given that the prevalence of child health issues and mortality remains an
important global priority, our focus on the impact of aid on child health outcomes, as
opposed to adult health outcomes, has important implications for policy. For instance,
taking into account the increased prevalence of child mortality related to missed
vaccinations, our focus on child vaccinations as an important health outcome
contributes to the discourse on factors worth considering to promote vaccine uptake
among households, and the need to reduce child mortality rates via vaccinations.
The DRC makes for an interesting case for at least four reasons. First, the DRC has
some of the worst vaccination rates worldwide. Evidence suggests that the DRC is a
major contributor to the two-thirds of children with no access to vaccination services
globally (Acharya et al. 2018). Compared to other countries, the vaccination rates in
the DRC have persistently declined. For instance, between 2019 and 2020,
vaccination coverage against diphtheria, hepatitis B, tetanus, haemophilus influenzae
type B and whooping cough dropped by up to 10%, while poliovirus vaccination
coverage decreased by 8.4% for inactivated polio vaccine and 5.4% for oral polio
13
vaccine. Coverage for vaccination against measles, chickenpox, yellow fever,
rotavirus and pneumococcal disease also fell by between 1.5% and 4.5% (UNICEF
2020a). Evidence indicates that the DRC is in the top 7 list of African countries with
children who have not received all basic vaccinations (Bobo et al., 2022). According
to Bobo et al. (2022), about 55% of the children in the DRC have not received all basic
vaccinations.
Second, the DRC is highly dependent on foreign aid. Annually, it receives more than
US$1 billion in ODA. In 2003 alone, it received approximately US$6.5 billion (World
Bank 2020a). Additionally, the donor community spends millions of dollars on the DRC
health sector every year, with approximately US$280 million in this sector in 2013 and
2017 (OECD 2022a). Compared to other aid-dependent countries in the region (e.g.,
Burundi, the Central African Republic, Rwanda, and South Sudan), the DRC ranks the
highest in terms of total aid received. For instance, between 1960 and 2019, the DRC,
South Sudan, Rwanda, and the Central African Republic received net official
development assistance and official aid approximately equal to $80.3 billion, $14
billion, $30.9 billion, and $13.1 billion, respectively (World Bank, 2022).
Third, the DRC faces numerous health issues. For instance, it has the world's second-
highest number of malaria cases and deaths (Lechthaler et al. 2019) and the world's
fifth-highest under-five mortality rate (Diese et al. 2018). Having endured 10 outbreaks
of the Ebola virus since 1976 and long periods of conflicts characterised by low
government investment, the DRC’s health infrastructure is in a poor state (Kalisya et
al. 2015). Medical facilities and medicines are limited (Laokri et al. 2018) while 33
million people in rural areas have no access to drinking water (UNICEF 2020b). In
2019, 6,045 lives were lost to measles – an issue that has been linked with very low
vaccination rates (Ducomble & Gignoux 2020).
Fourth, the DRC has poor institutions, characterised by high levels of corruption (see,
e.g., Pring & Vrushi 2019; Steurs 2019). Evidence shows that poor institutions
undermine aid efficacy by inducing corruption and mismanagement of donated
resources (see, e.g., Chong et al. 2009; Ijaiya & Ijaiya 2004). Due to poor institutional
quality in aid-recipient countries, aid fails to improve development outcomes, such as
health, poverty and economic growth (see, e.g., Leeson 2008; Lensink & White 2001;
Papanek 1972). This makes the DRC an interesting case study to examine whether
14
aid is likely to be effective without an enabling institutional environment free of
corruption and mismanagement.
We use two rounds of data from the Demographic and Health Surveys (DHS) (2007
and 2013–14) and district-level data on foreign aid projects from the AidData projects
database covering the period 2008 to 2014. We consider aid projects with physical
sites and examine the effect of aid on child health outcomes using a DID model.
Additionally, we examine how foreign aid influences gender gaps in child health
outcomes. Our results show that aid increases vaccination coverage and reduces
anaemia and mortality rates among children residing within a 25-kilometre (km)
distance from an aid project site. The results are robust to a number of checks,
including the use of alternative estimation approaches, alternative measures of
selected health outcomes, and use of different distance buffers, among others.
Further, aid narrows the gender gap in the prevalence of anaemia, mortality and
vaccinations against tuberculosis, DPT, polio, measles and vitamin A deficiency.
We contribute to several bodies of literature. First, we make additions to the broad aid
effectiveness literature exploring the impact of aid on various outcomes (see, e.g.,
Arndt et al. 2015; Baliamoune‐Lutz 2016; Dollar & Burnside 2000; Munyanyi et al.
2020; Yiew & Lau 2018), a subset of which has focused on health outcomes (see, e.g.,
Banchani & Swiss 2019; Feeny & Ouattara 2013; Thomson, M et al. 2017). Second,
we contribute to the growing literature on aid efficacy at the sub-national level. Our
study also relates to the literature that has examined factors influencing gender gaps
in health outcomes (see, e.g., Banda et al. 2017; Ross et al. 2012; Yu 2018). This
literature has explored gender gaps in various health outcomes, including mortality,
mental wellbeing and maternal health, but has focused on factors other than
development aid. For instance, Ross et al. (2012) examine the effect of education on
gender gaps in mortality and health. Similarly, Yu (2018) explores the association
between gender disparities and gender inequality in mental health. Banda et al. (2017)
investigate the role of gender inequality in maternal health in Zambia. In this context,
our study contributes to this literature by showing that foreign aid is also an important
factor that influences gender gaps in health, and thus, needs to be taken into account
when strategizing on how to narrow health-related gender gaps.
15
The remainder of the chapter proceeds as follows: The next section gives a brief
literature review on aid and health. Sections 2.3 and 2.4 describe the data and
methodology, respectively. Section 2.5 interprets the results, and Section 2.6
concludes the chapter.
2.2 A Brief Overview of Related Literature
Conceptually, aid may affect health directly or through a number of channels, including
economic growth, pro-poor public expenditures, and income poverty. Evidence shows
that aid directly allocated to the health sector improves access to healthcare and
enhances people's health practises (see, e.g., Banchani & Swiss 2019; Munyanyi et
al. 2020; Pickbourn & Ndikumana 2019). For instance, aid stabilises growth (Collier &
Goderis 2009; Jeanneney & Tapsoba 2012), triggers self-sustaining growth
(Rosenstein-Rodan 1943; Shleifer 2009), provides employment and entrepreneurial
opportunities (Jeanneney & Tapsoba 2012; Mahadea & Kaseeram 2018), and reduces
income poverty (Kaya et al. 2013). It is expected that households living in such
conducive environments characterised by high growth and a lower incidence of
poverty will have the economic means to afford a high standard of living that promotes
health.
Aid improves pro-poor expenditures (Mosley et al. 2004). These are expenditures that
can improve the welfare of poor households, including expenditures on education,
water supply, sanitation services, and health and other social services (Munyanyi et
al. 2020). By improving education outcomes (see, e.g., d’Aiglepierre & Wagner 2013;
Dreher & Lohmann 2015; Riddell & Niño-Zarazúa 2016), aid increases the probability
of living long and healthily (see, e.g., Davies et al. 2018; Raghupathi & Raghupathi
2020; Zajacova & Lawrence 2018). Educated people also accumulate significant
knowledge on health issues and thus make informed decisions about their health
(Hoffmann & Lutz 2019). Similarly, by improving access to improved water and
sanitation facilities (see, e.g., Gopalan & Rajan 2016; Ndikumana & Pickbourn 2017;
Wayland 2019), aid decreases the prevalence of water-borne diseases among
households (Wayland 2019).
The empirical findings on the impact of aid on health outcomes are mixed. Some cross-
country studies suggest that aid improves health outcomes (see, e.g., Banchani &
Swiss 2019; Bendavid & Bhattacharya 2014; Burnside & Dollar 1998; Gomanee et al.
16
2005; Pickbourn & Ndikumana 2019; Rasschaert et al. 2011). Some findings from this
literature show that countries with good policies are associated with lower infant
mortality rates (see, e.g., Burnside & Dollar 1998). Additionally, evidence suggests
that by financing public expenditures, aid improves welfare indicators, including infant
mortality and poverty (see, e.g., Arndt et al. 2015; Gomanee et al. 2005). Some studies
also show that the type of aid influences aid efficiency. For instance, evidence
suggests that health aid and education aid are more effective than total aid in reducing
infant and child mortality (see, e.g., Mishra & Newhouse 2009; Wolf 2007). Similarly,
aid targeted at the health sector is associated with improved prenatal and postnatal
care, and lower maternal mortality (Banchani & Swiss 2019), reduced under-five child
mortality rates (see, e.g., Bendavid & Bhattacharya 2014; Pickbourn & Ndikumana
2019), and increased life expectancy (see, e.g., Bendavid & Bhattacharya 2014;
Ziesemer 2016). Moreover, aid improves the coverage of medical drugs and
functionality of health facilities that provide services related to maternal and child
health (Rasschaert et al. 2011).
On the contrary, other cross-country studies show that aid does not improve health
outcomes (see, e.g., Mishra & Newhouse 2009; Pandolfelli et al. 2014; Thomson, M
et al. 2017; Wilson 2011), and this is the case for both total aid and health-sector aid
(Mishra & Newhouse 2009; Wilson 2011). International Monetary Fund (IMF) loans
have been found to increase maternal mortality rates (Pandolfelli et al. 2014).
Evidence also suggests that in some contexts, IMF structural adjustment programmes
worsen maternal and child health, undermine the availability of food and income and
reduce access to quality healthcare (Thomson, M et al. 2017). The inability of IMF
loans and programmes to improve health outcomes has been attributed to strict fiscal
targets set by the IMF for developing countries that are beneficiaries of their loans and
programmes. These inflexible targets often end up as major challenges that compel
countries to drain aid funds from social sectors, such as the health sector, to repay
debts (Kentikelenis et al. 2015; Kentikelenis et al. 2016; Stuckler et al. 2011).
Moving away from cross-country analyses, a very small strand of the aid effectiveness
literature recently examined aid effects on health outcomes at the sub-national level
(see, e.g., De & Becker 2015; Kotsadam et al. 2018; Marty et al. 2017; Odokonyero et
al. 2018). Using data from the AidData projects and five rounds of the DHS data for
Nigeria, Kotsadam et al. (2018) explore the effect of sub-national aid on infant
17
mortality. Applying a DID approach, they find that aid reduces infant mortality among
households living near aid projects. Odokonyero et al. (2018) examine the impact of
aid on disease prevalence and the productivity burden of disease as measured by the
number of productivity days lost due to sickness. Their results indicate that aid reduces
the productivity burden of disease but is less effective in lowering the prevalence of
the disease. Furthermore, people living near aid projects are found to benefit more
from aid. De and Becker (2015) and Marty et al. (2017) investigate the efficacy of sub-
national aid in Malawi. De and Becker (2015) find that sub-national aid lowers disease
severity and diarrhea incidence while Marty et al. (2017) find that sub-national aid
improves healthcare quality and reduces the prevalence of malaria, mostly when
targeted at parasitic control and health infrastructure projects.
In summary, the existing literature on aid and health shows that research on the effect
of aid on health outcomes at the sub-national level is limited. Against this background,
we contribute to this limited literature by exploring a comprehensive set of child health
outcomes and gender gaps in these outcomes.
2.3 Data and Variables
This study draws on data from two main sources. Health outcome variables are drawn
from the DHS. Specifically, we use two rounds of the DHS data for the DRC: 2007 and
2013–14. The DHS surveys have been conducted since 1984, and report information
on a number of household issues, including mortality, education, vaccination, and
disease prevalence (ICF 2018). Further, the surveys, which are nationally
representative, report information on the exact locations of surveyed clusters. With
such attributes, we are able to match these surveys with geo-referenced data from
other sources, and also make inferences about populations.
We match the DHS data with geo-referenced foreign aid data from the AidData
projects1. The AidData projects database, run by the College of William & Mary,
provides freely available data on sub-national aid (Tierney et al. 2011). The database
contains multilateral and bilateral aid projects data extracted from donor-published
reports, projects records, and the OECD Creditor Reporting System. Further, the
database includes information on the type of aid projects (e.g., industry, energy supply
1 https://www.aiddata.org/datasets
18
and generation, health) and their exact locations (e.g., state, region, district,
geographic coordinates). Since we are interested in the efficacy of aid at the sub-
national level, we limit our study to the district-level and focus on aid projects with
physical locations, consistent with existing literature (Marty et al. 2017). In that case,
we can examine the effect of aid on health outcomes among clusters located near aid
project sites. Our aid projects data for the DRC, at district-level, stretches from 2008
to 2014. A description of these projects is presented in Table A2.1 in the appendices.
2.3.1 Health outcomes
We explore seven child health outcomes, which capture child mortality, the prevalence
of anaemia and vaccinations. The prevalence of vaccinations is measured using a
dummy variable that captures vaccinations received. Given data availability, we focus
on: 1) BCG vaccine, which protects children from tuberculosis, 2) vitamin A vaccine or
supplements, which prevent child blindness, boost the immune system and enable
child growth and development, 3) DPT vaccine, which helps children fight against
diphtheria, pertussis and tetanus, 4) polio vaccine, which protects children against
paralysis caused by the poliovirus, and 5) measles vaccine, which safeguards children
from contracting measles, rubella and mumps. Child mortality and the prevalence of
anaemia are at the household level and are also measured using binary variables that
capture the incidence of each outcome.
In the context of vaccinations received and anaemia prevalence, the DHS asks
mothers: 1) whether their children have received the following vaccinations and or
supplements: BCG, DPT, polio, measles and vitamin A, and 2) whether any of their
children has anaemia. For each response recorded as ‘Yes’, one is assigned for the
binary variable. Further, the DHS provides information on child mortality. Specifically,
mothers are asked if they have given birth to a child who was alive but later died. For
those who answered ‘Yes’, the time the child was born and died is reported. If the child
died before reaching five years of age, one is assigned for the binary variable.
2.3.2 Sub-national aid
Consistent with the literature that has used the AidData datasets at the sub-national
level (Kotsadam et al. 2018; Odokonyero et al. 2018), we measure aid using a dummy
variable to reflect household proximity to aid projects. If a household is located within
19
a 25 km radius (i.e., our main buffer) from any aid project site, a code 1 is assigned
(Kotsadam et al. 2018; Odokonyero et al. 2018). Put differently, to measure aid, we
measure the distance from each household to the nearest aid project using the
geocodes provided. Our baseline results are based on a 25 km buffer, and thus
households living within this radius are considered the treatment group while those
outside of this range are the control group. In robustness checks, we also examine
alternative distance buffers to examine if proximity to an aid project influences the
effectiveness of aid. Distribution of the aid projects and DHS clusters in the DRC is
shown in Figure 2.1 in the appendix.
2.3.3 Covariates
We include a set of socio-demographic control variables consistent with the existing

literature on the determinants of household health (see, e.g., Awaworyi Churchill et al.
2019; Marty et al. 2017; Odokonyero et al. 2018). In the child health context, these
include education status, age and its squared term, gender, and employment status,
among others, for the household head and mother. Table A2.2 in the appendices
presents the summary statistics of all variables employed in our analysis.
2.4 Methodology
2.4.1 Main empirical specification
In this study, we seek to examine the effect of aid on health outcomes among children
living in the proximity of project sites. To this end, we adopt a repeated cross-sectional
DID approach in line with the existing literature on the efficacy of sub-national aid (see,
e.g., Marty et al. 2017; Odokonyero et al. 2018), and estimate the following linear
probability regression equation:
𝐶𝐻𝑖,𝑗,𝑡 = 𝛼0 + 𝛼1 𝑡𝑖𝑚𝑒 + 𝛼2 (𝑡𝑖𝑚𝑒 ∗ 𝑎𝑖𝑑𝜃𝑘𝑚𝑗,𝑡−1 ) + 𝐸𝐺𝑗,𝑡 + 𝜗𝑋𝑖,𝑗,𝑡 + 𝜀𝑖,𝑗,𝑡 (1)
where 𝐶𝐻 represents child health outcomes for household 𝑖 living in district 𝑗 in year
𝑡. 𝐸𝐺 represents district-level economic performance proxied by night-time light. 𝑋 is
a vector of covariates that were discussed earlier. To run a DID model, we need two
periods: 1) aid projects pre-implementation period – the period before aid projects are
introduced, and 2) aid projects implementation period – the period when aid projects
are set up and functioning. These two periods enable us to compare the likelihood of
a child: 1) getting vaccinated, 2) dying before the age of five, and 3) contracting
20
anaemia, before and after an aid project has been set up close by. 𝑡𝑖𝑚𝑒 is a binary
variable that equals 1 for years 2013–14 (i.e., aid projects implementation period) and
1 in 2007 (i.e., aid projects pre-implementation period). In accordance with the
literature on aid efficacy at the sub-national level, we define our treatment group (i.e.,
households benefiting from aid projects) as those households that are within a
specified buffer distance (i.e., 𝜃𝑘𝑚) from an aid project site (see, e.g., Kotsadam et al.
2018; Odokonyero et al. 2018). Households outside this specified threshold are
considered the control group. In this regard, sub-national foreign aid (𝑎𝑖𝑑𝜃𝑘𝑚) is a
binary variable that equals one if a household stays within this defined buffer distance
from an aid project location. For our main estimations and results, we use a 25 km
buffer distance. We derive the aid treatment effect by creating an interaction term of
the sub-national aid variable at time 𝑡 − 1 and the time variable (i.e., 𝑡𝑖𝑚𝑒 ∗
𝑎𝑖𝑑𝜃𝑘𝑚𝑗,𝑡−1 ). 𝛼2 is the coefficient of interest that measures the impact of sub-national
aid on child health outcomes. Consistent with the literature on aid effectiveness, we
lag our aid variable to control for the delay that aid may take to effect changes on our
outcomes of interest (see, e.g., Arshad et al. 2016; Kaya et al. 2013). We apply a one-
year lag (𝑡 − 1) and control for district fixed effects. 𝜀 is the error term.
The DID approach requires that the allocation of the treatment (i.e., aid) be
uncorrelated to our outcomes at baseline. Thus, for equation (1) to identify the causal
effect of foreign aid on child health outcomes, the establishment of foreign aid projects
needs to be exogenous to the focal health outcomes. We contend that the
establishment of aid projects is likely to be exogenous due to the way and pace in
which these aid projects are established, and the criteria for choosing project locations.
The process of deciding where, how much and when to allocate foreign aid rests with
the donors, and this process can take years (Bearce & Tirone 2010). Although aid
recipient countries may be facing various development challenges, which in some
cases may require immediate financial assistance, the decision to allocate aid
ultimately depends on donors and their within-country or internal processes, which are
often determined by political and bureaucratic systems that are not correlated with
socioeconomic conditions in aid recipient countries. Thus, there is always a time lag
when it comes to foreign aid allocation. Importantly, in the context of our study, we find
relatively homogenous child health outcomes across districts and communities in the
21
DRC prior to 2008 when aid projects were implemented. With such homogeneity, we
can intuitively expect that differences in child health outcomes across districts do not
influence donor decisions on where (in which districts) to implement aid projects. Put
differently, districts that benefit from aid projects are not chosen because they have
either poorer or better child health outcomes.2
In addition to internal bureaucratic processes in donor countries, the existing literature

demonstrates that donors typically focus on specific characteristics of recipient
countries to determine aid allocation. For instance, evidence suggests that donors
tend to establish aid projects mainly in areas that provide everyone with easier
accessibility to the benefits of the intervention (Brown et al. 2014; Safarpour et al.
2020). More generally, donors may consider the economic conditions of recipients,
and this could be an important factor that influences aid allocation.
Given that economic growth have been shown to be correlated with child health
outcomes in other contexts (see, e.g., Abdul Salam et al. 2015; Rashad & Sharaf
2018), this could be a potential threat to our identification strategy. To explore if this is
an issue in the context of our study, we examine the impact of economic performance
at the district level on aid allocation. Since we do not have data on GDP per capita at
the district level, we follow the literature that has used night-time light as a proxy for
economic performance (see, e.g., Ackermann et al. 2021; Appau et al. 2021; Awaworyi
Churchill et al. 2021; Hodler & Raschky 2014), and use this as an indicator for district
level economic performance in the DRC. The results, which are presented in Appendix
Table A2.3, show that district-level economic performance in the DRC is not correlated
with aid allocation. To further alleviate any concerns, we also control for district-level
economic growth in our child outcome models.
2.4.2 Oaxaca-Blinder decomposition model
To examine the gender gaps in child health outcomes, we use the Oaxaca-Blinder
decomposition model. This model quantifies the effects of the factors behind the
differences in health outcomes of female and male children (Blinder 1973; Oaxaca
2We demonstrate this in our data by showing that prior to the establishment of aid projects, there is no
statistically significant difference between child health outcomes for the control and treatment groups.
Results for balancing test are reported in Table A2.7.
22
1973). Further, the model allows us to explain the gaps in the means of given child
health outcomes between the two groups (i.e., female and male), and to draw
conclusions on whether child health outcomes converge or diverge over time as a
result of aid. To estimate the gender gaps, we run the standard Oaxaca-Blinder
decomposition model as:
𝐶𝐻𝑖,𝑓 − 𝐶𝐻𝑖,𝑚 = 𝜔
̂𝑓 (𝑋𝑓 − 𝑋𝑚 ) + 𝑋𝑚 (𝜔
̂𝑓 − 𝜔
̂𝑚 ) (2)
where 𝑓 and 𝑚 denote female and male, respectively. 𝐶𝐻𝑖,𝑓 and 𝐶𝐻𝑖,𝑚 represent the
mean values of child health outcome 𝑖 obtained after running linear regression models
on the determinants of that outcome variable 𝑖. 𝑋𝑓 and 𝑋𝑚 denote mean values of
factors that influence child health outcomes, including aid. 𝜔
̂𝑓 and 𝜔
̂𝑚 represent the
estimated effects of these factors. We take the mean level of male child health
outcome 𝑖 as the baseline and examine what it takes for it to converge to the mean
level of the female child health outcome 𝑖. We estimate equation (2) as a linear
probability model.
2.5 Results
2.5.1 Main results: Aid and child health outcomes
Table 2.1 presents the results for the association between sub-national aid and child
health outcomes.3 Panel A reports bivariate regression results without relevant
controls while Panel B adds on the set of covariates. Across both panels, Columns (1)
to (5) report results for the effects of aid on tuberculosis, DPT, polio, measles, and
vitamin A, respectively. Columns (6) and (7) report results for effects on anaemia
prevalence and mortality, respectively.
3 The full set of results showing all covariates are reported in appendix Table A2.4.
23
Table 2.1 – Main results
(1) (2) (3) (4) (5) (6) (7)

VARIABLES BCG DPT Polio Measles Vitamin A Anaemia children Child mortality
Panel A: Bivariate results

Total aid (25 km) 0.158*** 0.166*** 0.063*** 0.099*** 0.154*** -0.044*** -0.009***
(0.003) (0.003) (0.002) (0.004) (0.004) (0.005) (0.002)
Controls No No No No No No No
Observations 91,535 91,097 91,715 91,194 91,083 48,616 98,584
R-squared 0.025 0.025 0.014 0.007 0.272 0.004 0.002
Panel B: Multivariate results
Total aid (25 km) 0.105*** 0.116*** 0.033*** 0.049*** 0.105*** -0.025*** -0.004*
(0.005) (0.005) (0.004) (0.006) (0.005) (0.006) (0.002)
Controls Yes Yes Yes Yes Yes Yes Yes

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082
R-squared 0.069 0.071 0.048 0.048 0.289 0.019 0.006
Notes: Robust standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1
From Panel A, the unconditional estimates show that aid has a positive and statistically
significant effect on BCG, DPT, polio, measles, and vitamin A vaccinations. We find
that living in a household located within a 25 km distance from a foreign aid project
site increases the likelihood of being vaccinated against: 1) tuberculosis by 15.8
percentage points, 2) DPT by 16.6 percentage points, 3) polio by 6.3 percentage
points, 4) measles by 9.9 percentage points, and 5) vitamin A deficiency by 15.4
percentage points. Columns (5) and (6) show that aid has a negative and statistically
significant impact on anaemia prevalence among children and child mortality,
respectively. Specifically, aid reduces the likelihood of being anaemic by 4.4
percentage points and lowers the probability of mortality by 0.9 percentage points
among children from households within a 25 km radius of an aid project.
From Panel B, we find that with the inclusion of the relevant covariates, the effects of
aid remain robust. Specifically, we find that proximity to an aid project increases the
probability of being vaccinated against: 1) tuberculosis by 10.5 percentage points, 2)
DPT by 11.6 percentage points, 3) polio by 3.3 percentage points, 4) measles by 4.9
percentage points, and 5) vitamin A deficiency by 10.5 percentage points. In Column
(5), the results show that aid lowers the likelihood of anaemia prevalence among
children by 2.5 percentage points. Similarly, we observe the same impact of aid on
24
child mortality – where aid reduces child mortality by 0.4 percentage points (see
Column (6)).
The effects of the control variables are consistent with expectations and the literature.
For instance, mother’s educational status, birth order, and household head’s
employment status, all have the expected signs. Specifically, as shown in Appendix
Table A2.4, mother’s educational status has a positive and statistically significant
effect on tuberculosis, DPT, polio, measles and vitamin A vaccinations, and a negative
and statistically significant effect on anaemia prevalence and mortality among children.
These findings align with existing evidence, which suggests that educated mothers
raise healthy children (see, e.g., Chen & Li 2009; Mensch et al. 2019). The household
head’s employment status has a positive and statistically significant impact on all
vaccination indicators (see, Columns (1) to (5) of Table A2.4), which is consistent with
the literature that associates parental employment with improved children’s health
(see, e.g., Benson & Mokhtari 2011; Pieters & Rawlings 2020). The birth order variable
has a negative and statistically significant effect on all vaccination indicators (i.e.,
tuberculosis, DPT, polio, measles, and vitamin A deficiency). This implies that children
born after the first-born child are less likely to be immunised (Gavrielov-Yusim et al.
2012; Stein-Zamir & Israeli 2017). Similarly, economic growth, captured by night-time
light, increases the coverage of vaccination indicators (e.g., BCG, DPT, polio and
vitamin A deficiency) and lowers the prevalence of anaemia among children.
2.5.2 Robustness checks
We conduct a series of tests to examine the robustness of our results. First, we check
if our results are robust to the use of different distance buffers. Specifically, we
examine the effect of aid on child health outcomes, using buffer distances that are
both below and above the 25 km-threshold used in our main analysis (i.e., 5, 10, 15,
20, 50, and 75 km). This exercise allows us to determine whether a household's
proximity to an aid project site influences the magnitude of the effect of aid on child
health outcomes. The results, which are reported in Table 2.2, show that our results
are robust up to a 50 km distance buffer, after which aid mostly becomes statistically
insignificant and less effective. These results confirm findings from the existing
literature that aid efficacy is higher only when aid projects are positioned close to
household locations (see, e.g., Odokonyero et al. 2018; Wayland 2019).
25
Table 2.2 – Different distance buffers
(1) (2) (3) (4) (5) (6) (7)

Panel A: 5 km
Total aid (5 km) 0.088*** 0.089*** 0.022*** 0.031*** 0.124*** -0.050*** -0.002
(0.005) (0.005) (0.004) (0.007) (0.006) (0.007) (0.003)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.061 0.061 0.046 0.047 0.281 0.019 0.006
Panel B: 10 km
Total aid (10 km) 0.101*** 0.107*** 0.024*** 0.036*** 0.118*** -0.040*** -0.004
(0.005) (0.005) (0.004) (0.006) (0.006) (0.007) (0.003)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.063 0.064 0.046 0.047 0.281 0.019 0.006
Panel C: 15 km
Total aid (15 km) 0.104*** 0.115*** 0.027*** 0.036*** 0.132*** -0.038*** -0.003
(0.005) (0.005) (0.004) (0.006) (0.006) (0.006) (0.003)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.064 0.066 0.047 0.047 0.284 0.019 0.006
Panel D: 20 km
Total aid (20 km) 0.116*** 0.123*** 0.036*** 0.056*** 0.144*** -0.029*** -0.003
(0.005) (0.005) (0.004) (0.006) (0.005) (0.006) (0.002)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.067 0.069 0.047 0.049 0.287 0.019 0.006
Panel E: 50 km
Total aid (50 km) 0.088*** 0.095*** 0.034*** 0.040*** 0.110*** -0.018*** -0.007***
(0.005) (0.005) (0.004) (0.005) (0.005) (0.005) (0.002)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.064 0.065 0.047 0.048 0.283 0.018 0.006
Panel F: 75 km
Total aid (75 km) -0.012 -0.001 -0.020*** -0.016* 0.013 0.009 -0.017***
(0.008) (0.008) (0.006) (0.008) (0.008) (0.008) (0.004)
Observations 30,253 30,155 30,257 30,164 29,887 29,467 31,728

R-squared 0.150 0.172 0.072 0.096 0.134 0.101 0.014
Notes: All regressions include controls as in Table 2.1. Robust standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1
Second, we examine the sensitivity of our results to different types of foreign aid.
Consistent with the literature on aid effectiveness at the sub-national level, our main
results are based on total aid (see, e.g., Kotsadam et al. 2018; Marty et al. 2017;
Odokonyero et al. 2018). To examine the robustness of our results, we also consider
three types of aid that may influence child health, including 1) health aid, 2) water
26
supply and sanitation (WSS) aid, and 3) agricultural aid. We include agricultural aid
because more than 75% of the DRC’s rural population relies on agriculture for its
livelihood (FAO 2017). To test if our results are robust to different aid types, we run all
seven regressions presented in our main results, separately including these three aid
types as our main explanatory variables. The results, presented in Table 2.3, show
that these three aid types increase the coverage of all vaccination indicators and lower
anaemia prevalence and mortality among children.
Table 2.3 – Alternative aid types
(1) (2) (3) (4) (5) (6) (7)

Panel A: Health aid

Health aid (25 km) 0.114*** 0.112*** 0.041*** 0.045*** 0.103*** -0.015** -0.009***
(0.005) (0.005) (0.004) (0.006) (0.006) (0.006) (0.002)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.067 0.066 0.048 0.048 0.281 0.018 0.006
Panel B: Water supply & sanitation aid
WSS aid (25 km) 0.124*** 0.127*** 0.047*** 0.016** 0.156*** -0.036*** -0.005*
(0.005) (0.005) (0.004) (0.007) (0.006) (0.008) (0.003)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.064 0.065 0.048 0.047 0.284 0.019 0.006
Panel C: Agricultural aid
Agricultural aid (25 km) 0.139*** 0.163*** 0.057*** 0.096*** 0.183*** -0.091*** -0.006**
(0.005) (0.005) (0.004) (0.007) (0.006) (0.008) (0.003)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.066 0.069 0.048 0.050 0.287 0.021 0.006
*** p<0.01, ** p<0.05, * p<0.1
Third, we verify if our results are sensitive to the use of non-lagged foreign aid. Aid
takes time to influence outcomes, hence the need to lag the aid variable in our main
analysis (see, e.g., Kaya et al. 2013). We check the robustness of our results by
exploring the impact of non-lagged aid on child health outcomes. Panel A of Appendix
Table A2.5 presents the results. We find that our results remain robust. Fourth, we
check the robustness of our results to the use of an alternative regression model. For
ease of interpretation, our main results are based on a linear probability model.
However, given that our outcome variables are binary, we also apply logit regressions
27
to examine if our results remain robust. The results, which are reported in Panel B of
Table A2.5, show that our results remain robust.
Fourth, we test the sensitivity of our results to the individual DPT and polio vaccination
indicators. The DPT vaccine has three doses (i.e., DPT 1, DPT 2, and DPT 3), while
the polio vaccine has four (i.e., Polio 0, Polio 1, Polio 2 and Polio 3). In our main results,
a code 1 is assigned if a child received at least one of the DPT doses. Similarly, a
code 1 is assigned if a child received any of the four polio doses. For robustness
checks, we examine the effect of aid on these individual doses or vaccination
indicators. The results, reported in Appendix Table A2.6, show that aid increases the
likelihood of being vaccinated against: 1) DPT-1 by 12.4 percentage points, 2) DPT-2
by 13.9 percentage points, 3) DPT-3 by 13.4 percentage points, 4) Polio-0 by 21.1
percentage points, 5) Polio-1 by 1.9 percentage points, 6) Polio-2 by 3.1 percentage
points, and 7) Polio-3 by 2.8 percentage points. In this regard, the effect of aid on child
vaccination remains robust.
Fifth, we explore the sensitivity of our results to foreign aid intensity – that is, foreign
aid amounts. Our main results capture provide insight into how households are
influenced by their proximity to aid projects. However, the results do not provide
differentiation by project size. Put differently, by using proximity to aid projects, we are
not able to capture the impact of aid intensity or aid project scale. In robustness
checks, we examine the impact of foreign aid amounts allocated to aid project
establishments on child health outcomes. This allows us to understand if the size of
the aid project impacts on child health outcomes. The results for this exercise, which
are reported in Table 2.4, indicate that foreign aid intensity increases BCG, DPT, polio
and vitamin A deficiency vaccination coverage.
Table 2.4 – Foreign aid intensity
(1) (2) (3) (4) (5) (6) (7)

Aid intensity (25km) 0.005*** 0.005*** 0.002*** 0.001 0.010*** -0.001 0.000
(0.000) (0.000) (0.000) (0.001) (0.000) (0.001) (0.000)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.058 0.059 0.046 0.047 0.281 0.018 0.006
*** p<0.01, ** p<0.05, * p<0.1
28
Sixth, we examine the extent to which our results are influenced by political instability.
The literature shows that political instability undermines foreign aid efficacy and
allocation (see, e.g., de Ree & Nillesen 2009; Lis 2013; Munyanyi et al. 2020). We
examine whether this is also the case in the context of our study. We capture political
instability using armed conflict, and following the literature (see, e.g., Ackermann et al.
2021; Ayana et al. 2016; Wagner et al. 2018), we obtain data on conflict from the
Uppsala Conflict Data Program (UCDP) database.4 We set a dummy variable equal to
one if a household is located in an area classified as a conflict zone. To examine the
impact of conflict, we include an interaction term that captures the interaction between
conflict and aid. We also conduct a sub-sample analysis that focuses on household in
conflict zones and those outside conflict zones. We present the results in Table 2.5.
Table 2.5 – Foreign aid and conflict
(1) (2) (3) (4) (5) (6) (7)

Panel A: Households inside conflict zones
Total aid (25km) 0.004 0.016 -0.003 -0.039* 0.011 -0.069*** -0.015
(0.017) (0.018) (0.014) (0.022) (0.019) (0.023) (0.013)
Observations 10,761 10,694 10,754 10,706 10,711 9,966 11,190

R-squared 0.070 0.053 0.061 0.058 0.406 0.037 0.010
Panel B: Households outside conflict zones
Total aid (25km) 0.096*** 0.109*** 0.024*** 0.046*** 0.098*** -0.005 -0.003
(0.006) (0.006) (0.005) (0.007) (0.007) (0.007) (0.003)
Observations 33,873 33,676 33,965 33,719 33,708 32,985 35,892

R-squared 0.054 0.057 0.044 0.043 0.251 0.017 0.005
Panel C: Full sample and interaction term
Total aid (25km) 0.094*** 0.087*** 0.040*** 0.030** 0.094*** 0.008 0.009
(0.011) (0.011) (0.009) (0.012) (0.010) (0.012) (0.005)
Conflict 0.129*** 0.121*** 0.047*** 0.146*** 0.018* -0.107*** -0.007
(0.009) (0.010) (0.008) (0.011) (0.009) (0.011) (0.005)
Aid*Conflict -0.063*** -0.052*** -0.043*** -0.097*** 0.014 0.044** -0.012
(0.015) (0.016) (0.013) (0.019) (0.015) (0.019) (0.009)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.270 0.278 0.202 0.200 0.433 0.199 0.062
*** p<0.01, ** p<0.05, * p<0.1
4 More information on UCDP can be found here: https://ucdp.uu.se/.
29
Panels A and B report results for the effect of aid on child health outcomes in
households inside and outside the conflict zones, respectively. In Panel C, we report
results for the full sample, which includes the interaction term. From Panel A, where
we consider households living in conflict zones, we find the effect of aid child health
outcomes is statistically insignificant across all specification except Column (6), which
focuses on anaemia. In contrast, from Panel B, we observe that effect of aid on child
health outcomes is positive and statistically significant in Columns (1) to (5). Thus, for
households living outside conflict zones, foreign aid increases the coverage of all
vaccination indicators. The results from Panel C confirm this. Specifically, we find that
for tuberculosis, DPT, polio, and measles, the coefficient on the interaction term is
negative and statistically significant. These results imply that the positive impact of aid
on child health outcomes in relatively stronger in areas without conflict compared to
areas with conflict. The positive and significant coefficient on the interaction term in
the model for anaemia also confirms that the negative effect of aid on the prevalence
of anaemia is relatively weaker in conflict zones than in non-conflict zones.
Next, we verify the reliability of our results to DID assumptions. For our DID estimates
to be valid, in the absence of the treatment intervention (in our case, foreign aid
projects), the difference in outcomes between the control and treatment groups should
be constant over time. To test the validity of this assumption in our data, we run a
balancing test on all the child health outcomes at baseline conditions. By balancing on
the outcomes at baseline, we ensure that the treatment and control groups have
parallel trends – thus satisfying the common parallel trend assumption. For the
assumption to hold, the difference in outcomes between the control and treatment
groups should be statistically insignificant. The results, which are reported in Appendix
Table A2.7, show that the differences in child health outcomes between the two groups
are not statistically significant at the 5% level.
Another important consideration is the year-gap between the pre-implementation and

implementation periods, which is relatively wide and thus differences in outcomes may
be attributed to various factors other than aid over the period. As an additional check,
we separately examine the effect of aid on households in the control group in the 2007
and 2013–14 periods. The results for this exercise are reported in Appendix Table
A2.8. In Panel A, where we consider households outside the 25 km aid radius in 2007
(i.e., the control group at baseline), we find that being far away from aid projects fails
30
to improve all child health outcomes (i.e., the impact of foreign aid on the control
group’s outcomes is statistically insignificant). Similarly, we observe the same results
in Panel B, where we focus on the control group in 2013–14 (i.e., the control group
during the implementation period). Specifically, the results show that being outside the
25 km threshold of an aid project renders the impact of foreign aid on these child health
outcomes statistically insignificant. Based on the results, we argue that foreign aid
projects are the only factor explaining the differences in outcomes between the control
and treatment groups – thus, the wide year-gap is not an issue.
2.5.1 Gender differentials and child health outcomes
Table 2.6 shows the gender differentials in the mean values of the child health
outcomes obtained using the baseline data (i.e., data for 2007). The mean values for
males are higher than those of females for most of the outcomes. The results show
that there are significant female-male gaps of: 1) -1.720% for BCG vaccination
received, 2) -1.039% for polio vaccination received, 3) 6.523% for vitamin A
supplements received, and 4) -6.854% for anaemia reports.
The female-male gaps for the other outcomes are statistically insignificant. Thus, while
there is no evidence of a significant gender gap in child mortality rates and vaccination
uptake for DPT and measles, we find that the average vaccination rates for
tuberculosis, polio and anaemia are higher for males than females. In contrast,
vaccination against vitamin A deficiency is higher for females than males.
31
Table 2.6 – Gender differentials in mean values of child health outcomes
Female- Female-
Variables Male Female Total Male gap t-stat Male gap %
Panel A: BCG
BCG 0.748 0.735 0.741 -0.013 -2.501** -1.720%
(0.004) (0.004) (0.003) (0.005)
Observations 14,679 14,818 29,497
Panel B: DPT
DPT 0.734 0.728 0.731 -0.007 -1.337 -0.949%
(0.004) (0.004) (0.003) (0.005)
Observations 14,499 14,695 29,194
Panel C: Polio
Polio 0.820 0.811 0.816 -0.008 -1.882* -1.039%
(0.003) (0.003) (0.002) (0.005)
Observations 14,785 14,895 29,680
Panel D: Measles
Measles 0.614 0.614 0.614 -0.0003 -0.055 -0.049%
(0.004) (0.004) (0.003) (0.006)
Observations 14,646 14,686 29,332
Panel E: Vitamin A
Vitamin A 0.096 0.103 0.100 0.006 1.873* 6.523%
(0.002) (0.002) (0.002) (0.003)
Observations 14,881 14,943 29,824
Panel F: Anaemia children
Anaemia children 0.679 0.634 0.656 -0.045 -5.589*** -6.854%
(0.006) (0.006) (0.004) (0.008)
Observations 6,967 6,936 13,903
Panel G: Child mortality
Child mortality 0.081 0.081 0.081 0.0001 0.047 0.175%
(0.002) (0.002) (0.002) (0.003)
Observations 16,196 16,266 32,462
Notes: Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 2.7 presents the Oaxaca-Blinder estimated child health coefficients by gender.
In Panels A and B, we estimate multivariate regressions based on the female and
male samples, respectively.
32
Table 2.7 – Oaxaca-Blinder estimated child health coefficients, by gender
(1) (2) (3) (4) (5) (6) (7)
Panel A: Female sample
Total aid (25km) 0.116*** 0.136*** 0.035*** 0.038*** 0.123*** -0.001 0.000
(0.007) (0.007) (0.007) (0.008) (0.007) (0.008) (0.003)

Observations 22,363 22,258 22,380 22,237 22,249 21,469 23,555
R-squared 0.069 0.074 0.051 0.051 0.281 0.014 0.008
Panel B: Male sample
Total aid (25km) 0.112*** 0.113*** 0.036*** 0.061*** 0.122*** -0.054*** -0.009***
(0.007) (0.007) (0.006) (0.008) (0.007) (0.008) (0.003)

Observations 22,271 22,112 22,339 22,188 22,170 21,482 23,527
R-squared 0.068 0.068 0.047 0.050 0.289 0.027 0.005
*** p<0.01, ** p<0.05, * p<0.1
Both panels show that aid improves child health outcomes. However, the coefficients
on aid are mostly higher in the male sub-sample than in the female sub-sample.
Columns (1) to (5) of both panels show that aid increases the likelihood of getting
vaccinated against: 1) tuberculosis by 11.6 percentage points for females and 11.2
percentage points for males, 2) DPT by 13.6 percentage points for females and 11.3
percentage points for males, 3) polio by 3.5 percentage points for females and 3.6
percentage points for males, 4) measles by 3.8 percentage points for females and 6.1
percentage points for males and 5) vitamin A deficiency by 12.3 percentage points for
females and 12.2 percentage points for males. These results suggest that aid narrows
the gender gap in the vaccination against tuberculosis, DPT, polio, measles and
vitamin A deficiency. From Column (6), the results show that among female children,
aid has no significant effect on anaemia prevalence, while among male children, aid
reduces the probability of being anaemic by 5.4 percentage points. Column (7)
indicates that aid lowers the likelihood of male child mortality by 0.9 percentage points
and has a statistically insignificant effect on female child mortality. Thus, aid narrows
the gender gap in anaemia prevalence and mortality among male children.
33
Table 2.8 presents the Oaxaca-Blinder child health outcomes decomposition. Here,
the contribution of foreign aid towards explaining the gender gap in child health
outcomes is decomposed into the coefficient and endowment effects.
Table 2.8 – Oaxaca-Blinder child health outcomes decomposition

Endowment effects Coefficient effects Summation
Coefficient Coefficient
Panel A: BCG
Total aid (25km) 0.002*** 0.001
(0.0005) (0.003)
Total 0.004*** 0.004 -0.0003

(0.001) (0.004) (0.004)
Panel B: DPT
Total aid (25km) 0.002*** 0.007**
(0.001) (0.003)
Total 0.004*** -0.0003 -0.005
(0.001) (0.004) (0.004)
Panel C: Polio
Total aid (25km) 0.001*** -0.0003
(0.0002) (0.002)
Total 0.002** 0.001 -0.0003
(0.001) (0.003) (0.003)
Panel D: Measles
Total aid (25km) 0.001*** -0.006**
(0.0002) (0.003)
Total 0.002* 0.009** 0.007
(0.001) (0.005) (0.005)
Panel E: Vitamin A
Total aid (25km) 0.002*** 0.0002
(0.001) (0.003)
Total 0.003 0.015*** 0.013***

(0.003) (0.004) (0.005)
Panel F: Anaemia Children
Total aid (25km) -0.0002 0.015***
(0.0001) (0.003)
Total -0.002*** -0.038*** -0.036***
(0.001) (0.005) (0.005)
Panel G: Child Mortality
Total aid (25km) -0.0001* 0.003**
(0.00004) (0.001)
Total -0.0005*** -0.002 -0.002

(0.0002) (0.002) (0.002)
*** p<0.01, ** p<0.05, * p<0.1
34
We find that differences in endowments play the dominant role in explaining the
gender gap in child health outcomes. For instance, in the case of tuberculosis, polio
and vitamin A, the coefficient effects are statistically insignificant, suggesting that any
observed gender gaps in these health outcomes can be explained by differences in
endowments. Overall, we find evidence of a significant endowment effect for all child
outcomes except vaccination against anaemia. Depending on the outcome, the
endowment effect ranges between 0.001 and 0.002 for the vaccination rate models,
and -0.0001 for child mortality. Given the relatively small effect sizes, the results for
the endowment effect show that if female children had the same characteristics as
male children, their mean uptake of vaccination and mortality would only be marginally
different if they had the same exposure to foreign aid. This supports the findings that
aid narrows the gender gap in child health outcomes.
2.6 Conclusion
We examine the effect of sub-national aid on child health outcomes, capturing

vaccinations received (i.e., BCG, DPT, polio, measles, and vitamin A), anaemia
prevalence and mortality. We match data from the DHS and AidData projects for the
DRC, and apply the DID estimation technique to examine how proximity to aid projects
influences child health.
Our main results show that aid has a positive and statistically significant impact on
vaccination against: 1) tuberculosis (BCG), 2) diphtheria, pertussis, and tetanus
(DPT), 3) polio, 4) measles, and 5) vitamin A deficiency. Additionally, aid has a
negative and statistically significant effect on anaemia prevalence and mortality among
children. Our results are robust to a number of sensitivity checks, including the use of
alternative foreign aid types (e.g., health aid, WSS aid, and agricultural aid), use of
individual vaccination indicators, use of foreign aid intensity, use of non-lagged aid,
use of an alternative regression model, and controlling for conflict. Robustness checks
on the use of different distance buffers show that our results are only robust up to a
50 km buffer. Buffers greater than 50 km are associated with aid ineffectiveness. We
further examine the role of aid in explaining gender gaps in child health outcomes. We
find that aid narrows the gender gap in anaemia prevalence, mortality and vaccination
against tuberculosis, DPT, polio, measles and vitamin A deficiency.
35
Our results contradict findings from the literature, based on cross-country estimations,
that show that aid does not improve health outcomes (see, e.g., Pandolfelli et al. 2014;
Thomson, M et al. 2017). Our findings suggest that when viewed from a micro
perspective, aid is effective. Our findings that proximity to aid projects tend to promote
better child health outcomes lends support to the need for policies that promote the
establishment of community-based aid projects. DRC is known for its poor institutional
quality (see, e.g., Callaway 2018; Titeca & Edmond 2019). That aid positively
influences most of the outcomes considered in this study suggests that aid is likely to
be effective even among poor institutions. Our findings also show that aid can be
instrumental in addressing the development goals related to gender equality in human
capital accumulation, especially with regard to narrowing the gender gaps in health
outcomes. Aid can, therefore, be an important policy instrument when devising policies
aimed at achieving the goal of gender equality.
A limitation of this study, however, is that we do not have indicators of institutional

quality to examine the moderating role of institutions in the aid-health relationship.
Nonetheless, we speculate that the impact of aid on health outcomes could be
stronger than we observed here if institutions work effectively. Future studies
examining the impact of aid at the sub-national level could incorporate indicators of
institutional quality to understand the moderating role in the sub-national context.
36
References
Abdul Salam, A, Elsegaey, I, Khraif, R, AlMutairi, A & Aldosari, A 2015, 'Components and
public health impact of population growth in the Arab world', PloS One, vol. 10, no. 5, p.
e0124944.
Acharya, P, Kismul, H, Mapatano, MA & Hatløy, A 2018, 'Individual-and community-level

determinants of child immunization in the Democratic Republic of Congo: a multilevel
analysis', PloS One, vol. 13, no. 8, p. e0202742.
Ackermann, K, Churchill, SA & Smyth, R 2021, 'Mobile phone coverage and violent conflict',
Journal of Economic Behavior & Organization, vol. 188, pp. 269-287.
Aheto, JMK 2019, 'Predictive model and determinants of under-five child mortality: evidence
from the 2014 Ghana demographic and health survey', BMC Public Health, vol. 19, no. 1, p.
64.
Allen, M & Martinez, S 2005, 'The politics of children's issues: Challenges and opportunities
for advancing a children's agenda in the political arena', in R Lerner, F Jacobs & D Wertlieb
(eds), Applied Developmental Science – An Advanced Textbook, SAGE, London, pp. 325-
352.
Alvi, E & Senbeta, A 2012, 'Does foreign aid reduce poverty?', Journal of International
Development, vol. 24, no. 8, pp. 955-976.
Appau, S, Awaworyi Churchill, S, Smyth, R & Trinh, T-A 2021, 'The long-term impact of the
Vietnam War on agricultural productivity', World Development, vol. 146, p. 105613.
Arndt, C, Jones, S & Tarp, F 2015, 'Assessing foreign aid’s long-run contribution to growth
and development', World Development, vol. 69, pp. 6-18.
Arshad, M, Farooq, O, Sultana, N & Farooq, M 2016, 'The role of foreign aid in the
development of Pakistan’s education sector: a time-lag analysis', European Online Journal of
Natural and Social Sciences, vol. 5, no. 3, pp. 588-594.
Arvin, BM & Barillas, F 2002, 'Foreign aid, poverty reduction, and democracy', Applied
Economics, vol. 34, no. 17, pp. 2151-2156.
Awaworyi Churchill, S, Farrell, L & Smyth, R 2019, 'Neighbourhood ethnic diversity and mental
health in Australia', Health Economics, vol. 28, no. 9, pp. 1075-1087.
Awaworyi Churchill, S, Munyanyi, ME, Smyth, R & Trinh, T-A 2021, 'Early life shocks and
entrepreneurship: Evidence from the Vietnam War', Journal of Business Research, vol. 124,
pp. 506-518.
Ayana, EK, Ceccato, P, Fisher, JR & DeFries, R 2016, 'Examining the relationship between
environmental factors and conflict in pastoralist areas of East Africa', Science of the Total
Environment, vol. 557, pp. 601-611.

37
Banda, PC, Odimegwu, CO, Ntoimo, LF & Muchiri, E 2017, 'Women at risk: Gender inequality
and maternal health', Women & Health, vol. 57, no. 4, pp. 405-429.
Bearce, DH & Tirone, DC 2010, 'Foreign Aid Effectiveness and the Strategic Goals of Donor
Governments', The Journal of Politics, vol. 72, no. 3, pp. 837-851.
Bendavid, E & Bhattacharya, J 2014, 'The relationship of health aid to population health
improvements', JAMA Internal Medicine, vol. 174, no. 6, pp. 881-887.
Benson, L & Mokhtari, M 2011, 'Parental employment, shared parent–child activities and
childhood obesity', Journal of Family and Economic Issues, vol. 32, no. 2, pp. 233-244.
Blinder, AS 1973, 'Wage discrimination: reduced form and structural estimates', Journal of
Human Resources, vol. 1, pp. 436-455.
Bobo, FT, Asante, A, Woldie, M, Dawson, A & Hayen, A 2022, 'Child vaccination in sub-
Saharan Africa: Increasing coverage addresses inequalities', Vaccine, vol. 40, no. 1, pp. 141-
150.
Brown, D, Donini, A & Clarke, P 2014, Engagement of crisis-affected people in humanitarian

action, Active Learning Network for Accountability and Performance (ALNAP), Addis Ababa.
Burnside, C & Dollar, D 1998, Aid, the incentive regime, and poverty reduction, World Bank
Publications.
Callaway, A 2018, Powering Down Corruption: Tackling Transparency and Human Rights
Risks from Congo’s Cobalt Mines to Global Supply Chains [online]. The Enough Project,
Washington, pp. 1e28.
Campbell-Barr, V & Nygård, M 2014, 'Losing sight of the child? Human capital theory and its
role for early childhood education and care policies in Finland and England since the mid-
1990s', Contemporary Issues in Early Childhood, vol. 15, no. 4, pp. 346-359.
Chen, Y & Li, H 2009, 'Mother’s education and child health: Is there a nurturing effect?', Journal
of Health Economics, vol. 28, no. 2, pp. 413-426.
Chong, A, Gradstein, M & Calderon, C 2009, 'Can foreign aid reduce income inequality and
poverty?', Public Choice, vol. 140, no. 1-2, pp. 59-84.
Christensen, Z, Homer, D & Nielson, DL 2011, 'Dodging adverse selection: How donor type
and governance condition aid’s effects on school enrollment', World Development, vol. 39, no.
11, pp. 2044-2053.
Collier, P & Goderis, B 2009, 'Does aid mitigate external shocks?', Review of Development
d’Aiglepierre, R & Wagner, L 2013, 'Aid and universal primary education', Economics of
Education Review, vol. 37, pp. 95-112.
Davies, NM, Dickson, M, Smith, GD, Van Den Berg, GJ & Windmeijer, F 2018, 'The causal
effects of education on health outcomes in the UK Biobank', Nature Human Behaviour, vol. 2,
no. 2, pp. 117-125.
38
AidData, USA.
de Ree, J & Nillesen, E 2009, 'Aiding violence or peace? The impact of foreign aid on the risk
of civil conflict in sub-Saharan Africa', Journal of Development Economics, vol. 88, no. 2, pp.
301-313.
Diese, M, Kalonji, A, Izale, B, Villeneuve, S, Kintaudi, NM, Clarysse, G, Ngongo, N & Ntambue,
AM 2018, 'Community-based maternal, newborn, and child health surveillance: perceptions
and attitudes of local stakeholders towards using mobile phone by village health volunteers in
the Kenge Health Zone, Democratic Republic of Congo', BMC Public Health, vol. 18, no. 1,
pp. 1-12.
Dollar, D & Burnside, C 2000, 'Aid, policies, and growth', American Economic Review, vol. 90,
no. 4, pp. 847-868.
Ducomble, T & Gignoux, E 2020, 'Learning from a massive epidemic: measles in DRC', The
Lancet Infectious Diseases, vol. 20, no. 5, p. 542.
FAO 2017, The Democratic Republic of the Congo. Response Plan 2017–2018, Rome.
Feeny, S & Ouattara, B 2013, 'The effects of health aid on child health promotion in developing
countries: cross-country evidence', Applied Economics, vol. 45, no. 7, pp. 911-919.
Gavrielov-Yusim, N, Battat, E, Neumann, L, Friger, M & Balicer, RD 2012, 'Birth order and
private voluntary immunization—a study of 110,902 children', Vaccine, vol. 30, no. 2, pp. 442-
447.
Gomanee, K, Girma, S & Morrissey, O 2005, 'Aid, public spending and human welfare:
evidence from quantile regressions', Journal of International Development, vol. 17, no. 3, pp.
299-309.
Gopalan, S & Rajan, RS 2016, 'Has foreign aid been effective in the water supply and
sanitation sector? Evidence from panel data', World Development, vol. 85, pp. 84-104.
Gyimah‐Brempong, K, Racine, JS & Gyapong, A 2012, 'Aid and economic growth: Sensitivity
analysis', Journal of International Development, vol. 24, no. 1, pp. 17-33.
Hirano, Y & Otsubo, S 2014, Aid is good for the poor, The World Bank.
Hodler, R & Raschky, PA 2014, 'Regional favoritism', The Quarterly Journal of Economics,
vol. 129, no. 2, pp. 995-1033.
Hoffmann, R & Lutz, SU 2019, 'The health knowledge mechanism: evidence on the link
between education and health lifestyle in the Philippines', The European Journal of Health
39
ICF 2018, Demographic and Health Surveys Standard Recode Manual for DHS7, The
Demographic and Health Surveys Program, ICF, Rockville, Maryland, U.S.A.
Ijaiya, GT & Ijaiya, MA 2004, 'Foreign aid and poverty reduction in sub-Saharan Africa: A
cross-country investigation', South African Journal of Economic and Management Sciences,
vol. 7, no. 3, pp. 542-552.
Jeanneney, SG & Tapsoba, SJA 2012, 'Aid and income stabilization', Review of Development
Julian, G-P 2015, 'The Value of the Future: The Child as Human Capital and the Neoliberal
Labor of Race', Women's Studies Quarterly, vol. 43, no. 1/2, pp. 181-196.
Kalisya, LM, Salmon, M, Manwa, K, Muller, MM, Diango, K, Zaidi, R, Wendel, SK & Reynolds,
TA 2015, 'The state of emergency care in Democratic Republic of Congo', African Journal of
Emergency Medicine, vol. 5, no. 4, pp. 153-158.
Kaya, O, Kaya, I & Gunter, L 2013, 'Foreign aid and the quest for poverty reduction: Is aid to
agriculture effective?', Journal of Agricultural Economics, vol. 64, no. 3, pp. 583-596.
Kentikelenis, A, King, L, McKee, M & Stuckler, D 2015, 'The international monetary fund and
the Ebola outbreak', The Lancet Global Health, vol. 3, no. 2, pp. e69-e70.
Kentikelenis, AE, Stubbs, TH & King, LP 2016, 'IMF conditionality and development policy
space, 1985–2014', Review of International Political Economy, vol. 23, no. 4, pp. 543-582.
Klobodu, S, Dawson, J, Reed, D & Carpio, C 2018, 'Do maternal and child health promote
economic development? A case study of six sub-Saharan African countries', African Journal
of Food, Agriculture, Nutrition and Development, vol. 18, no. 3, pp. 13761-13775.
Laokri, S, Soelaeman, R & Hotchkiss, DR 2018, 'Assessing out-of-pocket expenditures for

primary health care: how responsive is the Democratic Republic of Congo health system to
providing financial risk protection?', BMC Health Services Research, vol. 18, no. 1, p. 451.
Lechthaler, F, Matthys, B, Lechthaler-Felber, G, Likwela, JL, Mavoko, HM, Rika, JM,

Mutombo, MM, Ruckstuhl, L, Barczyk, J & Shargie, E 2019, 'Trends in reported malaria cases
and the effects of malaria control in the Democratic Republic of the Congo', PloS One, vol. 14,
no. 7, p. e0219853.
Leeson, PT 2008, 'Escaping poverty: Foreign aid, private property, and economic
development', Journal of Private Enterprise, vol. 23, no. 2, pp. 39-64.
Lensink, R & White, H 2001, 'Are there negative returns to aid?', Journal of Development
Studies, vol. 37, no. 6, pp. 42-65.
Liew, C-Y, Mohamed, MR & Mzee, SS 2012, 'The impact of foreign aid on economic growth
of East African countries', Journal of Economics and Sustainable Development, vol. 3, no. 12,
pp. 129-188.
Lis, P 2013, 'Armed conflict, terrorism, and the allocation of foreign aid', The Economics of
Peace and Security Journal, vol. 8, no. 1.
40
Mahadea, D & Kaseeram, I 2018, 'Impact of unemployment and income on entrepreneurship
in post-apartheid South Africa: 1994–2015', The Southern African Journal of Entrepreneurship
and Small Business Management, vol. 10, no. 1, pp. 1-9.
Mallik, G 2008, 'Foreign Aid and Economic Growth: A Cointegration Analysis of the Six Poorest
African Countries', Economic Analysis & Policy, vol. 38, no. 2.
vol. 2, no. 1.
Mensch, BS, Chuang, EK, Melnikas, AJ & Psaki, SR 2019, 'Evidence for causal links between
education and maternal and child health: systematic review', Tropical Medicine & International
Health, vol. 24, no. 5, pp. 504-522.
Mishra, P & Newhouse, D 2009, 'Does health aid matter?', Journal of Health Economics, vol.
28, no. 4, pp. 855-872.
Mosley, P, Hudson, J & Verschoor, A 2004, 'Aid, poverty reduction and the ‘new
conditionality’', The Economic Journal, vol. 114, no. 496, pp. F217-F243.
Macmillan, UK.
Ndikumana, L & Pickbourn, L 2017, 'The impact of foreign aid allocation on access to social
services in sub-Saharan Africa: the case of water and sanitation', World Development, vol. 90,
pp. 104-114.
Oaxaca, R 1973, 'Male-female wage differentials in urban labor markets', International

Economic Review, vol., pp. 693-709.
Pandolfelli, LE, Shandra, J & Tyagi, J 2014, 'The international monetary fund, structural
adjustment, and women's health: a cross-national analysis of maternal mortality in Sub-
Saharan Africa', The Sociological Quarterly, vol. 55, no. 1, pp. 119-142.
Papachrisanthou, MM & Davis, RL 2019, 'The resurgence of measles, mumps, and pertussis',
The Journal for Nurse Practitioners, vol. 15, no. 6, pp. 391-395.
Papanek, GF 1972, 'The effect of aid and other resource transfers on savings and growth in
less developed countries', The Economic Journal, vol. 82, no. 327, pp. 934-950.
41
Pickbourn, L & Ndikumana, L 2016, 'The impact of the sectoral allocation of foreign aid on
gender inequality', Journal of International Development, vol. 28, no. 3, pp. 396-411.
Pickbourn, L & Ndikumana, L 2019, 'Does Health Aid Reduce Infant and Child Mortality from
Diarrhoea in Sub-Saharan Africa?', The Journal of Development Studies, vol. 55, no. 10, pp.
2212-2231.
Pieters, J & Rawlings, S 2020, 'Parental unemployment and child health in China', Review of
Economics of the Household, vol. 18, no. 1, pp. 207-237.
Pring, C & Vrushi, J 2019, Global Corruption Barometer: Africa 2019, Afrobarometer, Ghana.
Raghupathi, V & Raghupathi, W 2020, 'The influence of education on health: an empirical

assessment of OECD countries for the period 1995–2015', Archives of Public Health, vol. 78,
pp. 1-18.
Rashad, AS & Sharaf, MF 2018, 'Economic growth and child malnutrition in Egypt: new
evidence from national demographic and health survey', Social Indicators Research, vol. 135,
no. 2, pp. 769-795.
Rasschaert, F, Pirard, M, Philips, MP, Atun, R, Wouters, E, Assefa, Y, Criel, B, Schouten, EJ

& Van Damme, W 2011, 'Positive spill‐over effects of ART scale up on wider health systems
development: evidence from Ethiopia and Malawi', Journal of the International AIDS Society,
vol. 14, pp. S3-S3.
Rosenstein-Rodan, PN 1943, 'Problems of industrialisation of eastern and south-eastern

Europe', The Economic Journal, vol. 53, no. 210/211, pp. 202-211.
Ross, CE, Masters, RK & Hummer, RA 2012, 'Education and the gender gaps in health and
mortality', Demography, vol. 49, no. 4, pp. 1157-1183.
Safarpour, H, Fooladlou, S, Safi-Keykaleh, M, Mousavipour, S, Pirani, D, Sahebi, A, Ghodsi,

H, Farahi-Ashtiani, I & Dehghani, A 2020, 'Challenges and barriers of humanitarian aid
management in 2017 Kermanshah earthquake: a qualitative study', BMC Public Health, vol.
20, no. 1, pp. 1-10.
Samson, M, Fajth, G & François, D 2016, 'Cognitive capital, equity and child-sensitive social
protection in Asia and the Pacific', BMJ Global Health, vol. 1, no. Suppl 2, pp. i19-i26.
Sharma, R 2018, 'Health and economic growth: Evidence from dynamic panel data of 143
years', PloS One, vol. 13, no. 10, p. e0204940.
Shleifer, A 2009, 'Peter Bauer and the failure of foreign aid', Cato Journal, vol. 29, p. 379.
Stein-Zamir, C & Israeli, A 2017, 'Age-appropriate versus up-to-date coverage of routine

childhood vaccinations among young children in Israel', Human Vaccines &
Immunotherapeutics, vol. 13, no. 9, pp. 2102-2110.
Steurs, L 2019, 'European aid and health system strengthening: an analysis of donor
approaches in the DRC, Ethiopia, Uganda, Mozambique and the global fund', Global Health
Action, vol. 12, no. 1, p. 1614371.
42
Stuckler, D, Basu, S & McKee, M 2011, 'International Monetary Fund and aid displacement',
International Journal of Health Services, vol. 41, no. 1, pp. 67-76.
Tayefi Nasrabadi, M, García, EH & Pourzakarya, M 2021, 'Let children plan neighborhoods
for a sustainable future: a sustainable child-friendly city approach', Local Environment, vol. 26,
no. 2, pp. 198-215.

Tierney, MJ, Nielson, DL, Hawkins, DG, Roberts, JT, Findley, MG, Powers, RM, Parks, B,
Wilson, SE & Hicks, RL 2011, 'More dollars than sense: Refining our knowledge of
development finance using AidData', World Development, vol. 39, no. 11, pp. 1891-1906.
Titeca, K & Edmond, P 2019, 'The political economy of oil in the Democratic Republic of Congo
(DRC): Corruption and regime control', The Extractive Industries and Society, vol. 6, no. 2, pp.
542-551.
UNICEF 2019, WHO and UNICEF warn of a decline in vaccinations during COVID-19,
UNICEF, New York, <https://www.unicef.org/press-releases/who-and-unicef-warn-decline-
vaccinations-during-covid-19>.
UNICEF 2020a, Declining vaccination rates in the Democratic Republic of the Congo could
lead to resurgence in deadly diseases, UNICEF, <https://www.unicef.org/press-
releases/declining-vaccination-rates-democratic-republic-congo-could-lead-resurgence-
deadly>.
UNICEF 2020b, Water, Sanitation and Hygiene, UNICEF, The Democratic Republic of Congo.
Wagner, Z, Heft-Neal, S, Bhutta, ZA, Black, RE, Burke, M & Bendavid, E 2018, 'Armed conflict
and child mortality in Africa: a geospatial analysis', The Lancet, vol. 392, no. 10150, pp. 857-
865.
156.
Wilson, SE 2011, 'Chasing success: health sector aid and mortality', World Development, vol.
39, no. 11, pp. 2032-2043.
Wolf, S 2007, 'Does aid improve public service delivery?', Review of World Economics, vol.
143, no. 4, pp. 650-672.


<https://databank.worldbank.org/reports.aspx?source=World-Development-Indicators#>
Yiew, T-H & Lau, E 2018, 'Does foreign aid contribute to or impeded economic growth?',
Journal of International Studies vol. 11, no. 3.
43
Yu, S 2018, 'Uncovering the hidden impacts of inequality on mental health: a global study',
Translational Psychiatry, vol. 8, no. 1, pp. 1-10.
Zajacova, A & Lawrence, EM 2018, 'The relationship between education and health: reducing
disparities through a contextual approach', Annual Review of Public Health, vol. 39, pp. 273-
289.
Ziesemer, T 2016, 'The Impact of Development Aid on Education and Health: Survey and New
Evidence for Low‐income Countries from Dynamic Models', Journal of International
44
Chapter 2 – Appendices
Appendix 2.A
Table A2.1 – District-level aid projects in the DRC
Nature of aid project Number of aid projects

Agriculture 97
Education 12
Emergency response 54
Energy generation and supply 15
Fishing 7
Forestry 8
General environmental protection 63
Government and civil society 181
Industry 2
Health 201
Mineral resources and mining 9
Population policies/ programmes and reproductive health 20
Trade policy and regulations 52
Transport and storage 77
Water supply and sanitation 74
Other 14
Source: Authors’ computation using data from AidData projects
45
Appendix 2.B
Table A2.2 – Summary statistics
Variable Description Mean SD
BCG 1 if child is vaccinated against BCG 0.76 0.427
DPT-1 1 if child is vaccinated against DPT-1 0.735 0.442
DPT 1 if child is vaccinated against DPT-1 or DPT-2 or DPT-3 0.737 0.441
Polio-0 1 if child is vaccinated against Polio-0 0.44 0.496
Polio 1 if child is vaccinated against Polio-0 or Polio-1 or Polio-2 or Polio-3 0.863 0.344
Measles 1 if child is vaccinated against Measles 0.616 0.486
Vitamin A 1 if child is vaccinated against Vitamin A deficiency 0.463 0.499
Anaemia children 1 if child is anaemic 0.62 0.485
Child mortality 1 if child died before reaching five years of age 0.067 0.25
Mother's age Mother's age in years 30.056 7.431
Mother's age-squared Mother's age squared 958.611 465.848
Mother primary 1 if the mother's highest education level is primary level 0.43 0.495
Mother secondary 1 if the mother's highest education level is secondary level 0.322 0.467
Mother tertiary 1 if the mother's highest education level is tertiary level 0.012 0.11
Household size Number of persons in the household 6.811 3.138
Household head 1 if the household head is currently working 0.739 0.439
employed
Birth order Birth order of the child 4.19 2.687
Male child 1 if child is male 0.499 0.5
Protestant 1 if the household belongs to the Protestant religion 0.193 0.395
Islam 1 if the household belongs to the Islam religion 0.011 0.103
Catholic 1 if the household belongs to the Catholic religion 0.182 0.386
Kimbanguist 1 if the household belongs to the Kimbanguist religion 0.02 0.14
Total aid (5 km) 1 if the household/child is within a 5 km distance from any aid project 0.161 0.367
WSS aid (25 km) 1 if the household/child is within a 25 km distance from a Water & 0.138 0.345
Sanitation aid project
Health aid (25 km) 1 if the household/child is within a 25 km distance from a Health aid 0.222 0.416
project
Agricultural aid (25 km) 1 if the household is within a 25 km distance from an Agricultural aid 0.125 0.331
project
46
Appendix 2.C
Table A2.3 – Effect of economic growth on total aid
(1)
VARIABLES Total Aid
Night-time lights 1.228

(1.560)
Observations 270
R-squared 0.431
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
47
Appendix 2.D
Table A2.4 – Main results
(1) (2) (3) (4) (5) (6) (7)

VARIABLES BCG DPT Polio Measles Vitamin A Anaemia Child
children mortality
Total aid (25km) 0.105*** 0.116*** 0.033*** 0.049*** 0.105*** -0.025*** -0.004*
(0.005) (0.005) (0.004) (0.006) (0.005) (0.006) (0.002)
Birth order -0.009*** -0.008*** -0.004*** -0.023*** -0.013*** -0.003** 0.005***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001)
Male child -0.003 0.001 -0.001 -0.009** -0.014*** 0.038*** 0.002
(0.004) (0.004) (0.003) (0.005) (0.004) (0.005) (0.002)
Mother primary 0.088*** 0.085*** 0.075*** 0.069*** 0.046*** 0.026*** -0.006**
(0.006) (0.006) (0.005) (0.006) (0.005) (0.006) (0.003)
Mother secondary 0.213*** 0.225*** 0.149*** 0.175*** 0.096*** -0.036*** -0.018***
(0.006) (0.006) (0.005) (0.006) (0.006) (0.007) (0.003)
Mother tertiary 0.279*** 0.307*** 0.197*** 0.192*** 0.138*** -0.170*** -0.039***
(0.009) (0.009) (0.007) (0.019) (0.017) (0.024) (0.005)
Household head 0.024*** 0.022*** 0.038*** 0.039*** 0.013*** -0.002 0.005**
employed
(0.005) (0.005) (0.004) (0.005) (0.004) (0.005) (0.002)
Household size 0.002*** 0.001 0.000 -0.001 -0.000 -0.004*** -0.002***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000)
Mother’s age 0.010*** 0.015*** 0.015*** 0.032*** 0.028*** -0.017*** 0.001
(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.001)
Mother’s age-squared -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** 0.000*** -0.000**
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Protestant -0.016*** 0.009 0.013*** 0.022*** -0.040*** -0.017** -0.003
(0.006) (0.006) (0.005) (0.007) (0.007) (0.007) (0.003)
Islam 0.005 0.015 0.019 0.064*** -0.164*** 0.112*** 0.063***
(0.022) (0.024) (0.016) (0.025) (0.024) (0.022) (0.015)
Catholic 0.060*** 0.078*** 0.023*** 0.092*** 0.007 -0.065*** -0.007***
(0.006) (0.006) (0.005) (0.007) (0.007) (0.007) (0.003)
Kimbanguist -0.055*** -0.010 -0.006 -0.106*** -0.052*** -0.006 -0.001
(0.015) (0.015) (0.011) (0.017) (0.016) (0.017) (0.007)
Night-time lights 0.008*** 0.007*** 0.003*** 0.001 0.016*** -0.002** -0.000
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.069 0.071 0.048 0.048 0.289 0.019 0.006
Notes: Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
48
Appendix 2.E
Table A2.5 – Non-lagged aid and logit model
(1) (2) (3) (4) (5) (6) (7)

VARIABLES BCG DPT Polio Measles Vitamin A Anaemia Children Child Mortality
Panel A: Non-lagged aid

Total aid (25km) 0.120*** 0.131*** 0.038*** 0.054*** 0.129*** -0.027*** -0.005*
(0.005) (0.005) (0.004) (0.006) (0.005) (0.006) (0.002)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

R-squared 0.069 0.071 0.048 0.049 0.285 0.019 0.006
Panel B: Logit model
Total aid (25km) 0.740*** 0.716*** 0.402*** 0.219*** 0.564*** -0.115*** -0.115**
(0.031) (0.029) (0.039) (0.024) (0.025) (0.024) (0.056)
Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

*** p<0.01, ** p<0.05, * p<0.1
49
Appendix 2.F
Table A2.6 – Individual vaccination indicators
(1) (2) (3) (4) (5) (6) (7)

VARIABLES DPT 1 DPT 2 DPT 3 Polio 0 Polio 1 Polio 2 Polio 3
Total aid (25km) 0.124*** 0.139*** 0.134*** 0.211*** 0.019*** 0.031*** 0.028***
(0.005) (0.005) (0.006) (0.006) (0.004) (0.005) (0.006)
Observations 44,370 44,345 44,345 44,719 44,179 43,375 43,375

R-squared 0.070 0.069 0.065 0.086 0.042 0.044 0.030
*** p<0.01, ** p<0.05, * p<0.1
50
Appendix 2.G
Table A2.7 – Parallel trend analysis (balancing approach)
Variable Control Treatment Diff p-value

BCG 0.663 0.788 0.064 0.107
(0.473) (0.409) (0.036)
DPT 0.656 0.776 0.070 0.133
(0.475) (0.417) (0.043)
Polio 0.771 0.842 0.030 0.424
(0.420) (0.365) (0.036)
Measles 0.544 0.656 0.072 0.074
(0.498) (0.475) (0.036)
Vitamin A 0.091 0.104 0.007 0.692
(0.288) (0.306) (0.017)
Anaemia Children 0.708 0.624 -0.061 0.143
(0.455) (0.484) (0.039)
Child mortality 0.099 0.070 -0.006 0.268
(0.299) (0.255) (0.005)
Notes: Standard errors in parentheses.
51
Appendix 2.H
Table A2.8 – Parallel trend analysis (supplemental analysis)
(1) (2) (3) (4) (5) (6) (7)

VARIABLES BCG DPT Polio Measles Vitamin A Anaemia Children Child Mortality
Panel A: Households outside the 25km radius in 2007
Total aid (25km) 0.044 -0.006 0.007 -0.005 0.012 -0.010 -0.022
(0.045) (0.050) (0.046) (0.057) (0.029) (0.056) (0.022)
Observations 5,413 5,341 5,465 5,363 5,515 5,205 5,888

R-squared 0.446 0.434 0.381 0.321 0.280 0.268 0.095
Panel B: Households outside the 25km radius in 2013–14
Total aid (25km) -0.043 0.032 0.005 -0.011 0.074 -0.045 -0.003
(0.077) (0.084) (0.067) (0.081) (0.069) (0.069) (0.035)
Observations 17,645 17,586 17,651 17,606 17,429 17,413 18,605

R-squared 0.353 0.386 0.257 0.272 0.325 0.276 0.091
*** p<0.01, ** p<0.05, * p<0.1
52
Appendix 2.I
Figure 2.1 – Distribution of DHS clusters and aid projects
Source: Author’s computation using QGIS 3.12
53
Chapter 3
Foreign Aid and Energy Poverty: Sub-National Evidence from Senegal
3.1 Introduction
Energy poverty, typically defined as households’ inability to afford or access energy,

is a pressing issue for developing and developed countries. Approximately 40% of the
world’s population, most of whom live in developing countries, still rely on traditional
energy sources (e.g., coal, animal waste and wood), and 13% of the population remain
unconnected to electricity (United Nations 2020). Given the costs associated with
establishing infrastructure for electricity (e.g., grid extension), utility companies in
developing countries have little incentive to install electricity lines, especially in rural
areas, because the population is sparsely distributed, increasing the marginal cost of
electricity installation (Baldwin et al. 2015). Most households in such areas tend to rely
on traditional energy sources, with a few drawing on energy from off-grid energy
alternatives, such as solar.
Over the past decade, off-grid solar solutions have played a pivotal role in improving
access to energy to millions of households, especially in developing countries (World
Bank 2020b). However, greater investment is required for the off-grid solar sector to
significantly contribute to universal access to electricity by 2030 (World Bank 2020b).
Put differently, despite the significant progress that has been made, millions of
households remain in energy poverty. In developed countries, where household
reliance on traditional energy sources is relatively low, energy poverty still poses
significant challenges. For instance, evidence suggests that over 3.8 million
households in France are energy poor, while in the United Kingdom (UK), the United
States (US), and Australia, energy poverty rates have been shown to be as high as
10%, 12% and 14%, respectively (Burlinson et al. 2018; Legendre & Ricci 2015; Mohr
2018; Nance 2013; Thomson et al. 2016).
The implications of energy poverty have been well-documented. For instance, air
pollution resulting from the use of traditional sources of energy has been linked with
high mortality rates. An average of 4.3 million people dies each year due to indoor
pollution from using these traditional solid fuels (Austin & Mejia 2017). In other
contexts, some studies have shown that energy poverty undermines health, wellbeing,
54
and homeownership (see, e.g., Awaworyi Churchill et al. 2020; Llorca et al. 2020;
Munyanyi et al. 2021; Prakash & Munyanyi 2021; Thomson, H, Snell, C, et al. 2017).
At the same time, as the energy poverty problem continues to grow, many developing
countries have been beneficiaries of large amounts of foreign aid, some of which has
been targeted towards achieving sustainable energy goals. Across developing
countries, billions of dollars in foreign aid are spent annually on energy (OECD 2022a),
yet millions of people, for instance, in Africa, still have no access to modern energy
services (Ouedraogo 2017), including access to electricity and clean cooking fuels
(IEA 2020a). Importantly, evidence specifically on the impact of foreign aid on energy
poverty is scarce.
We examine the effect of foreign aid on energy poverty. To do so, we match geo-
referenced sub-national foreign aid and DHS datasets for Senegal and apply a DID
approach. We use the information on total aid and sectoral aid and measure energy
poverty multidimensionally. We identify the location of aid projects and examine if
proximity to different aid projects influences energy poverty. Using data on aid projects
over the period 1994 to 2013 and household information from five waves of the DHS,
we find that foreign aid has a negative effect on energy poverty. Our results are robust
to the use of different distance buffers, different types of aid, alternative cut-off points
for the multidimensional measure of energy poverty, and alternative indicators of
energy poverty. Additionally, we test several channels through which aid may influence
energy poverty. In particular, we examine income poverty, economic growth,
education, and health as potential channels and find that income poverty, economic
growth, and education are channels through which aid transmits to energy poverty.
We contribute to the literature by investigating the impact of foreign aid on energy

poverty. Examining this relationship is important, given that addressing energy poverty
requires better insight into the factors that influence it. By examining this relationship,
we complement the small literature that has studied the determinants of energy
poverty (see, e.g., Ambrose 2015; Awaworyi Churchill & Smyth 2020; Azpitarte et al.
2015; Boardman 2010; Bouzarovski & Tirado Herrero 2017). This literature has, thus
far, focused on factors such as income, household dwelling, socioeconomic
characteristics, and energy efficiency, with much emphasis on developed countries.
For instance, Bouzarovski and Tirado Herrero (2017) show that energy prices and
55
expenditures significantly influence energy poverty in the Czech Republic, Hungary
and Poland. Azpitarte et al. (2015) link energy poverty to the socioeconomic profiles
of households in Australia. Moving away from these traditional factors, including
energy prices and socioeconomic characteristics, Awaworyi Churchill and Smyth
(2020) show that sociocultural factors, such as ethnic diversity, are important
determinants of energy poverty in Australia. We make an important contribution to the
literature by diverging from the focus on developed countries and providing evidence
for an important developing country. In addition, we show that foreign aid is an
important factor that can help reduce energy poverty; thus, in devising policies to
address energy poverty, this factor should be taken into account.
Our study is also related to the large body of literature that has examined the impact
of foreign aid on different outcomes. This literature has examined the impact of aid
inflows on economic growth (see, e.g., Feeny 2005; Feeny & McGillivray 2010; Galiani
et al. 2017; Gyimah‐Brempong et al. 2012), health outcomes (see, e.g., Banchani &
Swiss 2019; Feeny & Ouattara 2013; Pickbourn & Ndikumana 2019), and various
development outcomes (see, e.g., Alvi & Senbeta 2012; Arndt et al. 2015; Munyanyi
et al. 2020). However, no study has investigated the effect of foreign aid on energy
poverty. Further, this literature has mostly provided evidence using cross-country
analyses based on aggregate statistics, which can have the limitation of failing to
control for unobservable differences across countries, leading to spurious
relationships between aid and many outcomes (Kotsadam et al. 2018).
Our study contributes to the small number of studies that examine how aid affects
household outcomes using sub-national or microdata (see, e.g., Kotsadam et al. 2018;
Marty et al. 2017; Odokonyero et al. 2018). The use of sub-national data, as opposed
to national-level data, has the advantage of allowing us to examine the efficacy of aid
exactly where it has been geographically established. Much of what is known about
the existing literature on aid efficacy at the sub-national level is limited to Nigeria,
Malawi, and Uganda, and focuses on health outcomes (see, e.g., De & Becker 2015;
Kotsadam et al. 2018; Marty et al. 2017; Odokonyero et al. 2018).
Overall, while existing literature has focussed on the impact of aid and explored
various determinants of energy poverty, the potential effect of aid on energy poverty
has not been explored. We argue that foreign aid can influence various household
56
outcomes that can alter the probability of being in energy poverty. Thus, in addition to
examining the direct effects of foreign aid on household energy poverty, we also
examine several channels through which it is likely to occur.
The rest of the chapter is organised as follows. Sections 3.2 and 3.3 discuss the
related literature and channels through which aid may influence energy poverty,
respectively. Section 3.4 explains why the study is positioned in Senegal, while
Section 3.5 describes the data and variables used. Sections 3.6, 3.7 and 3.8 present
the methodology, discuss the study’s findings, and conclude, respectively.
3.2 Related Literature
Given the correlation between energy poverty and income poverty, this section
discusses the literature on aid and income poverty, together with related literature that
has examined effects on energy poverty using alternative measures or proxies.
3.2.1 Aid and income poverty
Some studies regard energy poverty as an indication of income poverty. Thus,

households with low income or those who live in income poverty are also likely to be
in energy poverty (see, e.g., Pye et al. 2015; Rademaekers et al. 2016). Although there
is some merit to this argument, this may not always be the case, given that energy
poverty is a combination of income poverty, high energy bills, inefficient homes, poor
local area infrastructure and environmental conditions. For instance, lack of access to
electricity, which may be an issue with local area infrastructure, is not necessarily a
reflection of household income poverty. However, a well-established literature has
demonstrated that income poverty is a major determinant of energy poverty (see, e.g.,
Ashagidigbi et al. 2020; Ningi et al. 2020; Qurat-ul-Ann & Mirza 2021).
The relationship between aid and income poverty is well-documented in the literature
(see, e.g., Alvi & Senbeta 2012; Arndt et al. 2015; Chong et al. 2009; Kaya et al. 2013).
However, whether foreign aid is negatively or positively associated with income
poverty is ambiguous. On the one hand, aid inflows have been associated with higher
income poverty (see, e.g., Arvin & Barillas 2002). Here, it is argued that without an
enabling environment that ensures allocated aid is appropriately managed, aid is likely
to create an overreliance on foreign support that eventually worsens poverty. For
instance, evidence suggests that aid is more effective in a good institutional
57
environment and with proper donor coordination (see, e.g., Bigsten & Tengstam 2015;
Bourguignon & Platteau 2017; Ijaiya & Ijaiya 2004; Minasyan et al. 2017), although in
other contexts, even with the support of democracy and strict donor monitoring, aid
fails to improve outcomes, including eradicating poverty (Briggs 2012). According to
Briggs (2012), the failure of aid to effect positive changes is due to government
interference in the allocation and establishment of aid projects. Notably, aid recipient
governments have been found to exploit aid for political reasons – strategically
allocating foreign aid to secure votes (Brass 2012; Briggs 2012). In the end, aid does
not flow to areas where it is most needed.
On the other hand, foreign aid has also been linked with lower income poverty (see,
e.g., Alvi & Senbeta 2012; Arndt et al. 2015; Bahmani‐Oskooee & Oyolola 2009;
Hirano & Otsubo 2014). The effect of aid is also more pronounced when targeted
towards specific sectors (see, e.g., Kaya et al. 2013; Mosley & Suleiman 2007). For
instance, agricultural sectorial aid reduces income poverty by increasing per capita
income growth and improving policies aimed at supporting the poor (Kaya et al. 2013).
3.2.2 Aid and energy-related poverty
One of the adverse effects of energy poverty is increased pollution due to the use of
traditional energy sources (Austin & Mejia 2017). Thus, the use of traditional energy
sources has been used as a proxy for energy poverty (Barnes et al. 2011). Several
macro-level studies have documented evidence on the impact of foreign aid on
pollution and the generation and use of renewable energy sources (see, e.g., Häfele
2017; Kablan 2018; Kim 2018; Mahalik et al. 2021; Mak Arvin & Lew 2009). For
instance, foreign aid has a positive and significant effect on electricity output – with a
larger impact observed in richer countries (Häfele 2017), given that aid
disproportionately flows to wealthier areas (Briggs 2017; Briggs 2021). Briggs (2017)
argues that aid flows to richer areas as opposed to poorer areas for at least three
reasons: 1) donors may not actually care where aid is allocated within aid recipient
countries, 2) donors may care about sub-national aid allocation but have no sound
information on where the most vulnerable live, and 3) sometimes donors are
manipulated by aid recipient governments that may have other agendas other than
reaching the poor.
58
Evidence suggests that aid reduces air pollution (see, e.g., Mahalik et al. 2021; Mak
Arvin & Lew 2009; Sharma et al. 2019) by supporting investments in renewable and
cleaner energy sources (Kablan 2018). Thus, foreign aid bridges financial gaps and
encourages the use of environmentally friendly energy sources (Mak Arvin & Lew
2009). Some evidence suggests that the effect of aid on pollution and adoption of
clean energy depends on the type of donor, such that multilateral aid tends to have
more pronounced effects than bilateral aid (Boly 2018). Given that multilateral donors
have voting arrangements that prohibit any aid recipient country’s government from
forcing its preferences during the aid allocation process, these multilateral donors are
better than bilateral donors at directing foreign aid to poor areas – hence the more
pronounced impact of multilateral aid (Briggs 2017).
Aid for technical cooperation helps transfer clean technologies to developing

countries, which encourages the production and use of clean fuels (Kim 2018). In
another context, aid has been shown to positively influence renewable energy
development in the early stages of social structural change (e.g., urbanisation) and
technical progress; however, as these social and economic structures develop beyond
a threshold, aid is likely to hinder renewable energy development (Wang et al. 2021).
Overall, these studies have mostly focused on macro-level dynamics and do not
provide nuanced evidence on how aid allocation at the local-level influences
household energy poverty. The current study fills this knowledge gap.
3.3 Channel Analysis: Why Should Foreign Aid Affect Energy Poverty?
Conceptually, foreign aid can positively or negatively influence energy poverty,

depending on the transmission channel. In what follows, we discuss how foreign aid
is linked to energy poverty and the potential mechanisms through which aid may
influence energy poverty.
3.3.1 Income poverty
The link between energy poverty and income poverty (see, e.g., Awaworyi Churchill &
Smyth 2020; Boardman 1991; Bouzarovski & Petrova 2015; Pachauri & Spreng 2011;
Sadath & Acharya 2017) suggests that foreign aid could influence energy poverty via
its effect on income poverty. However, the new effect via the income poverty channel
is ambiguous, given that aid may either increase or decrease income poverty.
59
3.3.2 Human capital and health
Aid improves education and health outcomes. In particular, aid increases school
enrolment rates (see, e.g., Christensen et al. 2011; Dreher & Lohmann 2015; Dreher
et al. 2008; Michaelowa & Weber 2006; Riddell & Niño-Zarazúa 2016), reduces
repetition rates (d’Aiglepierre & Wagner 2013) and promotes quality education and
continuous learning (Yogo 2017). By making funds available to develop the education
sector of many countries, aid significantly boosts human capital development (Wolf
2007).
Healthwise, aid enhances prenatal and postnatal care, which in turn reduces maternal
mortality (Banchani & Swiss 2019). Similarly, by improving health infrastructure (e.g.,
antiretroviral therapy and bed nets), financing health education (Bendavid &
Bhattacharya 2014), and supplementing the public health expenditure budget
(Pickbourn & Ndikumana 2019), aid lowers infant and child mortality. Households that
are located closer to aid projects enjoy improved health outcomes (Kotsadam et al.
2018). In vulnerable communities with a higher prevalence of diseases, aid improves
the coverage of treatment drugs (Rasschaert et al. 2011), enhances health care quality
(Marty et al. 2017) and quickens sickness-recovery time (Odokonyero et al. 2018).
Due to the effects of aid on the prevalence of diseases, aid boosts life expectancy
(Bendavid & Bhattacharya 2014).
Improved health and education outcomes that result from aid inflows increase the
probability of higher income from well-paying jobs, making it easier to afford energy
services (Njiru & Letema 2018).
3.3.3. Institutional quality
Aid undermines the institutional quality of recipients. Specifically, aid weakens

governance (see, e.g., Asongu & Nwachukwu 2016; Bräutigam & Knack 2004) and
triggers corruption when aid-recipient governments are given complete control over
the use of the donated funds (Asongu & Nwachukwu 2016). In particular, aid is found
to promote local corruption around active aid projects (Isaksson & Kotsadam 2018)
and has been used to elicit favours that involve corruption from the leaders of
developing countries (for a review, see Tan‐Mullins et al. 2010; Tull 2006). Given that
the execution of aid projects is often characterised by discretion, it remains vulnerable
60
to corruption (Tavares 2003) and rent dissipation, which ultimately results in reduced
productive public expenditure (Svensson 2000).
Heightened corruption often weakens informal institutions such as trust, and with
formal and informal institutions compromised, the provision of public goods, such as
energy services, can be curtailed (Awaworyi Churchill & Smyth 2020).
3.3.4 Aggregate economic activity/growth
Aid promotes economic growth (see, e.g., Dollar & Burnside 2000; Doucouliagos &
Paldam 2008; Galiani et al. 2017; Gyimah‐Brempong et al. 2012; Minoiu & Reddy
2010). This growth-enhancing effect of aid is even more pronounced in countries with
desirable conditions, including good fiscal, monetary and trade policies (Dollar &
Burnside 2000), good climate-related circumstances (Dalgaard et al. 2004), and good
aid absorption capacity (Collier et al. 2002).
Early theorists posit that aid provides the capital required to trigger self-sustaining
growth in aid-recipient countries (see, e.g., Lewis 1954; Rosenstein-Rodan 1943). In
this regard, aid kickstarts economic growth and triggers a cycle in which investments
create income and boost the economic return to be used to further investment (Shleifer
2009). Stable economic growth and high economic growth rates associated with aid
(see, e.g., Collier & Goderis 2009; Guillaumont & Wagner 2014) often lead to high
investment (Iamsiraroj 2016), lower unemployment (see, e.g., Aghion & Howitt 1994;
Bartolucci et al. 2018; Maitah et al. 2015), and high entrepreneurship levels (Mahadea
& Kaseeram 2018), all of which increase wage-earning opportunities and
consequently lower energy poverty. Put differently, in an environment characterised
by stable growth, high investments, low unemployment and more entrepreneurship
opportunities, many individuals would be expected to have greater access to, or at
least afford better energy services.
Aid, though, need not promote economic growth. Aid has been linked to poorer
economic growth (see, e.g., Liew et al. 2012; Mallik 2008; Moyo 2009). Aid
undermines capital accumulation and labour supply in the long run (Gong & Zou 2001),
creates aid dependency, corruption and market distortions (see, e.g., Brazys 2018;
Moyo 2009; Yiew & Lau 2018), and weakens governance (Bräutigam & Knack 2004),
61
all of which are relevant for economic growth. If this is the case, aid is expected to
exacerbate energy poverty via the economic growth channel.
3.3.5 Other channels
Aid may also influence energy poverty through other channels, such as inequality and
energy efficiency. Aid improves income equality (Bourguignon et al. 2009) and
reduces gender inequality (see, e.g., Baliamoune‐Lutz 2016; Pickbourn & Ndikumana
2016). Education aid narrows the male-female literacy gap (Pickbourn & Ndikumana
2016) and increases women’s participation in politics (Baliamoune‐Lutz 2016). By
ensuring both income and gender equality, aid promotes equal opportunities and
access to resources, including energy services.
Aid increases energy efficiency (Maruta & Banerjee 2020) and reduces energy
intensity (Kretschmer et al. 2013). Energy efficiency refers to using less energy to
perform tasks – that is, alleviating energy waste (IEA 2020b). Aid increases energy
efficiency by building and renovating energy technologies in aid-recipient countries
(Maruta & Banerjee 2020). Targeted aid reduces energy intensity and improves
energy efficiency by providing financial resources to develop infrastructure and
technical knowledge to do so (Kretschmer et al. 2013). By promoting energy efficiency,
aid alleviates energy poverty, given that energy inefficiency is one of the causes of
energy poverty (Mattioli et al. 2017).
3.4 Why Senegal?
We situate our study in Senegal for three main reasons. First, Senegal is a major
foreign aid-receiving country and has been an aid recipient since the early 1960s –
receiving over US$40 billion in aid as of 2020 (World Bank 2020a). In terms of aid
magnitude, Senegal receives more aid than other countries in the region, particularly
its neighbouring countries such as Guinea, Guinea-Bissau, and Mauritania. Between
2002 and 2020, Senegal received about US$11 billion in total ODA from DAC
members5, while Guinea, Guinea-Bissau and Mauritania received about US$3 billion,
US$900 million, and US$2 billion, respectively (OECD 2022a). Additionally, for the
5DAC members include Australia, Austria, Belgium, Canada, Denmark, the European Union, Finland,
France, Germany, Greece, Ireland, Italy, Japan, South Korea, Luxembourg, the Netherlands, New
Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the UK, and the US.
62
same period, the DAC members disbursed foreign aid for the energy sector of
approximately US$281 million, US$9 million, US$3 million, and US$33 million to
Senegal, Guinea, Guinea-Bissau, and Mauritania, respectively. Thus, the combined
value of aid received by Guinea, Guinea-Bissau and Mauritania is less than what
Senegal alone received when considering total aid as well as aid to the energy sector.
In the same context, international financial institutions, such as the World Bank, have
also disbursed financial assistance to Senegal over many years. In 2019, the World
Bank approved US$180 million geared towards improving Senegal’s energy sector,
promoting sustainable energy, enhancing the performance of the electricity utility,
lowering the cost of electricity and boosting access to energy services by the most
vulnerable (World Bank 2020d). Thus, in addition to being one of the major aid-
receiving countries, Senegal has also benefited substantially from aid to the energy
sector, making it an important case study to examine how aid has influenced energy
poverty.
Second, Senegal has witnessed some interesting transformations with regard to

access to modern energy. In particular, Senegal has been a leader in sub-Saharan
Africa on reforming the energy sector – being one of the first African countries to open
the energy sector to private participation coupled with the establishment of some of
the largest renewable energy projects in Africa (USAID 2021b). It is among the five
African countries, together with Kenya, Ghana, Rwanda, and Ethiopia, that
significantly contributed to the increase in the continent’s electricity coverage between
2013 and 2019. Over this period, the number of people without access to electricity
fell from 610 million to 580 million (IEA 2019a).
As part of a comprehensive strategy known as the Plan Senegal Emergent, the

Senegalese government has prioritised improving its energy sector, lowering the cost
of power generation, reducing reliance on imported liquid fuels, and increasing access
to electricity (USAID 2021b). Given that the country enjoys more than 2,000 hours of
sunshine annually (GIZ 2021), there has also been an increase in access to electricity
via solar-based energy solutions for those not connected to the grid (Ba 2018). Thus,
between 2000 and 2019, the percentage of the Senegalese population with access to
electricity rose from about 31% to approximately 70%, which is a relatively high
proportion for the sub-Saharan Africa region (IEA 2019a; USAID 2021b). Additionally,
due to financial incentives schemes such as the liquefied petroleum gas (LPG) subsidy
63
program in Senegal, the use of LPG as a cooking fuel is growing significantly (Ba
2018). Specifically, approximately 30% of the population uses LPG, although this is
still relatively low (IEA 2019b).
Despite these remarkable achievements, the overreliance on traditional energy

resources, such as firewood and charcoal, as the main cooking fuels is very high (Ba
2018), and the proportion of the rural population with access to electricity remains
below 50% (USAID 2021b). Thus, with regard to the multidimensional nature of energy
poverty, Senegal has made remarkable progress in some areas, for instance, access
to electricity, but still lags behind in areas such as clean energy use for cooking. These
dynamics make Senegal an interesting case study.
Third, Senegal is regarded as one of the most politically stable African countries
(Vengroff & Creevey 2018). Having gained independence in 1960, Senegal has well-
consolidated democratic institutions (Ba 2018). With such a politically stable
environment and its quality institutions, it is possible to achieve the exact impact of aid
on outcomes without the interference of political instability and weak institutions, which
the literature associates with aid inefficacy (see, e.g., de Ree & Nillesen 2009;
Doucouliagos & Paldam 2008; Lis 2013). Thus, the institutional environment in
Senegal makes it a useful case study to understand the impact of aid without weak
institutional interference.
The data used in this study are from the DHS and AidData projects 6. Data from
AidData are freely available online at the College of William & Mary’s research lab and
contain information on geo-coded development aid projects from 96 donor institutions
from 1945 (Tierney et al. 2011). The database includes data on the projects’ exact
locations by providing geo-codes (i.e., GPS points). It also includes information on the
type of projects to which aid is assigned (e.g., education, water supply and sanitation,
agriculture). At the district level, the dataset used in this study includes Senegal’s aid
projects from 1994 to 2013. Appendix Table A3.1 describes these projects.
64
The DHS dataset consists of household surveys carried out since 1984 in several
countries. The surveys report household demographic details, including health,
mortality and other socioeconomic indicators (ICF 2018). This dataset also contains
geo-codes for all surveyed clusters, which makes it possible to match it with other geo-
referenced datasets such as the AidData information. The study uses five rounds of
the DHS data for Senegal: 1992–1993, 1997, 2005, 2010–2011, and 2012–2013 (ICF
2020). Appendix Table A3.2 reports summary statistics of the variables used.
We geographically match the sub-national aid and demographic surveys datasets.

Consistent with Marty et al. (2017), we include projects with exact geo-codes and at
the district level, enabling us to match households to specific areas where aid projects
have been executed. By confining the analysis to such projects, we focus on projects
that have physical sites and analyse how proximity to project sites influences energy
poverty.
3.5.1 Energy poverty
(a) Measuring energy poverty
The literature has proposed and used various indicators of energy poverty, including
expenditure-based and consensual-based indicators, which, due to the availability of
data, have been widely used in developed countries (see, e.g., Awaworyi Churchill &
Smyth 2020; Biermann 2016; Hills 2012; Robinson et al. 2018; Thomson, H,
Bouzarovski, S, et al. 2017).
Expenditure-based indicators are objective indicators, which capture the cost of

energy services and the ability of households to pay for such services. These
indicators define energy poverty using the information on the household’s energy
expenditures and often focus on the share of energy expenditure relative to income or
absolute expenditure in monetary terms to a normative threshold (Awaworyi Churchill
& Smyth 2021). Here, a household would be considered energy poor if its energy
expenditure exceeds a certain threshold. By applying different thresholds, these
indicators are able to capture the severity of energy poverty (Rademaekers et al.
2016).
There are three types of expenditure-based indicators that have been used in the
literature. First is the high share of energy costs indicator. This indicator considers a
65
household as energy poor if its energy costs, as a share of income or in absolute
terms, lie above a certain threshold. Here, a relatively high share of energy
expenditure in the household budget is often linked with energy poverty. Previous
research has suggested a 10% threshold, and thus, a household that spends more
than 10% of its income on energy is deemed energy poor (see, e.g., Awaworyi
Churchill & Smyth 2020; Boardman 2010; Bouzarovski & Petrova 2015). This indicator
has often been criticised for the arbitrariness of the threshold measure, and thus,
alternative thresholds such as 5%, 15% and 20% have also been used (Awaworyi
Churchill & Smyth 2020). Additionally, this indicator has been criticised for
understating and overstating energy poverty rates.
Second is the insufficient energy spending indicator, which compares household

energy expenditure to a minimum threshold that is deemed necessary for the
household to enjoy basic energy services. A common indicator in this tradition is the
Hidden Energy Poverty indicator, which is an objective energy poverty indicator that
considers a household energy poor if its absolute energy expenditure is below half the
national median energy expenditure (see, e.g., Munyanyi et al. 2021; Rademaekers
et al. 2016). Using this indicator makes it possible to identify households experiencing
difficulties in meeting the minimum energy consumption threshold.
Third is the low available income indicator, which classifies a household as energy
poor if, after energy costs, the household is left with little income (i.e., these measures
identify households whose incomes, after covering energy costs, lie below a certain
threshold) (Hills 2012). Hills (2012) proposes the low income-high cost (LIHC) indicator
of energy poverty, which takes into account the high costs of energy and the low-
income status of households. Using the LIHC indicator, households are energy poor if
they report high energy costs above the median and their equivalised income after
energy-related expenditure is below the official poverty line.
Contrary to the objective indicators of energy poverty, consensual-based indicators

are subjective. They are indicators derived from household perspectives on their ability
to meet basic energy services. These subjective measures are easier to implement,
have the benefit of reflecting how households feel about being energy poor, and can
help us understand perceived energy poverty better than expenditure-based indicators
(see, e.g., Awaworyi Churchill & Smyth 2020; Rademaekers et al. 2016; Thomson, H,
66
Snell, C, et al. 2017). However, these measures can also be unreliable, given that they
depend on respondents’ confidence to acknowledge the inability to heat their homes
or keep them adequately cool when required. Thus, the use of these measures may
understate the true extent of energy poverty.
The literature has also used multidimensional measures of energy poverty that
combine various indicators of energy poverty, including the objective and subjective
measures discussed above (see, e.g., Awaworyi Churchill & Smyth 2020; Munyanyi
et al. 2021; Nussbaumer et al. 2012b; Sadath & Acharya 2017; Zhang et al. 2019).
Besides the dimensions of affordability captured in most expenditure-based measures

of energy poverty, the literature has also focused on the energy access dimension of
energy poverty (see, e.g., Acheampong et al. 2021; Poblete-Cazenave & Pachauri
2021). This is typically the case in developing countries where access to clean energy
sources is a problem. Here, the most basic indicator of energy poverty pertains to
access to electricity, which has been a primary target for most policymakers in
developing countries, including Senegal, Kenya, Ghana, Rwanda and Ethiopia (IEA
2019b). Accordingly, over time the share of the population with access to electricity
has increased remarkably. However, the multidimensional nature of energy poverty
has meant that many households in developing countries, even those typically
considered wealthy, are deemed energy poor when it comes to using clean energy
sources for cooking. This is consistent with energy poverty in developed countries,
where depending on the dimension of energy poverty, some households may be
energy poor in one dimension but not in another (Awaworyi Churchill & Smyth 2021).
(b) Measuring energy poverty in Senegal
Unavailability of data on the costs or expenditures related to energy in most surveys

conducted in most developing countries has meant that energy poverty in developing
countries typically focuses on energy access. The focus on energy access in
developing countries is also consistent with the idea that energy access, and in
particular clean energy use, is more of an issue in developing countries than
developed countries (World Bank - ESMAP 2021). Accordingly, the Multidimensional
Energy Poverty Index (MEPI), which reflects the economic conditions and clean
energy adoption rate of developing countries, is often used. The measure of energy
poverty in Senegal follows this tradition.
67
We use the MEPI following Nussbaumer et al. (2012b). We assign relative weights on
five main forms of energy use, including cooking, lighting, communication,
entertainment/education, and connecting household appliances. In addition to
questions that focus on energy access for cooking and lighting, the affordability
dimension in the MEPI is captured via survey questions that collect information on the
inability of households to use or afford basic appliances that require energy. This can
be linked with energy rationing, which has been discussed in the literature as a
component of energy poverty. Thus, households without access to or that are unable
to afford certain basic appliances, such as communication devices, entertainment or
education devices and other electrical devices, are deemed energy poor (Mendoza Jr
et al. 2019; Nussbaumer et al. 2012b; Nussbaumer et al. 2013).
In the context of our study, the DHS asks respondents: 1) if they have access to
electricity, 2) about the type of fuels they use for cooking, 3) if they own a mobile phone
or landline, 4) if they possess electrical appliances such a fridge, and 5) if they own a
TV or radio. A household is considered energy-deprived or poor: 1) if it does not have
access to electricity, 2) if it uses coal, dung, charcoal or agricultural waste for cooking
instead of electricity, LPG, natural gas, or biogas, 3) if it does not own a mobile phone
or landline, 4) if it does not possess a fridge, and 5) if it does not own a TV or radio. A
summary of these five indicators/energy uses and weights applied is provided in
Appendix Table A3.3.
The indicators for cooking, lighting (i.e., electricity), electronic appliances,

communication, and entertainment indicators are assigned weights of 0.4, 0.2, 0.13,
0.13, and 0.13, respectively. These weights are determined considering their relative
importance in the case of Senegal and other developing countries (see, e.g., Koomson
& Danquah 2021; Mendoza Jr et al. 2019; Nussbaumer et al. 2012a). For instance,
compared to the other three indicators, cooking and lighting (i.e., electricity) carry more
weights due to their importance in energy poverty in Africa (see, e.g., Adusah-Poku &
Takeuchi 2019; Nussbaumer et al. 2013). The inability to use clean sources of energy
that are not harmful to health when cooking is considered energy deprivation
(Mendoza Jr et al. 2019; Nussbaumer et al. 2013). In Senegal, even wealthy
households that have access to electricity tend to continue using highly polluting fuels,
such as charcoal and wood for cooking. There are at least two potential reasons
behind this. First, the insufficient electricity generation capacity coupled with the high
68
electricity costs compel households to use other less expensive cooking fuels such as
charcoal and wood (AFDB 2013; Streatfeild 2018). Second, most African countries
use hydroelectric power stations that depend on weather conditions (AFDB 2013).
During periods of bad weather such as heavy flooding or poor rainfall, power
disruptions and shortages occur more frequently (AFDB 2013), resulting in heightened
load shedding (Bwalya Umar et al. 2021; Timilsina & Steinbuks 2021). With such
random load shedding, households are more likely to resort to other energy sources,
including the polluting ones. The higher weight assigned to cooking relative to other
dimensions reflects this deprivation, which is quite pronounced even in the case of
households that are able to afford electricity. Additionally, the higher weight assigned
to cooking reflects the fact that in developing countries such as Senegal, cooking is a
more critical need, and thus energy for cooking is typically assigned a higher weight
than energy for lighting (Koomson & Danquah 2021).
Besides cooking, lighting also weighs higher than the other dimensions. The most
fundamental definition of energy poverty in developing countries such as Senegal is
the inability to access energy; thus, lack of access to electricity is one of the most
important indicators of energy poverty (Nussbaumer et al. 2012b; Nussbaumer et al.
2013). While many wealthy households have access to electricity, several others do
not have access. Frequent power disruptions because of load shedding also mean
that many households, even those connected to the grid, do not often have access to
electricity (Bwalya Umar et al. 2021; Timilsina & Steinbuks 2021). Thus, access to
electricity for lighting also weighs relatively higher because of the inequality in
electricity access. In further checks, to examine the robustness of our results, we
employ alternative weights, including equal weights to each of the indicators.
The five indicators are used to generate the MEPI using a formula akin to the
multidimensional income poverty indicator. The energy deprivation score is a weighted
sum of the number of household deprivations and ranges between 0 and 1 for each
household. The mathematical expression for energy deprivation is given as:
𝐷𝑖 = 𝑤1 𝐼1𝑖 + 𝑤2 𝐼2𝑖 + 𝑤3 𝐼3𝑖 + 𝑤4 𝐼4𝑖 + 𝑤5 𝐼5𝑖 (1)
where 𝐷𝑖 is the household energy deprivation score, 𝐼𝑖 = 1 if a household is deprived

in indicator 𝑖 and 𝐼𝑖 = 0 if otherwise. 𝑤𝑖 is the weight attached to indicator 𝑖 with
∑𝐷
𝑖=1 𝑤𝑖 = 1. A household is considered energy poor if its energy deprivation score is
69
at least 0.5. The study uses two measures: the energy deprivation score (MEPI score)
and the energy deprivation binary variable based on the 0.5 threshold (MEPI status).
Figure 3.1 shows the household clusters that are energy poor.
From Figure 3.1, we observe that energy poverty is clustered in Dakar, the capital city
of Senegal, which is inconsistent with research that shows that energy poverty is
higher in rural areas. One possible explanation for this could be related to
demographic transitions. Specifically, higher populations in urban areas driven by
rural-urban migration mean that many energy and income poor households from rural
areas relocate to urban areas to enhance their livelihoods (FAO 2020; Herrera-
Almanza & Sahn 2020).
3.5.2 Forms of energy support and sub-national aid
Energy support takes many forms, including support from non-governmental

organisations (NGOs), private firms, independent power producers, and multilateral
institutions. Over the years, NGOs have been found to promote small-scale, localised
and environmentally friendly electricity generation technologies, including wind and
hydroelectric turbines, biogas digesters and solar photovoltaic panels (see, e.g., Balint
2006; Hossain et al. 2017; MacLean & Brass 2015). Some of these technologies are
offered for free to the residents, as long as the residents agree to cooperate in local
environmental protection efforts (Balint 2006).
The private sector also plays a significant role in the provision of energy. Given that
electrification is a capital-intensive activity, and the donor community and public sector
have limited resources, the private sector is viewed as a potential source of capital to
finance energy-related projects (Williams et al. 2015). In some countries, despite the
high levels of risk and potentially low returns of such projects, the private sector has
made significant contributions to society by installing decentralised microgrid based
projects to cater for those far away from the grid (MacLean & Brass 2015; Williams et
al. 2015).
Independent power producers (IPPs) – which are licensed power stations constructed,
owned, and managed by investors (Okedu et al. 2018), are increasingly becoming
important contributors to meeting the power needs in developing countries, including
Africa (Eberhard et al. 2018). These IPPs have long-term power purchase agreements
70
with a utility (Eberhard et al. 2018) and are licensed to generate electricity to increase
electricity supply and foster sustainable development (Okedu et al. 2018).
Multilateral financing is also another form of energy support, which involves multilateral
organisations disbursing financial assistance to enhance the energy sector (Tirpak &
Adams 2008). The most common multilateral institutions that have been allocating
billions of dollars in aid for the energy sector in developing countries are the World
Bank, the International Monetary Fund and the DAC (OECD 2022a; World Bank
2020d).
In our analysis, the measures of foreign aid projects do not consider these different
forms of energy support, given that the data source we use contains data on aid only
allocated by donor agencies and multilateral institutions (Tierney et al. 2011). We thus
focus on aid projects funded by donor agencies and multilateral institutions. Consistent
with the literature on aid efficacy at the sub-national level, aid is a dummy variable
(see, e.g., Kotsadam et al. 2018; Marty et al. 2017; Odokonyero et al. 2018) that equals
one if a household lives within a certain distance from an aid project. In line with the
literature, the main buffer distance in this study is 25 kilometres (km) (see, e.g.,
Kotsadam et al. 2018; Odokonyero et al. 2018). That is, households living within a 25
km distance from an aid project are coded 1, otherwise 0. Figure 3.2 shows the
distribution of aid projects and the DHS clusters, together with the household clusters
that fall within the 25 km buffer distance.
3.5.3 Covariates
In accordance with the literature, we control for the characteristics of the household
head. These include educational status, employment status, marital status, residence,
gender and age. We also control for household size and district economic
performance. Educational attainment and employment status influence labour market
conditions that could influence energy poverty (see, e.g., Ashagidigbi et al. 2020; Ningi
et al. 2020; Qurat-ul-Ann & Mirza 2021). These two variables are expected to be
negatively correlated with energy poverty. A similar effect is also expected among 1)
male-headed households – as men have a higher chance of securing employment
opportunities or other sources of income than women (Jayachandran 2021); 2)
households staying in urban areas – given that compared to rural areas, most urban
areas have more reliable access to modern energy fuels such as electricity (Baldwin
71
et al. 2015); 3) married household heads – given that married people tend to strive to
provide adequately for their dependents and family (Ruales 2017); and 4) districts with
enhanced economic performance – as higher levels of economic growth often mean
more employment opportunities and better provision of public services, including clean
energy (Chand et al. 2017; Hung-Pin 2014). A priori, the effect of household size and
age on energy poverty is ambiguous and could go either way. We control for age given
that it has an influence on household choices, including energy choices, while we
control for household size given that it determines the level of household energy
consumption.
3.5.4 Channels of influence
We examine if income poverty, economic growth, education and health status are
channels through which aid influences energy poverty. To measure income poverty,
we use the wealth index, which is a commonly used proxy for household poverty
available in the DHS (see, e.g., Gachanja & Kinyanjui 2016). In the absence of reliable
household income or expenditure data (as in our case), this index is argued to be a
better alternative (Filmer & Pritchett 2001). The measure of income poverty is a
dummy variable that equals 1 if the wealth index score for a given household is
positive, otherwise 0. We measure education using the highest level of education
attained by the household head. Health status is measured using a dummy variable
that equals one if the household head has not had a cough, fever or diarrhoea in the
last two weeks before the interview. Given the absence of measures of gross domestic
product (GDP) per capita at the sub-national level, consistent with the geographic
locations of households and aid project locations, we use satellite night vision data as
proxy for economic growth. Night-time light is increasingly used in the literature as a
measure of economic activity (see, e.g., Ackermann et al. 2021; Appau et al. 2021;
Awaworyi Churchill et al. 2021). We use satellite data, covering the years 1992 to
2013, from the National Oceanic and Atmospheric Administration’s Earth Observation
Group (NOAA 2020).7 The measure of economic growth is the mean of night-time light
per year. Night-time light data are provided in 30 arc-second grids and measured in
bits (Zhao et al. 2019).
7 https://www.ngdc.noaa.gov/
72
3.6 Empirical Methodology
The nature of the AidData dataset enables us to conduct Geospatial Impact

Evaluations (GIEs). Evidence suggests that GIEs create conditions similar to
Randomised Controlled Trials (RCTs) with observational data (AidData 2021;
BenYishay et al. 2017). Notably, like RCTs, GIEs are able to estimate the impact of a
specific treatment or intervention by comparing similar individuals whose only
difference was exposure to the treatment or intervention. However, unlike RCTs, GIEs
employ precise geographic data to establish this counterfactual retroactively, without
randomly assigning individuals/participants to control and treatment groups (AidData
2021).
In practice, to estimate the intended or unintended impacts of an intervention or

treatment, GIEs depend on sub-nationally georeferenced interventions as well as the
outcome, covariate data and quasi-experimental approaches of causal inference
(BenYishay et al. 2017). Some of the quasi-experimental approaches that have been
used in association with GIEs in the literature include the propensity score matching,
DID, and regression discontinuity design (see, e.g., Amare et al. 2018; Dolan et al.
2019; Duque et al. 2018; Heo et al. 2019; Skoufias et al. 2012).
GIEs have many benefits: 1) they enable studies to evaluate the impact of programs
or treatments where it is not feasible or ethically possible to conduct RCTs; 2) they
can be implemented retrospectively and remotely (e.g., they can even be applied in
conflict-torn areas or fragile state settings); 3) they control for omitted variables and
potential confounders at fine geographic levels; and 4) they offer external validity and
generalisability (BenYishay et al. 2017).
Consistent with the work of Marty et al. (2017) on the impact of sub-national foreign
aid, we use a repeated cross-sectional DID strategy. The DID approach is a statistical
method that imitates an experimental research design utilising observational study
data (Abadie 2005; Bertrand et al. 2004). It focuses on the differential impact of a
treatment on a ‘treatment group’ compared to a ‘control group’. The DID method
derives the impact of a treatment on an outcome by comparing the average change
over time in the outcome for the treatment group with the average change for the
control group (Angrist & Pischke 2008). In the context of our study, the DID model
allows us to compare differences in outcomes (i.e., household energy poverty)
73
between the control and treatment groups before and after foreign aid projects (i.e.,
the treatment) have been established.
While DHS surveys do not follow the same individuals or households over time, a
repeated cross-section strategy is ideal because these surveys are nationally
representative (Marty et al., 2017). Using this DID strategy, we test whether the
likelihood of a household facing energy poverty is affected by how close the household
is to an aid project. Aid projects data are available for the period 1994 to 2013, while
data for DHS are from 1992 to 2013. The period before the start of aid projects’
implementation (i.e., 1992 and 1993) will represent the baseline years, and in that
regard, a code 0 is assigned. For 1994 onwards (i.e., aid implementation years), they
will represent the treatment period or end line years, and a code 1 is assigned.
Consistent with the literature, the study estimates the following DID equation:
𝐸𝑃𝑖,𝑗,𝑡 = 𝛼0 + 𝛼1 𝑇𝑖𝑚𝑒 + 𝛼2 𝐴𝑖𝑑𝑖,𝑗,𝑡−1 + 𝛼3 (𝑇𝑖𝑚𝑒 × 𝐴𝑖𝑑𝑖,𝑗,𝑡−1 ) + 𝛼4 𝐸𝐺𝑗,𝑡 (2)

+ 𝜗𝑋𝑖,𝑗,𝑡 + 𝜀𝑖,𝑗,𝑡
where 𝐸𝑃 represents the energy poverty level of household 𝑖 living in district 𝑗 in year
𝑡. As discussed earlier, we capture energy poverty using: 1) a binary variable (i.e.,
MEPI status) equal to one if the household’s energy deprivation score is at least 0.5,
and 2) the total household deprivation score (i.e., MEPI score). 𝑇𝑖𝑚𝑒 is a binary
variable that equals 1 for the aid projects implementation period (i.e., 1994–2013) and
0 for the aid projects pre-implementation period (i.e., 1992 and 1993). 𝐴𝑖𝑑 is a binary
treatment variable equal to one if household 𝑖 in district 𝑗 is living within close proximity
to an aid project. In line with the literature, for the main results, we consider close
proximity to be within a 25 km distance from any aid project (see, e.g., Odokonyero et
al. 2018). That is, households living within the 25 km radius of an aid project are the
treatment group, while those living beyond this specified distance represent the control
group. In further checks, we consider several distance buffers both below and above
the 25 km threshold (e.g., 5, 10, 15, 20, 50, 75 and 100 km). 𝑇𝑖𝑚𝑒 × 𝐴𝑖𝑑 is the
treatment effect, which is derived by creating an interaction between the 𝑇𝑖𝑚𝑒 and 𝐴𝑖𝑑
variables. Given that the effect of aid on most development outcomes is not immediate
(see, e.g., Arshad et al. 2016; Kaya et al. 2013), we use the lag of the aid variable (i.e.,
𝑡 − 1). In further checks, we consider the current period (𝑡) and subsequent lag (𝑡 −
74
2). 𝑋 is a vector of household and individual control variables that may be correlated
with energy poverty, including educational status, employment status, marital status,
gender, age and household size. 𝐸𝐺 represents the economic performance of district
𝑗 in year 𝑡 , captured by night-time light. 𝜀 is the error term. We control for district fixed
effects to account for time-invariant and unobservable factors.
𝛼3 is the coefficient of interest, which measures the effect of aid (i.e., the effect of living
close to an aid project) on energy poverty. To estimate the causal effect of aid using
the DID model, two important conditions need to be met. First, for the model to be
valid, the treatment should be unrelated to the outcome at baseline. In the case of our
study, the establishment of aid projects or allocation of foreign aid must not be
determined by household energy poverty.
Given donors’ criteria for choosing aid project locations, coupled with the manner and
pace in which aid is allocated to aid-recipient countries, we posit that aid allocation is
not influenced by household energy poverty. Specifically, donors are typically the
decision-makers for the aid allocation process. Aid allocations tend to depend on the
within-country processes of donors and are heavily influenced by national budgets, as
well as political and bureaucratic systems in the donor countries. These conditions or
donor country characteristics are independent of energy poverty in the recipient
countries, or in this case, Senegal. Evidence also indicates that the aid allocation
process can even take several years before aid is finally disbursed to aid recipient
countries (Bearce & Tirone 2010). Thus, the economic conditions of aid-receiving
countries can experience significant changes before aid is allocated. These conditions,
which change over time independent of aid allocation, do not influence internal
conditions in donor countries that ultimately determine whether aid is allocated.
A potential threat to our identification strategy is when aid is allocated based on the
aid-recipient country’s level of economic growth – which the literature connects to a
number of development-related outcomes, including energy poverty (see, e.g.,
Acharya & Sadath 2019; Filippidis et al. 2021). To ensure that this is not a problem,
we examine the effect of district-level economic performance on aid allocation. Given
that there is no data on GDP per capita at the district level, we follow the literature that
has used night-time light to measure economic development at the sub-national level
(see, e.g., Ackermann et al. 2021; Appau et al. 2021; Awaworyi Churchill et al. 2021;
75
Hodler & Raschky 2014). The results reported in Appendix Table A3.4 show no
correlation between district-level economic performance and energy poverty in
Senegal.
The second condition for the DID model is that of internal validity. To ensure the
internal validity of the DID model, the treatment and control groups must have parallel
trends in outcomes. This implies that in the absence of foreign aid, the difference in
outcomes between the two groups should be constant over time. We demonstrate the
fulfilment of this assumption by running a balancing test, showing that the difference
in outcomes between the control and treatment groups is not statistically significant at
baseline conditions.
3.7 Results
3.7.1 Main results
Table 3.1 presents the main results on the effect of sub-national foreign aid on energy
poverty among households. Columns 1 and 2 of Table 3.1 report bivariate regression
results, whereas Columns 3 and 4 show multivariate regression results. These results
show, in terms of coefficient signs and statistical significance, that aid has an identical
impact on both measures of energy poverty (i.e., MEPI status and MEPI deprivation
score).

(1) (2) (3) (4)
VARIABLES MEPI status MEPI score MEPI status MEPI score
Time -0.014** -0.012*** -0.000 -0.003*

(0.005) (0.002) (0.005) (0.002)
Total Aid (25 km) -0.016 -0.039*** 0.033*** -0.009*
(0.012) (0.005) (0.011) (0.005)
Time*Total Aid (25 km) -0.032*** -0.009* -0.033*** -0.009**
(0.011) (0.005) (0.010) (0.004)
Controls No No Yes Yes

Observations 17,227 17,227 17,227 17,227
R-squared 0.130 0.269 0.224 0.421
Notes: ‘MEPI status’ represents energy deprivation dummy, and ‘MEPI score’ stands for energy
deprivation score. All regressions include district fixed effects. Controls include educational status,
residence, age, household size, marital status, employment status, gender and economic performance.
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
76
The bivariate regression results suggest that aid has a negative and statistically
significant effect on energy poverty. Specifically, from Column 1, we find that aid
reduces the likelihood of being energy-poor by 3.2 percentage points among
households staying within a 25 km distance of an aid project. From Column 2, the
results show that moving from outside of the aid treatment zone into the treatment
zone (i.e., within the 25 km radius of an aid project) generates a 0.009-unit decrease
in energy poverty among households. In Columns 3 and 4, where we control for the
relevant covariates, we find that the effect of aid and the direction of the effect remain
consistent. The complete set of results showing all covariates (i.e., multivariate results)
are reported in Appendix Table A3.5. Specifically, the multivariate regression results
in Column 3 show that aid reduces the likelihood of being energy poor by 3.3
percentage points among households within the 25 km radius from an aid project.
Column 4 shows aid generates a 0.009-unit decrease in energy poverty among
households living within a 25 km distance of an aid project.
The effects of the control variables are generally consistent with expectations. For
instance, being employed, male, highly educated and married have negative effects
on energy poverty, while staying in urban areas and larger household sizes lower the
probability of being energy poor. Economic performance also lowers energy poverty.
The effects of foreign aid are relatively large with economic relevance. For instance,
comparing the standardised coefficients, we find that the effect of aid is relatively
stronger in magnitude compared to marital status and employment status. From Table
A3.2, the average MEPI status and scores are 0.825 and 0.657, respectively. A
standardised coefficient of -0.042 in the MEPI status regression implies that proximity
to an aid project is associated with a 5.1% (0.042/0.825) decline in energy poverty
measured using the MEPI status. Similarly, a standardised coefficient of -0.024 implies
a 3.7% (0.024/0.657) decline in energy poverty.
3.7.2 Sensitivity checks
In this section, we test the sensitivity of our results by running a number of robustness
checks. First, we examine the efficacy of aid at different distance buffers. Evidence
suggests that aid is most effective when households are nearer to aid projects (see,
e.g., Kotsadam et al. 2018; Odokonyero et al. 2018). To examine if this is the case for
energy poverty, we examine the robustness to different distance buffers that reflect
77
the distance between households and aid projects. Specifically, we consider
alternative radii including 5, 10, 15, 20, 75 and 100 km. Table 3.2 presents the results.
Table 3.2 – Alternative distance buffers

(1) (2)
VARIABLES MEPI status MEPI score
Panel A: 5 km buffer
Time *Total Aid (5 km) -0.020** -0.011***
(0.008) (0.003)
Observations 17,227 17,227

R-squared 0.223 0.421
Panel B: 10 km buffer
Time *Total Aid (10 km) -0.013* -0.001
(0.008) (0.003)

R-squared 0.223 0.421
Panel C: 15 km buffer
Time *Total Aid (15 km) -0.026*** -0.009**
(0.008) (0.003)

R-squared 0.223 0.419
Panel D: 20 km buffer
Time *Total Aid (20 km) -0.041*** -0.013***
(0.009) (0.004)

R-squared 0.224 0.419
Panel E: 50 km buffer
Time *Total Aid (50 km) -0.054*** -0.026***
(0.019) (0.010)

R-squared 0.223 0.419
Panel F: 75 km buffer
Time *Total Aid (75 km) 0.023 0.040*
(0.044) (0.021)

R-squared 0.223 0.419
Panel G: 100 km buffer
Time *Total Aid (100 km) 0.023 0.040*
(0.044) (0.021)

R-squared 0.223 0.419
Notes: All regressions above include district fixed effects and control variables.
78
We find that, consistent with the main results, for buffers below the 75 km threshold,
aid has a negative and statistically significant effect on energy poverty (see Panels A
to E). However, for the 75 km and 100 km distance buffers, foreign aid fails to reduce
energy poverty. As shown, in Column 2 of Panels F and G, being 75–100 km away
from an aid project establishment increases energy poverty. This finding lends support
to the existing literature that associates proximity to an aid project site with increased
aid efficacy (see, e.g., Kotsadam et al. 2018; Odokonyero et al. 2018; Wayland 2019).
Second, we check the robustness of our results by examining the effect of different
types of aid on energy poverty. Our results are based on total aid or any form of aid
project. Here, we restrict this to specific aid projects to examine if the dynamics of aid
are altered depending on project type. Specifically, we separately examine the effect
of production aid, social infrastructure aid and economic infrastructure aid on energy
poverty. Production aid refers to total aid meant for production sectors, including
agriculture, construction, mining, trade, industry, tourism, fishing and forestry.
Economic infrastructure aid is total aid for sectors such as energy, banking and
financial services, communications, business and other services, and transport and
storage. Social infrastructure aid is total aid availed to sectors focused on education,
health, government, water supply and sanitation, and other social infrastructure and
services. The results, which are reported in Table 3.3, suggest that all aid types have
a negative and statistically significant effect on energy poverty. Thus, the finding that
aid reduces households’ probability of being energy-poor is reinforced.
79
Table 3.3 – Types of foreign aid
(1) (2)
Panel A: Social Infrastructure & Services Aid
Time *Social Aid (25 km) -0.030*** -0.010***
(0.009) (0.004)

R-squared 0.223 0.420
Panel B: Economic Infrastructure & Services Aid
Time *Economic Aid (25 km) -0.024*** -0.009***
(0.008) (0.003)

R-squared 0.223 0.420
Panel C: Production Aid
Time * Production Aid (25 km) -0.015* -0.008***
(0.008) (0.003)

R-squared 0.223 0.420
Third, we verify if our results are sensitive to the use of different cut-off points and
weights for the MEPI. In the measure of MEPI used in our main analysis, a household
is considered energy poor if the household energy deprivation score is at least 0.5. To
examine the robustness of our results, we adopt alternative cut-off points. Specifically,
we use 0.3 and 0.7 as alternative cut-off points to examine how robust our results are
to relatively conservative and liberal thresholds. Table 3.4 reports the results using
these alternative cut-off points. Column 1 presents results for MEPI status based on
0.3 as the cut-off, while Column 2 presents results using 0.7 as the cut-off. We find
that our results are robust to these alternative thresholds.
80
Table 3.4 – Alternative cut-offs for energy poverty index
Dependent variable: MEPI status

(1) (2) (3)
VARIABLES Panel A: Panel B: Panel C:
0.3 Cut-off 0.7 Cut-off Using equal weights
Time *Total Aid (25 km) -0.010** -0.010** -0.015***

(0.004) (0.004) (0.006)
Observations 17,208 17,208 17,208

R-squared 0.422 0.422 0.409
Robust standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1
In Column 3 of Table 3.4, we check the robustness of our results to alternative weights
for the indicators included in the MEPI. Consistent with the literature, our main
regression results are based on the measure of MEPI that applied a weight of 0.4 to
cooking fuel, 0.2 to electricity access and 0.13 to each of the other dimensions. In
Column 3, where we test the sensitivity of the results to an index that places equal
weights on each of the indicators, we find that our results remain robust.
Fourth, we examine the robustness of our results to the individual energy poverty
indicators used in deriving the multidimensional index. The results reported in Table
3.5 show that aid has a positive effect on improving each indicator.
Table 3.5 – Indicators of energy poverty
(1) (2) (3) (4) (5)

VARIABLES Cooking Electricity Electric Entertainment Communication
fuels access appliances means means
Time *Total Aid (25 km) 0.004** 0.032*** 0.016** 0.031*** 0.037***
(0.002) (0.008) (0.007) (0.008) (0.006)
Observations 17,227 17,227 17,227 17,227 17,227

R-squared 0.029 0.449 0.221 0.108 0.096
*** p<0.01, ** p<0.05, * p<0.1
Specifically, aid increases the likelihood of using clean cooking fuels by 0.4 percentage
points, accessing electricity by 3.2 percentage points, using electric appliances (e.g.,
fridge) by 1.6 percentage points, using entertainment means (e.g., radio or TV) by 3.1
81
percentage points, and using communication means (e.g., landline or mobile phone)
by 3.7 percentage points.
Fifth, we verify the sensitivity of our results to foreign aid amounts (i.e., foreign aid
intensity) allocated to the aid projects establishments. In the main results, we capture
the impact of aid using a binary variable equal to 1 if a household is staying within the
25 km distance from an aid project. In robustness checks, we examine the effect of
foreign aid amounts (commitments and actual disbursements) on energy poverty. The
findings presented in Table 3.6 show that aid has a negative and statistically significant
impact on energy poverty, thus reinforcing our main results.
Table 3.6 – Foreign aid intensity
(1) (2)
Panel A: Aid commitments
Time *Total Aid (25 km) -0.004*** -0.001***
(0.001) (0.000)

R-squared 0.224 0.419
Panel B: Aid disbursements
Time *Total Aid (25 km) -0.006*** -0.003***
(0.001) (0.000)

R-squared 0.226 0.423
*** p<0.01, ** p<0.05, * p<0.1
Sixth, in Table 3.7, we check the robustness of the results to the 1) exclusion of
districts fixed effects and 2) application of different lags on the foreign aid variable.
The main results include district fixed effects to take care of time-invariant and
unobservable factors. However, given that in some instances, foreign aid may be more
concentrated in some districts than in others, as an additional check, we exclude
district fixed effects and explore foreign aid impacts on energy poverty.
82
Table 3.7 – Exclusion of district fixed effects, and lagged and non-lagged
foreign aid
(1) (2)
Panel A: No district fixed effects
Time *Total Aid (25 km) -0.042*** -0.015***
(0.010) (0.004)
Controls Yes Yes

District fixed effects No No
R-squared 0.191 0.354
Panel B: Non-lagged foreign aid
Time *Total Aid (25 km) 0.002 -0.018***
(0.008) (0.003)
Controls Yes Yes

District fixed effects Yes Yes
R-squared 0.223 0.419
Panel C: 2-year lag
Time *Total Aid (25 km) -0.015** -0.016***
(0.007) (0.003)
Controls Yes Yes

District fixed effects Yes Yes
R-squared 0.223 0.419
*** p<0.01, ** p<0.05, * p<0.1
The results, which are shown in Panel A of Table 3.7, show that the impact of aid on
energy poverty remains negative and statistically significant. Additionally, in the main
results, we capture the impact of aid in the previous period (𝑡 − 1). In further checks,
we consider the impact of foreign aid in the current period (𝑡) and the second (𝑡 − 2)
period. The results of this exercise, which are reported in Panels B and C of Table 3.7,
confirm the finding that aid negatively affects energy poverty.
Seventh, we verify the sensitivity of our results to different waves of the DHS data. The
results are presented in Table A3.6. Across all columns and panels, we find that the
effect of aid on energy poverty is negative and statistically significant. Additionally, we
observe that the effect of aid (i.e., the coefficient magnitude of aid) fluctuates over
time. Between 1997 and 2005, the effect of aid declines, but it peaks when the 2010–
83
11 data are used, before declining again when using the 2012–13 data. However,
despite these fluctuations, the effect of aid on energy poverty remains robust.
Last, we present evidence on the DID parallel trend assumption. We verify if, in the
absence of treatment (in this case, foreign aid projects), the difference between the
control and treatment groups is constant over time. To do this, we run a balancing test
on all our main energy poverty indicators (e.g., MEPI status and MEPI score) and
individual energy poverty indicators (e.g., cooking fuels, electricity access, electric
appliances, entertainment means, and communication means) at baseline conditions.
The assumption is fulfilled if the differences in all the energy poverty indicators
between the control and treatment groups are not statistically significant. The
balancing test results, presented in Appendix Table A3.7, show that for all the energy
poverty indicators except cooking fuels, the differences in outcomes between the two
groups at the 5% level are not statistically significant. Thus, the parallel trend
assumption is satisfied.
3.7.3 Mechanisms
Section 3.3 discusses several channels through which aid might influence energy
poverty. The data used in this study allow us to examine the potential role of income
poverty, economic growth, and education and health status as mechanisms through
which aid transmits to energy poverty. We use a two-staged approach consistent with
the literature to examine the validity of these variables as mechanisms (see, e.g.,
Awaworyi Churchill et al. 2019). As a first step, for a variable (in the context of this
study, income poverty, economic growth, education or health status) to qualify as a
transmission channel, it needs to be correlated with aid. Second, the coefficient on aid
should reduce in magnitude or become statistically insignificant with the inclusion of
the potential mechanism variable as an additional covariate in the regression linking
foreign aid to energy poverty. We examine the potential channel variables in
alternating models.
Table 3.8 reports the effects of aid on our potential channel variables. Columns 1, 2,
3 and 4 report results for the effects of aid on income poverty, education, health and
economic growth, respectively.
84
Table 3.8 – Effects of aid on mechanisms
(1) (2) (3) (4)

VARIABLES Wealth Education Health Night-time light
Total Aid (25 km) 0.129*** 0.010* -0.021*** 7.508***

(0.006) (0.005) (0.006) (0.205)
Observations 29,707 29,707 29,707 22,061

R-squared 0.328 0.088 0.021 0.610
Notes: All regressions include district fixed effects.
*** p<0.01, ** p<0.05, * p<0.1
From Column 1, we find a positive effect of aid on wealth. Thus, aid is associated with
lower income poverty (i.e., higher wealth). From Columns 2 and 4, we find that aid is
associated with higher educational attainment and enhanced economic activity,
respectively. In Column 3, the results show that aid is associated with lower health
outcomes.
Table 3.9 reports results for alternating models that include the potential channel
variables as additional covariates in regressions examining the impact of aid on MEPI.
Panel A reports results for effects on MEPI status, while Panel B reports results for
effects on MEPI score.
85
Table 3.9 – Effects of mechanisms
Dependent variable: MEPI
VARIABLES (1) (2) (3) (4) (5)
Panel A: Effects on MPI status
Total Aid (25 km) -0.033*** 0.011** -0.031*** -0.032*** -0.010*
(0.006) (0.005) (0.006) (0.006) (0.006)
Wealth -0.354***
(0.006)
Education -0.179***
(0.007)
Health 0.031***
(0.006)
Night-time light -0.003***
(0.000)
Observations 22,078 22,078 22,078 22,078 22,078

R-squared 0.128 0.255 0.160 0.129 0.135
Panel B: Effects on MPI score
Total Aid (25 km) -0.030*** 0.003 -0.029*** -0.030*** -0.017***
(0.002) (0.002) (0.002) (0.002) (0.003)
Wealth -0.264***
(0.002)
Education -0.083***
(0.003)
Health 0.011***
(0.003)
Night-time light -0.002***
(0.000)
Observations 22,078 22,078 22,078 22,078 22,078

R-squared 0.245 0.574 0.277 0.246 0.254
Notes: All regressions include district fixed effects.
*** p<0.01, ** p<0.05, * p<0.1
We first estimate the regression with only aid as the explanatory variable to ensure
that it uses the same sample when potential channel variables are included as
additional covariates. The results, reported in Column 1 of Table 3.9, are consistent
with aid having a negative effect on energy poverty. We find that lower-income poverty
(i.e., higher wealth) is associated with lower energy poverty. Regarding human capital,
education is associated with lower energy poverty, while health positively affects
energy poverty. Higher economic growth is also associated with lower energy poverty.
Further, with the inclusion of income poverty in the model, we find that the coefficient
of aid reduces, and in Panel B, it becomes statistically insignificant. The inclusion of
86
education and economic growth (i.e., night-time light) as additional covariates reduces
the size of the coefficient on aid. These results confirm that income poverty, education
and economic growth are potential channels through which aid transmits to energy
poverty.
Based on the results presented in Table 3.9, we find that wealth – a proxy for income
poverty, is the main transmission channel (based on the coefficient size) between
foreign aid and energy poverty. A potential explanation for this could be the fact that
many aid projects are mainly focused on empowering recipients by enhancing their
incomes – as a way of helping these recipients achieve other important outcomes that
depend on income (e.g., health, education, among others).
3.8 Conclusion
Energy poverty is a major challenge in developed and developing countries. In

developing countries, foreign aid has been touted as a potential intervention that
promotes universal access to reliable, sustainable, affordable, and modern energy
services, thus reducing energy poverty. However, no study has explored the impact of
foreign aid on energy poverty. Using DHS and sub-national aid data, we contribute to
the literature on aid and energy poverty by examining the impact of aid on energy
poverty in Senegal. We apply a DID estimation strategy and run a linear probability
equation to examine the effects of proximity to aid projects on energy poverty.
Our main finding is that sub-national aid has a negative and statistically significant
effect on energy poverty. This general conclusion is robust to all sensitivity checks
conducted. These sensitivity checks include using different distance buffers, aid types,
cut-off points and weights for the MEPI. The results obtained from using other types
of foreign aid suggest that households close to production, economic infrastructure
and social infrastructure aid projects significantly reduce their likelihood of being
energy poor. Further, these results show that economic infrastructure aid (i.e., aid for
energy, banking and financial services, communications, business and other services,
and transport and storage) has a larger impact than production aid (i.e., aid for
agriculture, construction, mining, trade, industry, tourism, fishing and forestry) and a
lower impact than social infrastructure aid (i.e., aid for education, health, government,
water supply and sanitation, and other social infrastructure and services).
87
The lower effect of economic infrastructure aid on energy poverty could be because
this category of aid is targeted at areas that have less impact on energy poverty. For
example, aid for communications, transport and storage, has little to do with
eliminating energy poverty among households. While we would expect the impact of
aid towards the energy sector to have a relatively stronger effect, our findings suggest
that the relatively weak effect of economic infrastructure aid is driven by the other
sectors such as communication, transport, banking and financial services, among
others, all of which are part of economic infrastructure. A limitation of our study,
therefore, is the inability to examine the direct effects of aid targeted to only the energy
sector given data availability issues. Social infrastructure aid is more effective in
eradicating energy poverty, given that it is allocated to sectors that are more likely to
influence the capacity of households to address energy poverty via the income
channel. For instance, aid directed to improve education and health provides people
with better economic opportunities (e.g., better jobs, improved savings and knowledge
on clean energy) to afford improved energy services, such as electricity and clean
cooking fuels.
We also examine a number of channels through which aid might influence energy
poverty. The results show that income poverty, education and economic growth are
channels through which aid transmits to energy poverty. Thus, policies aimed at
promoting aid, economic growth, education and boosting household wealth (i.e.,
lowering income poverty) are important to ensuring energy poverty is reduced. Further,
the finding that proximity plays a role in the effectiveness of aid suggests that aid is
likely to benefit households if appropriately managed – with aid projects spread out
across communities.
Our findings have important implications for policies aimed at reducing energy poverty.
While there is an ongoing debate on the usefulness of aid, our findings demonstrate
that aid projects that are appropriately implemented can have significant effects in
reducing energy poverty. Given that our study focuses on the location of aid projects
and their proximity to households, our findings lend support to the need for carefully
planned projects that take into consideration the location of beneficiaries. This will
ensure the optimal benefits of aid, especially on energy poverty, can be achieved.
88
Our study has two main limitations. First, compared to other foreign aid sources, the
AidData database understates the volumes of aid availed to developing countries
(Odokonyero et al. 2018). However, this does not pose much of a problem, given that
the results of this study will only remain conservative in this case. Second, due to data
availability issues, this study could not examine some of the potential channels through
which aid influences energy poverty. These include energy efficiency and income
inequality, which could be considered in future studies subject to data availability.
89
References
Abadie, A 2005, 'Semiparametric difference-in-differences estimators', The Review of
Economic Studies, vol. 72, no. 1, pp. 1-19.
Acharya, RH & Sadath, AC 2019, 'Energy poverty and economic development: Household-
level evidence from India', Energy and Buildings, vol. 183, pp. 785-791.
Acheampong, AO, Erdiaw-Kwasie, MO & Abunyewah, M 2021, 'Does energy accessibility

improve human development? Evidence from energy-poor regions', Energy Economics, vol.
96, p. 105165.
Adusah-Poku, F & Takeuchi, K 2019, 'Energy poverty in Ghana: Any progress so far?',
Renewable and Sustainable Energy Reviews, vol. 112, pp. 853-864.
AFDB 2013, The high cost of electricity eeneration in Africa, African Development Bank,
Abidjan, <https://blogs.afdb.org/blogs/afdb-championing-inclusive-growth-across-
africa/post/the-high-cost-of-electricity-generation-in-africa-11496>.
Aghion, P & Howitt, P 1994, 'Growth and unemployment', The Review of Economic Studies,
vol. 61, no. 3, pp. 477-494.
AidData 2021, Geospatial Impact Evaluations, AidData: William & Mary Williamsburg, Virginia,
<https://www.aiddata.org/gie>.
Alvi, E & Senbeta, A 2012, 'Does foreign aid reduce poverty?', Journal of International
Amare, M, Jensen, ND, Shiferaw, B & Cissé, JD 2018, 'Rainfall shocks and agricultural
productivity: Implication for rural household consumption', Agricultural Systems, vol. 166, pp.
79-89.
Ambrose, AR 2015, 'Improving energy efficiency in private rented housing: Why don't
landlords act?', Indoor and Built Environment, vol. 24, no. 7, pp. 913-924.
Angrist, JD & Pischke, J-S 2008, Mostly harmless econometrics, Princeton University Press,
Princeton.
Arndt, C, Jones, S & Tarp, F 2015, 'Assessing foreign aid’s long-run contribution to growth
and development', World Development, vol. 69, pp. 6-18.
Arshad, M, Farooq, O, Sultana, N & Farooq, M 2016, 'The role of foreign aid in the
development of Pakistan’s education sector: a time-lag analysis', European Online Journal of
Natural and Social Sciences, vol. 5, no. 3, pp. 588-594.
Arvin, BM & Barillas, F 2002, 'Foreign aid, poverty reduction, and democracy', Applied
90
Ashagidigbi, WM, Babatunde, BA, Ogunniyi, AI, Olagunju, KO & Omotayo, AO 2020,
'Estimation and determinants of multidimensional energy poverty among households in
Nigeria', Sustainability, vol. 12, no. 18, p. 7332.
Asongu, SA & Nwachukwu, JC 2016, 'Foreign aid and governance in Africa', International
Review of Applied Economics, vol. 30, no. 1, pp. 69-88.
Austin, KF & Mejia, MT 2017, 'Household air pollution as a silent killer: women’s status and
solid fuel use in developing nations', Population and Environment, vol. 39, no. 1, pp. 1-25.
pp. 506-518.
Awaworyi Churchill, S & Smyth, R 2020, 'Ethnic diversity, energy poverty and the mediating
role of trust: Evidence from household panel data for Australia', Energy Economics, vol. 86, p.
104663.
Awaworyi Churchill, S & Smyth, R 2021, 'Energy poverty and health: panel data evidence from
Australia', Energy Economics, vol. 97, p. 105219.
Awaworyi Churchill, S, Smyth, R & Farrell, L 2020, 'Fuel poverty and subjective wellbeing',
Energy Economics, vol. 86, p. 104650.
Azpitarte, F, Johnson, V & Sullivan, D 2015, Fuel poverty, household income and energy
spending: an empirical analysis for Australia using HILDA data, Brotherhood of St. Laurence,
Fitzroy, Victoria.
Ba, AS 2018, The energy policy of the Republic of Senegal, Université Paris Dauphine, Paris.
Bahmani‐Oskooee, M & Oyolola, M 2009, 'Poverty reduction and aid: cross‐country evidence',
International Journal of Sociology and Social Policy, vol.
Baldwin, E, Brass, JN, Carley, S & MacLean, LM 2015, 'Electrification and rural development:
issues of scale in distributed generation', Wiley Interdisciplinary Reviews: Energy and
Environment, vol. 4, no. 2, pp. 196-211.

Balint, PJ 2006, 'Bringing solar home systems to rural El Salvador: lessons for small NGOs',
Energy Policy, vol. 34, no. 6, pp. 721-729.
Barnes, DF, Khandker, SR & Samad, HA 2011, 'Energy poverty in rural Bangladesh', Energy
Policy, vol. 39, no. 2, pp. 894-904.
91
Bartolucci, F, Choudhry, MT, Marelli, E & Signorelli, M 2018, 'GDP dynamics and
unemployment changes in developed and developing countries', Applied Economics, vol. 50,
no. 31, pp. 3338-3356.
Bendavid, E & Bhattacharya, J 2014, 'The relationship of health aid to population health
improvements', JAMA Internal Medicine, vol. 174, no. 6, pp. 881-887.
BenYishay, A, Runfola, D, Trichler, R, Dolan, C, Goodman, S, Parks, B, Tanner, J, Heuser,

S, Batra, G & Anand, A 2017, 'A primer on geospatial impact evaluation methods, tools, and
applications', in AidData Working Paper# 44, AidData at William & Mary Williamsburg, VA.
Bertrand, M, Duflo, E & Mullainathan, S 2004, 'How much should we trust differences-in-
differences estimates?', The Quarterly Journal of Economics, vol. 119, no. 1, pp. 249-275.
Biermann, P 2016, How fuel poverty affects subjective well-being: Panel evidence from
Germany, Oldenburg Discussion Papers in Economics.
Bigsten, A & Tengstam, S 2015, 'International coordination and the effectiveness of aid', World
Boardman, B 1991, 'Fuel Poverty: From Cold Homes to Affordable Warmth Bellhaven London',
vol.
Boardman, B 2010, Fixing fuel poverty: challenges and solutions. London; Sterling, VA:
Earthscan.
Boly, M 2018, 'CO 2 mitigation in developing countries: the role of foreign aid', CERDI - Center
for Studies and Research on International Development, Clermont-Ferrand.
Bourguignon, F, Levin, V & Rosenblatt, D 2009, 'International redistribution of income', World

Bourguignon, F & Platteau, J-P 2017, 'Does aid availability affect effectiveness in reducing
poverty? A review article', World Development, vol. 90, pp. 6-16.
Bouzarovski, S & Petrova, S 2015, 'A global perspective on domestic energy deprivation:
Overcoming the energy poverty–fuel poverty binary', Energy Research & Social Science, vol.
10, pp. 31-40.
Bouzarovski, S & Tirado Herrero, S 2017, 'Geographies of injustice: the socio-spatial

determinants of energy poverty in Poland, the Czech Republic and Hungary', Post-Communist
Economies, vol. 29, no. 1, pp. 27-50.
Brass, JN 2012, 'Why do NGOs go where they go? Evidence from Kenya', World
Bräutigam, DA & Knack, S 2004, 'Foreign aid, institutions, and governance in sub-Saharan
Africa', Economic Development and Cultural Change, vol. 52, no. 2, pp. 255-285.
Brazys, S 2018, 'Aid dependence as aid persistence? Non-declining aid and growth', Journal
of International Relations and Development, vol. 21, no. 3, pp. 717-738.
92
Briggs, RC 2012, 'Electrifying the base? Aid and incumbent advantage in Ghana', The Journal
of Modern African Studies, vol. 50, no. 4, pp. 603-624.
Briggs, RC 2017, 'Does foreign aid target the poorest?', International Organization, vol. 71,
no. 1, pp. 187-206.
Briggs, RC 2021, 'Why Does Aid Not Target the Poorest?', International Studies Quarterly,
vol. 65, no. 3, pp. 739-752.
Burlinson, A, Giulietti, M & Battisti, G 2018, 'The elephant in the energy room: Establishing the
nexus between housing poverty and fuel poverty', Energy Economics, vol. 72, pp. 135-144.
Bwalya Umar, B, Chisola, MN, Mushili, BM, Kunda-Wamuwi, CF, Kafwamba, D, Membele, G
& Imasiku, EN 2021, 'Load-shedding in Kitwe, Zambia: Effects and implications on household
and local economies', Development Southern Africa, vol., pp. 1-18.
Chand, K, Tiwari, R & Phuyal, M 2017, 'Economic Growth and Unemployment Rate: An
Empirical Study of Indian Economy', Journal of Indian Economy, vol. 4, no. 2, pp. 130-137.
Chong, A, Gradstein, M & Calderon, C 2009, 'Can foreign aid reduce income inequality and
poverty?', Public Choice, vol. 140, no. 1-2, pp. 59-84.
Christensen, Z, Homer, D & Nielson, DL 2011, 'Dodging adverse selection: How donor type
and governance condition aid’s effects on school enrollment', World Development, vol. 39, no.
11, pp. 2044-2053.
Collier, P & Goderis, B 2009, 'Does aid mitigate external shocks?', Review of Development
Collier, P, Hoeffler, A & Collier, P 2002, Aid, policy, and growth in post-conflict societies, The
World Bank.
d’Aiglepierre, R & Wagner, L 2013, 'Aid and universal primary education', Economics of
Education Review, vol. 37, pp. 95-112.
Dalgaard, CJ, Hansen, H & Tarp, F 2004, 'On the empirics of foreign aid and growth', The
Economic Journal, vol. 114, no. 496, pp. F191-F216.
AidData, USA.
301-313.
Dolan, CB, BenYishay, A, Grépin, KA, Tanner, JC, Kimmel, AD, Wheeler, DC & McCord, GC
2019, 'The impact of an insecticide treated bednet campaign on all-cause child mortality: A
geospatial impact evaluation from the Democratic Republic of Congo', PloS One, vol. 14, no.
2, p. e0212890.
Dollar, D & Burnside, C 2000, 'Aid, policies, and growth', American Economic Review, vol. 90,
no. 4, pp. 847-868.
93
Doucouliagos, H & Paldam, M 2008, 'Aid effectiveness on growth: A meta study', European
Journal of Political Economy, vol. 24, no. 1, pp. 1-24.
Dreher, A, Nunnenkamp, P & Thiele, R 2008, 'Does aid for education educate children?
Evidence from panel data', The World Bank Economic Review, vol. 22, no. 2, pp. 291-314.
Duque, V, Rosales-Rueda, M & Sanchez, F 'How do early-life shocks interact with subsequent
human-capital investments? Evidence from administrative data',
Eberhard, A, Gratwick, K & Kariuki, L 2018, 'Kenya's lessons from two decades of experience
with independent power producers', Utilities Policy, vol. 52, pp. 37-49.
FAO 2020, Characteristics, patterns and drivers of rural migration in Senegal, Food and
Agriculture Organization of the United Nations (FAO), Rome.
Feeny, S 2005, 'The impact of foreign aid on economic growth in Papua New Guinea', Journal
of Development Studies, vol. 41, no. 6, pp. 1092-1117.
Feeny, S & McGillivray, M 2010, 'Aid and growth in small island developing states', The Journal
of Development Studies, vol. 46, no. 5, pp. 897-917.
Filippidis, M, Tzouvanas, P & Chatziantoniou, I 2021, 'Energy poverty through the lens of the
energy-environmental Kuznets curve hypothesis', Energy Economics, vol., p. 105328.
Filmer, D & Pritchett, LH 2001, 'Estimating wealth effects without expenditure data—or tears:
an application to educational enrollments in states of India', Demography, vol. 38, no. 1, pp.
115-132.
Gachanja, PM & Kinyanjui, GK 2016, 'Household Poverty Determinants in Kenya: A

Demographic and Health Survey Wealth Index Approach', Journal of Pan African Studies, vol.
9, no. 3.
GIZ 2021, Successful in Senegal, Deutsche Gesellschaft für Internationale Zusammenarbeit

(GIZ), Bonn, <https://www.giz.de/en/worldwide/68202.html>.
Gong, L & Zou, Hf 2001, 'Foreign aid reduces labor supply and capital accumulation', Review
of Development Economics, vol. 5, no. 1, pp. 105-118.
Guillaumont, P & Wagner, L 2014, 'Aid effectiveness for poverty reduction: Lessons from
cross‐country analyses, with a special focus on vulnerable countries', Revue D'économie du
Développement, vol. 22, no. HS01, pp. 217-261.
Gyimah‐Brempong, K, Racine, JS & Gyapong, A 2012, 'Aid and economic growth: Sensitivity
analysis', Journal of International Development, vol. 24, no. 1, pp. 17-33.
94
Häfele, S 2017, Does Foreign Aid to the Energy Sector improve Electricity Supply in
Developing Countries?: Evidence from Panel Data, GRIN Verlag.
Heo, S, Nori-Sarma, A, Lee, K, Benmarhnia, T, Dominici, F & Bell, ML 2019, 'The use of a
quasi-experimental study on the mortality effect of a heat wave warning system in Korea',
International Journal of Environmental Research and Public Health, vol. 16, no. 12, p. 2245.
Herrera-Almanza, C & Sahn, DE 2020, 'Childhood determinants of internal youth migration in

Senegal', Demographic Research, vol. 43, pp. 1335-1366.
Hills, J 2012, Getting the measure of fuel poverty: Final Report of the Fuel Poverty Review,
Centre for the Analysis of Social Exclusion, LSE, UK, London.
Hirano, Y & Otsubo, S 2014, Aid is good for the poor, The World Bank.
vol. 129, no. 2, pp. 995-1033.
Hossain, MF, Hossain, S & Uddin, MJ 2017, 'Renewable energy: Prospects and trends in
Bangladesh', Renewable and Sustainable Energy Reviews, vol. 70, pp. 44-49.
Hung-Pin, L 2014, 'Renewable energy consumption and economic growth in nine OECD
countries: bounds test approach and causality analysis', The Scientific World Journal, vol.
2014, p. 919167.
Iamsiraroj, S 2016, 'The foreign direct investment–economic growth nexus', International

Review of Economics & Finance, vol. 42, pp. 116-133.
ICF 2018, Demographic and Health Surveys Standard Recode Manual for DHS7, The
Demographic and Health Surveys Program, ICF, Rockville, Maryland, U.S.A.
ICF 2020, Senegal: Demographic and Health Surveys, ICF, Dakar, Senegal,
<https://www.dhsprogram.com/pubs/pdf/FR367/FR367.pdf>.
IEA 2019a, Access to electricity, International Energy Agency, Paris,

<https://www.iea.org/reports/sdg7-data-and-projections/access-to-electricity>.
IEA 2019b, Africa Energy Outlook 2019, IEA, France.
IEA 2020a, Energy access: Achieving modern energy for all by 2030 is possible, International
Energy Agency (IEA), Paris, <https://www.iea.org/topics/energy-access>.
IEA 2020b, Energy Efficiency 2020, International Energy Agency (IEA), Paris,
<https://www.iea.org/reports/energy-efficiency-2020>.
Ijaiya, GT & Ijaiya, MA 2004, 'Foreign aid and poverty reduction in sub-Saharan Africa: A
cross-country investigation', South African Journal of Economic and Management Sciences,
vol. 7, no. 3, pp. 542-552.
Isaksson, A-S & Kotsadam, A 2018, 'Chinese aid and local corruption', Journal of Public
Economics, vol. 159, pp. 146-159.
Jayachandran, S 2021, 'Social norms as a barrier to women’s employment in developing

countries', IMF Economic Review, vol., pp. 1-20.
95
Kablan, S 2018, 'An Analysis of the Links between Foreign Aid and Co 2 Emissions in Cities',
in Aid Effectiveness for Environmental Sustainability, Springer, pp. 185-214.
Kaya, O, Kaya, I & Gunter, L 2013, 'Foreign aid and the quest for poverty reduction: Is aid to
agriculture effective?', Journal of Agricultural Economics, vol. 64, no. 3, pp. 583-596.
Kim, JE 2018, 'Technological capacity building through energy aid: Empirical evidence from
renewable energy sector', Energy policy, vol. 122, pp. 449-458.
Koomson, I & Danquah, M 2021, 'Financial inclusion and energy poverty: Empirical evidence
from Ghana', Energy Economics, vol. 94, p. 105085.
Kretschmer, B, Hübler, M & Nunnenkamp, P 2013, 'Does foreign aid reduce energy and
carbon intensities of developing economies?', Journal of International Development, vol. 25,
no. 1, pp. 67-91.
Legendre, B & Ricci, O 2015, 'Measuring fuel poverty in France: Which households are the
most fuel vulnerable?', Energy Economics, vol. 49, pp. 620-628.
Lewis, WA 1954, 'Economic development with unlimited supplies of labour', Southern

Methodist University, Texas.
Liew, C-Y, Mohamed, MR & Mzee, SS 2012, 'The impact of foreign aid on economic growth
of East African countries', Journal of Economics and Sustainable Development, vol. 3, no. 12,
pp. 129-188.
Llorca, M, Rodriguez-Alvarez, A & Jamasb, T 2020, 'Objective vs. subjective fuel poverty and
self-assessed health', Energy Economics, vol. 87, p. 104736.
MacLean, LM & Brass, JN 2015, 'Foreign aid, NGOs and the private sector: New forms of
hybridity in renewable energy provision in Kenya and Uganda', Africa Today, vol. 62, no. 1,
pp. 57-82.
Mahadea, D & Kaseeram, I 2018, 'Impact of unemployment and income on entrepreneurship

in post-apartheid South Africa: 1994–2015', The Southern African Journal of Entrepreneurship
and Small Business Management, vol. 10, no. 1, pp. 1-9.
Mahalik, MK, Villanthenkodath, MA, Mallick, H & Gupta, M 2021, 'Assessing the effectiveness
of total foreign aid and foreign energy aid inflows on environmental quality in India', Energy
Policy, vol. 149, p. 112015.
Maitah, M, Toth, D & Kuzmenko, E 2015, 'Exploring the relationship between economic growth
and employment in the Czech Republic and Belgium', Review of European Studies, vol. 7, no.
11, pp. 115-124.
Mak Arvin, B & Lew, B 2009, 'Foreign aid and ecological outcomes in poorer countries: an
empirical analysis', Applied Economics Letters, vol. 16, no. 3, pp. 295-299.
96
Mallik, G 2008, 'Foreign Aid and Economic Growth: A Cointegration Analysis of the Six Poorest
African Countries', Economic Analysis & Policy, vol. 38, no. 2.
vol. 2, no. 1.
Maruta, AA & Banerjee, R 2020, 'Does energy aid improve energy efficiency in developing
countries?', Empirical Economics, vol., pp. 1-34.
Mattioli, G, Lucas, K & Marsden, G 2017, 'Transport poverty and fuel poverty in the UK: From
analogy to comparison', Transport Policy, vol. 59, pp. 93-105.
Mendoza Jr, CB, Cayonte, DDD, Leabres, MS & Manaligod, LRA 2019, 'Understanding
multidimensional energy poverty in the Philippines', Energy Policy, vol. 133, p. 110886.
Michaelowa, K & Weber, A 2006, 'Chapter 18 Aid Effectiveness in the Education Sector: A
Dynamic Panel Analysis', Theory and Practice of Foreign Aid, vol., pp. 357-385.
Minasyan, A, Nunnenkamp, P & Richert, K 2017, 'Does aid effectiveness depend on the quality
of donors?', World Development, vol. 100, pp. 16-30.
Minoiu, C & Reddy, SG 2010, 'Development aid and economic growth: A positive long-run
relation', The Quarterly Review of Economics and Finance, vol. 50, no. 1, pp. 27-39.
Mohr, TM 2018, 'Fuel poverty in the US: evidence using the 2009 Residential Energy
Consumption Survey', Energy Economics, vol. 74, pp. 360-369.
Mosley, P & Suleiman, A 2007, 'Aid, agriculture and poverty in developing countries', Review
of Development Economics, vol. 11, no. 1, pp. 139-158.
Macmillan, UK.
Munyanyi, ME, Mintah, K & Baako, KT 2021, 'Energy-related deprivation and housing tenure
transitions', Energy Economics, vol. 98, p. 105235.
Nance 2013, Relative energy poverty in Australia, St Kitts Associates, Adelaide.
Ningi, T, Taruvinga, A & Zhou, L 2020, 'Determinants of energy security for rural households:
The case of Melani and Hamburg communities, Eastern Cape, South Africa', African Security
Review, vol. 29, no. 4, pp. 299-315.
Njiru, CW & Letema, SC 2018, 'Energy Poverty and Its Implication on Standard of Living in
Kirinyaga, Kenya', Journal of Energy, vol. 2018, pp. 1-12.
NOAA 2020, Version 4 DMSP-OLS Nighttime Lights Time Series, National Oceanic and
Atmospheric Administration’s Earth Observation Group,
<https://www.ngdc.noaa.gov/eog/dmsp/downloadV4composites.html>.
97
Nussbaumer, P, Bazilian, M & Modi, V 2012a, 'Measuring energy poverty: Focusing on what
matters', Renewable and Sustainable Energy Reviews, vol. 16, no. 1, pp. 231-243.
Nussbaumer, P, Bazilian, M & Modi, V 2012b, 'Measuring energy poverty: Focusing on what
matters', Renewable & Sustainable Energy Reviews, vol. 16, no. 1, pp. 231-243.
Nussbaumer, P, Nerini, FF, Onyeji, I & Howells, M 2013, 'Global insights based on the
multidimensional energy poverty index (MEPI)', Sustainability, vol. 5, no. 5, pp. 2060-2076.
Okedu, KE, Kenu, S, Idowu, K & Uhunmwangho, R 2018, 'Impact of the independent power
producers and national integrated power projects on the deregulation of the Nigerian electricity
sector', Journal of Sustainable Development in Africa, vol. 20, no. 1.
Ouedraogo, NS 2017, 'Modeling sustainable long-term electricity supply-demand in Africa',

Applied Energy, vol. 190, pp. 1047-1067.
Pachauri, S & Spreng, D 2011, 'Measuring and monitoring energy poverty', Energy Policy, vol.
39, no. 12, pp. 7497-7504.
Pickbourn, L & Ndikumana, L 2019, 'Does Health Aid Reduce Infant and Child Mortality from
Diarrhoea in Sub-Saharan Africa?', The Journal of Development Studies, vol. 55, no. 10, pp.
2212-2231.
Poblete-Cazenave, M & Pachauri, S 2021, 'A model of energy poverty and access: Estimating
household electricity demand and appliance ownership', Energy Economics, vol. 98, p.
105266.
Prakash, K & Munyanyi, ME 2021, 'Energy poverty and obesity', Energy Economics, vol. 101,
p. 105428.
Pye, S, Dobbins, A, Baffert, C, Brajković, J, Deane, P & De Miglio, R 2015, 'Addressing energy
poverty and vulnerable consumers in the energy sector across the EU', L'Europe en
Formation, vol. no. 4, pp. 64-89.
Qurat-ul-Ann, A-R & Mirza, FM 2021, 'Determinants of multidimensional energy poverty in

Pakistan: a household level analysis', Environment, Development and Sustainability, vol., pp.
1-45.
Rademaekers, K, Yearwood, J, Ferreira, A, Pye, S, Hamilton, I, Agnolucci, P, Grover, D,

Karásek, J & Anisimova, N 2016, Selecting indicators to measure energy poverty, Trinomics:
Rotterdam, Netherlands.
Rasschaert, F, Pirard, M, Philips, MP, Atun, R, Wouters, E, Assefa, Y, Criel, B, Schouten, EJ

& Van Damme, W 2011, 'Positive spill‐over effects of ART scale up on wider health systems
98
development: evidence from Ethiopia and Malawi', Journal of the International AIDS Society,
vol. 14, pp. S3-S3.
Robinson, C, Bouzarovski, S & Lindley, S 2018, '‘Getting the measure of fuel poverty’: The
geography of fuel poverty indicators in England', Energy Research & Social Science, vol. 36,
pp. 79-93.
Rosenstein-Rodan, PN 1943, 'Problems of industrialisation of eastern and south-eastern

Europe', The economic journal, vol. 53, no. 210/211, pp. 202-211.
Ruales, J 2017, 'Determinants of Carbon Emission among Households in Eastern Visayas,

Philippines', Review of Socio-Economic Research and Development Studies, vol. 1, no. 1.
Sadath, AC & Acharya, RH 2017, 'Assessing the extent and intensity of energy poverty using
Multidimensional Energy Poverty Index: Empirical evidence from households in India', Energy
Policy, vol. 102, pp. 540-550.
Sharma, K, Bhattarai, B & Ahmed, S 2019, 'Aid, growth, remittances and carbon emissions in
Nepal', The Energy Journal, vol. 40, no. 1.
Shleifer, A 2009, 'Peter Bauer and the failure of foreign aid', Cato Journal, vol. 29, p. 379.
Skoufias, E, Katayama, RS & Essama-Nssah, B 2012, 'Too little too late: welfare impacts of
rainfall shocks in rural Indonesia', Bulletin of Indonesian Economic Studies, vol. 48, no. 3, pp.
351-368.
Streatfeild, JE 2018, 'Low electricity supply in sub-saharan Africa: causes, implications, and
remedies', Journal of International Commerce and Economics, vol., p. 1.
Svensson, J 2000, 'Foreign aid and rent-seeking', Journal of International Economics, vol. 51,
no. 2, pp. 437-461.
Tan‐Mullins, M, Mohan, G & Power, M 2010, 'Redefining ‘aid’in the China–Africa context',
Development and Change, vol. 41, no. 5, pp. 857-881.
Tavares, J 2003, 'Does foreign aid corrupt?', Economics Letters, vol. 79, no. 1, pp. 99-106.
Thomson, H, Bouzarovski, S & Snell, C 2017, 'Rethinking the measurement of energy poverty
in Europe: A critical analysis of indicators and data', Indoor and Built Environment, vol. 26, no.
7, pp. 879-901.
Thomson, H, Snell, C & Bouzarovski, S 2017, 'Health, well-being and energy poverty in
Europe: A comparative study of 32 European countries', International Journal of
Environmental Research and Public Health, vol. 14, no. 6, p. 584.
Thomson, H, Snell, CJ & Liddell, C 2016, 'Fuel poverty in the European Union: a concept in
need of definition?', People, Place & Policy Online, vol., pp. 5-24.
99
Timilsina, G & Steinbuks, J 2021, 'Economic costs of electricity load shedding in Nepal',
Renewable and Sustainable Energy Reviews, vol. 146, p. 111112.
Tirpak, D & Adams, H 2008, 'Bilateral and multilateral financial assistance for the energy sector
of developing countries', Climate Policy, vol. 8, no. 2, pp. 135-151.
Tull, DM 2006, 'China's engagement in Africa: scope, significance and consequences', The
Journal of Modern African Studies, vol. 44, no. 3, pp. 459-479.
United Nations 2020, The Energy Progress Report 2020, United Nations (UN), New York.
USAID 2021, Senegal Energy Sector Overview, United States Agency for International
Development, Washington DC, <https://www.usaid.gov/powerafrica/senegal>.
Vengroff, R & Creevey, L 2018, 'Senegal: the evolution of a quasi democracy', in Political
Reform in Francophone Africa, Routledge, pp. 204-222.
Wang, Q, Guo, J & Dong, Z 2021, 'The positive impact of official development assistance
(ODA) on renewable energy development: Evidence from 34 Sub-Saharan Africa Countries',
Sustainable Production and Consumption, vol. 28, pp.532-542.
156.
Williams, NJ, Jaramillo, P, Taneja, J & Ustun, TS 2015, 'Enabling private sector investment in
microgrid-based rural electrification in developing countries: A review', Renewable and
Sustainable Energy Reviews, vol. 52, pp. 1268-1281.
Wolf, S 2007, 'Does aid improve public service delivery?', Review of World Economics, vol.
143, no. 4, pp. 650-672.
World Bank - ESMAP 2021, Preparing feasibility studies for the financing of geothermal
projects, World Bank: Energy Sector Management Assistance Program (ESMAP),
Washington DC.
World Bank 2020a, DataBank: World Development Indicators,

World Bank 2020b, Off-Grid Solar Market Trends Report 2020, World Bank, Washington DC,
<https://www.worldbank.org/en/topic/energy/publication/off-grid-solar-market-trends-report-
2020>.
World Bank 2020c, World Bank to Promote Sustainable Energy and Digital Development in
Senegal, World Bank, Washington DC, <https://www.worldbank.org/en/news/press-
release/2019/12/18/world-bank-to-promote-sustainable-energy-and-digital-development-in-
senegal>.
Yiew, T-H & Lau, E 2018, 'Does foreign aid contribute to or impeded economic growth?',
Journal of International Studies vol. 11, no. 3, pp. 21-30.
Yogo, TU 2017, 'Assessing the effectiveness of foreign aid in the education sector in Africa:
the case of primary education', African Development Review, vol. 29, no. 3, pp. 389-402.
100
Zhang, D, Li, J & Han, P 2019, 'A multidimensional measure of energy poverty in China and
its impacts on health: An empirical study based on the China family panel studies', Energy
policy, vol. 131, pp. 72-81.
Zhao, M, Zhou, Y, Li, X, Cao, W, He, C, Yu, B, Li, X, Elvidge, CD, Cheng, W & Zhou, C 2019,
'Applications of satellite remote sensing of nighttime light observations: Advances, challenges,
and perspectives', Remote Sensing, vol. 11, no. 17, p. 1971.
101
Appendix 3.A
Table A3.1 – Description of aid projects in Senegal
Type of aid project Number of projects

Agriculture 34
Forestry 13
Industry 9
Health 77
Education 45
Government and civil society 24
Trade policy and regulations 2
Transport and storage 77
Water supply and sanitation 31
Business and other services 3
Energy generation and supply 36
Communications 1
Other social infrastructure and services 37
102
Appendix 3.B
Variable Description Mean SD

Age Age of household head 52.369 14.637
Age squared Square of age 2956.763 1600.082
Urban 1 if the respondent lives in the urban areas 0.370 0.483
Married 1 if the respondent is married 0.887 0.316
Gender 1 if the household head is male 1.235 0.424
Household size Number of persons in household 12.159 7.977
Employment 1 if the respondent is currently working 0.441 0.497
Primary education 1 if the household head’s highest education level is primary level 0.131 0.338
Secondary education 1 if the household head’s highest education level is secondary level 0.076 0.265
Tertiary education 1 if the household head’s highest education level is tertiary level 0.026 0.160
Education 1 if the household head’s highest education level is primary or 0.092 0.289
secondary or tertiary
Wealth index Household wealth index score -0.442 84.591
Wealth 1 if the household wealth index score is positive 0.523 0.499
Electricity access 1 if the household has access to electricity 0.462 0.499
Cooking fuels 1 if the household uses clean cooking fuels 0.006 0.080
Electric – appliances 1 if the household has electric appliances (e.g., fridge) 0.176 0.381
Entertainment means 1 if the household owns a TV or radio 0.800 0.400
Communication 1 if household owns a landline or mobile phone 0.730 0.444
means
Health 1 if the household head has not had a cough, fever, diarrhoea in the 0.239 0.426
last two weeks before survey interview
MEPI status 1 if the household is energy-poor (according to the multidimensional 0.825 0.380
energy poverty index)
MEPI score Multidimensional energy poverty score 0.657 0.179
Night-time light Mean of night-time light 6.497 16.523
Total aid (5 km) 1 if the household is within a 5 km distance from any aid project 0.286 0.452
Economic aid (25 km) 1 if the household is within a 25 km distance from an Economic 0.471 0.499
infrastructure aid project.
Social aid (25 km) 1 if the household is within a 25 km distance from a social 0.524 0.499
infrastructure aid project.
Production aid (25 km) 1 if the household is within a 25 km distance from a production aid 0.359 0.480
project
103
Appendix 3.C
Table A3.3 – Indicators and weights for MEPI
Dimension Indicator (weight) Energy poor/deprived if…
Accessibility Cooking fuel type/energy Clean/modern cooking fuel Household uses other cooking fuels
source (0.4) beside electricity, LPG, natural gas,
or biogas
Electricity Access to electricity (0.2) Household has no access to

electricity
Affordability Household electronic Ownership of household Household has no fridge

appliances appliances (0.13)
Communication Ownership of Household does not own a mobile

communication means phone or landline
(0.13)
Household electronic Ownership of an Household does not own a TV or

entertainment appliances entertainment appliance radio
(0.13)
104
Appendix 3.D
Table A3.4 – Effect of economic growth on foreign aid allocation
(1) (2)
VARIABLES Total Aid Commitments Total Aid Disbursements
Night-time lights -0.020 -0.016

(0.013) (0.010)
Observations 561 561

R-squared 0.152 0.152
Notes: All regressions include district and year fixed effects.
*** p<0.01, ** p<0.05, * p<0.1
105
Appendix 3.E
Table A3.5 – Main results
(1) (2) (3) (4)
VARIABLES MEPI status MEPI score MEPI status MEPI score
Time -0.014** -0.012*** -0.000 -0.003*

(0.005) (0.002) (0.005) (0.002)
Total Aid (25 km) -0.016 -0.039*** 0.033*** -0.009*
(0.012) (0.005) (0.011) (0.005)
Time *Total Aid (25 km) -0.032*** -0.009* -0.033*** -0.009**
(0.011) (0.005) (0.010) (0.004)
[-0.042] [-0.024]
Male -0.086*** -0.056***
(0.009) (0.003)
[-0.089] [-0.128]
Age 0.002* -0.001**
(0.001) (0.000)
[0.069] [-0.109]
Age squared -0.000*** -0.000
(0.000) (0.000)
[-0.138] [0.021]
Married 0.013 -0.020***
(0.013) (0.005)
[0.009] [-0.034]
Employed -0.011* -0.002
(0.006) (0.002)
[-0.012] [-0.003]
Primary education -0.077*** -0.044***
(0.009) (0.004)
[-0.066] [-0.081]
Secondary education -0.201*** -0.094***
(0.014) (0.004)
[-0.133] [-0.135]
Tertiary education -0.387*** -0.127***
(0.027) (0.007)
[-0.137] [-0.099]
Household size -0.005*** -0.003***
(0.000) (0.000)
[-0.116] [-0.174]
Urban area -0.224*** -0.154***
(0.011) (0.004)
[-0.276] [-0.416]
Night-time lights -0.001*** -0.001***
(0.000) (0.000)
[-0.045] [-0.064]
Observations 17,227 17,227 17,227 17,227
R-squared 0.130 0.269 0.224 0.421
Notes: MEPI status represents the energy deprivation dummy, and MEPI score stands for energy
deprivation score. All regressions include district fixed effects. Robust standard errors in parentheses.
Standardised coefficients in brackets. *** p<0.01, ** p<0.05, * p<0.1
106
Appendix 3.F
Table A3.6 – Aid and energy poverty over time
(1) (2)
Panel A: 1997 and below
Time *Total Aid (25 km) -0.088*** -0.038***
(0.013) (0.009)

R-squared 0.041 0.015
Panel B: 2005
Time *Total Aid (25 km) -0.068*** -0.027***
(0.010) (0.008)

R-squared 0.058 0.021
Panel C: 2010–11
Time *Total Aid (25 km) -0.149*** -0.064***
(0.020) (0.016)

R-squared 0.051 0.020
Panel D: 2012–13
Time *Total Aid (25 km) -0.079*** -0.023**
(0.015) (0.011)

R-squared 0.045 0.007
*** p<0.01, ** p<0.05, * p<0.1
107
Appendix 3.G
Table A3.7 – Parallel trend analysis
Variable Control Treatment Diff p-value

Main outcomes
MEPI status 0.412 0.490 0.047 0.422
(0.492) (0.500) (0.056)
MEPI score 0.358 0.407 -0.052 0.125
(0.417) (0.369) (0.031)
Individual indicators of energy poverty
Cooking fuels 0.001 0.003 0.009 0.038
(0.029) (0.057) (0.004)
Electricity access 0.130 0.436 0.172 0.069
(0.336) (0.496) (0.084)
Electric appliances 0.057 0.188 -0.045 0.451
(0.231) (0.390) (0.058)
Entertainment means 0.751 0.837 0.050 0.295
(0.433) (0.369) (0.045)
Communication means 0.088 0.259 0.104 0.166
(0.284) (0.438) (0.070)
Notes: This exercise is based on baseline conditions.
Standard errors in parentheses.
108
Appendix 3.H
Figure 3.1 – Distribution of DHS clusters with energy poverty
109
Appendix 3.I
Figure 3.2 – Distribution of aid projects and DHS clusters
110
Chapter 4
Temperature Shocks, Household Borrowing and the Mediating Role of Foreign

Aid: Evidence from Household Panel Data for Malawi
4.1 Introduction
Temperature shocks remain a threat to livelihoods globally (Dell et al. 2014). Generally
defined as temperature deviations from historical values, temperature shocks
undermine many outcomes, including health, economic, social, and political
outcomes. For instance, the literature on the relationship between weather shocks and
health shows that temperature extremes heighten mortality among the elderly, infants
and people with pre-existing health problems; amplify pre-existing medical issues; lead
to poor mental health; and trigger other health-related issues, such as heat strokes,
heat cramps, dehydration, disruption of health infrastructure, and cardiovascular and
respiratory diseases (see, e.g., Andalón et al. 2016; Calleja-Agiusa et al. 2020; CDC
2020; Hajat et al. 2010; Li et al. 2020). Some recent evidence connects cold
temperature shocks to the increased risk of the spread of the COVID-19 virus (see,
e.g., Mecenas et al. 2020; Ujiie et al. 2020). At the same time, temperature shocks
compromise the growth of economies by undermining labour productivity, investment
and output from production sectors, among others (see, e.g., Acevedo et al. 2020;
Colacito et al. 2019; Dell et al. 2012; Dellink et al. 2017). Evidence shows that a 1°C
rise in temperature, for example, lowers economic growth by 1.3 percentage points
(Lee et al. 2020). Assuming these temperature shocks persist, an estimated output
loss of about 9% is expected by 2100, particularly in low-income countries (Acevedo
et al. 2020). In the context of social and political outcomes, the literature shows that
temperature shocks initiate and facilitate conflict (see, e.g., Bellemare 2015; Bollfrass
& Shaver 2015; Salehyan & Gleditsch 2006) and drive households into severe poverty
(see, e.g., Azzarri & Signorelli 2020; Baez et al. 2020).
Empirically, a few studies have started examining the impact of weather shocks on
household financial decisions (Adjognon et al. 2020; Newman & Tarp 2020; Pelka et
al. 2015). Employing a two-step estimation technique on sequential logit and linear
probability models, Pelka et al. (2015) examined the effect of rainfall shocks on the
repayment performance on loans granted to small-scale farmers in Madagascar.
111
Similarly, using rainfall information, Newman and Tarp (2020) explored the impact of
rainfall shocks on farmers’ investment decisions in Viet Nam. In the same context, in
Nigeria, Adjognon et al. (2020) investigated the effect of weather-induced productivity
shocks on the repayment behaviour of rural credit market borrowers.
Much of what we know on weather shocks and household financial conditions and
decisions has focused on rainfall-related shocks and how these shocks influence loan
repayment performance and borrowers’ investment decisions and repayment
behaviours. Far less is known about how temperature shocks influence household
borrowing. Thus far, to the best of our knowledge, no study has specifically
investigated the relationship between temperature shocks and household borrowing.
Accordingly, we provide the first study to fill this important gap in the literature by
examining the impact of temperature shocks on household borrowing in Malawi. The
impact of temperature shocks on household borrowing here is ambiguous. It might be
that shocks lead to an increase in the demand for credit or the loss of income and
assets associated with temperature shocks 1) completely discourage borrowing, or 2)
reduce the amount of money households are able to borrow.
We are motivated to focus on household borrowing by the literature that associates

household borrowing with several important household and community-related
outcomes (see, e.g., Alter et al. 2018; Berger et al. 2016; Brewer 2020; Clayton et al.
2015; Hiilamo & Grundy 2020; Keese & Schmitz 2011; van Beek et al. 2021). This
literature suggests that in the long term, household borrowing increases household
debt, which is linked with poor health (e.g., undermined physical and mental health,
reduced life expectancy, depression and increased premature mortality), job loss,
heightened crime, and food insecurity, among others. Hence, by examining the effect
of temperature shocks on household borrowing, we improve our understanding of
household borrowing determinants and influence the designing of policies that govern
credit borrowing.
In addition to exploring the effects of temperature shocks, we also examine the

mediating role of foreign aid between temperature shocks and household borrowing.
The motivation behind focusing on foreign aid is that most governments in developing
countries lack the required domestic resources to fund their socio-economic
development initiatives (e.g., government transfers) and rely on foreign aid (Munyanyi
112
et al. 2020). At the same time, a large proportion of households in developing countries
live below the poverty datum line and highly depend on government transfers (see,
e.g., Hanna & Olken 2018; Shepherd et al. 2011). This means that, without foreign aid
or any other form of external or internal assistance to supplement the government
budget, most governments may fail to adequately provide such transfers, which may
compel households to engage in credit borrowing. However, the allocation of foreign
aid is influenced by a number of factors, including conditions in the aid recipient
country (e.g., conflict, institutions and policies, poverty and economic growth levels,
education and health statuses, and the presence of weather-induced natural disasters)
(see, e.g., Alesina & Dollar 2000; de Ree & Nillesen 2009; Mahembe & Odhiambo
2019; Weck-Hannemann & Schneider 2019). Therefore, by examining the mediating
role of foreign aid, we shed light on how temperature shocks influence foreign aid and
how these shocks transmit to household borrowing via aid.
We use the Malawian household data covering the years 2010–11, 2013 and 2016,
collected from the World Bank Microdata Library to examine the effect of temperature
shocks on household borrowing. Using district-level identifiers, we geographically
match this household data with the weather and foreign aid data from the University
of Delaware’s online database and the AidData website, respectively. Consistent with
the literature, household borrowing is measured using: 1) a binary variable that equals
one if the household borrowed on credit in the previous 12 months, 2) an indicator that
captures the number of loans obtained in the past 12 months, and 3) the amount of
credit borrowed over the past 12 months. In line with the literature, we measure
temperature shocks using temperature deviations from long-run values, while foreign
aid is captured using total foreign aid disbursements. Estimating a panel fixed effects
model and controlling for relevant fixed effects and household characteristics, we find
that temperature shocks positively affect household borrowing. These results are
robust to several sensitivity checks, including the use of alternative indicators of
household borrowing and temperature shocks and the choice of lags, among others.
Moreover, we explore foreign aid as a potential transmission channel through which
temperature shocks may influence household borrowing, and find that: total aid;
education aid; health aid; rural development aid; and water, sanitation and irrigation
aid are the channels through which the impacts of temperature shocks on household
borrowing are mediated.
113
Malawi makes for an interesting case study for at least three reasons. First, worldwide,
Malawi is regarded as one of the countries that are most vulnerable to climate change
and climate variability due to its disproportionate reliance on the agricultural sector for
food, exports and employment (Warnatzsch & Reay 2019). Specifically, the Malawian
agriculture sector contributes approximately 32% to GDP, 64.1% of the total labour
force and 80% of the nation’s export revenues (FAO 2021). However, given that
rainfall is no longer predictable, and temperatures have risen, reaching up to 40˚C in
many districts in Malawi (Mountain Research Initiative 2021), weather shock-induced
natural disasters such as floods and droughts are continuously putting pressure on the
country’s water supplies and food security (FAO 2021).
Second, Malawi heavily relies on foreign aid and has been a major aid recipient
country since 1964 (Page 2019). Over the last five decades, the country has received
more than US$33 billion in aid (World Bank 2020a) and has been funding a large part
of its national budget using aid. For instance, in 2011, foreign aid funded about 40%
of the Malawian national budget (Page 2019). Third, Malawi’s financial sector has
transformed in the past decade. As of 2009, the sector was concentrated, small, and
serviced a narrow clientele, mainly urban dwellers (World Bank 2021). However,
between 2011 and 2018, the country experienced digital transformation and increased
credit access, raising the proportion of the adult population with access to financial
services from 19% to 40% (World Bank 2021). In addition, the country has been
receiving significant amounts of capital to boost financial services' coverage and the
financial sector’s performance. In 2020, Malawi received US$86 million credit from the
World Bank to improve access to financial services (World Bank 2020c).
We make important contributions to several strands of literature. First, we contribute

to the literature on the determinants of household borrowing (see, e.g., Cui et al. 2017;
Fungáčová & Weill 2015; Kiplimo et al. 2015). We demonstrate that temperature
shocks are a significant determinant. Additionally, we also show that aid is an
important mediator and further contribute to the literature on the impact of and
effectiveness of foreign aid (see, e.g., Feeny & Ouattara 2013; Mahembe & Odhiambo
2019; Munyanyi et al. 2020).
Second, we make contributions to the literature that has examined the effects of
various types of shocks (see, e.g., Burke et al. 2015; Colacito et al. 2019; García-
114
Gómez 2011; Matz et al. 2015; Minale 2018). A large proportion of this literature has
examined the impacts of shocks in price, agriculture productivity, health, income, and
weather (e.g., rainfall and temperature), on outcomes such as food security, migration,
employment, health, economic growth, investment, poverty, conflict, and disease
prevalence, among others. We, however, contribute to this literature by specifically
focusing on the effects of temperature shocks on household borrowing.
A closely related number of studies in this literature examine the nexus between
weather shocks and the loan market (see, e.g., Adjognon et al. 2020; Newman & Tarp
2020; Pelka et al. 2015). These studies have improved our understanding on how
weather-related shocks, particularly, rainfall influence credit risk and borrowers’
behaviours and investment decisions. We add to these studies by showing that
temperature shocks influence aid allocation and that this serves as an important
mechanism through which temperature shocks transmit to household borrowing.
Thirdly, we significantly contribute to the climate economy literature, which associates

the understanding of climate change impacts with improved probability of devising
policies that may address the adverse consequences of climate change and extreme
weather events (see, e.g., Cachon et al. 2012; Graff Zivin & Neidell 2014; Seppanen
et al. 2006). Specifically, we answer the call from Dell et al. (2014), who concluded
that there is a need to explore other outcomes which may have a significant impact on
economic development. In response, we examine household borrowing as the
outcome – an important factor in the economic development equation (see, e.g., Amoo
et al. 2017; Duican & Pop 2015). In this regard, we contribute to both the literature on
the economic impacts of climate change and the climate economy literature.
The rest of the study is structured as follows: next, we explore relationships between
temperature shocks, household borrowing and foreign aid. Section 4.3 explains the
data and variables, whereas Section 4.4 details the methodology. Sections 4.5 and
4.6 interpret the results and conclude, respectively.
4.2 Literature Review
This section discusses how foreign aid mediates the relationship between temperature
shocks and household borrowing, and how temperature shocks impact household
115
borrowing. We explore these discussions in turn, and later evaluate the mediating role
of foreign aid between temperature shocks and household borrowing.
4.2.1 Temperature shocks and household borrowing: the mediating role of

foreign aid
Here, we postulate that temperature shocks create conditions that can either attract or
deter the flow of foreign aid, influencing household borrowing in the process. We
discuss at least four conditions that are associated with temperature shocks, which
can influence aid allocation.
(i) The ‘lower economic growth condition’
Temperature shocks are typically associated with lower levels of economic growth.
Evidence suggests that temperature shocks weaken economic growth by undermining
investment, labour productivity, trade, political stability, and performance of production
sectors, such as agriculture, forestry, industry and fishing (see, e.g., Acevedo et al.
2020; Colacito et al. 2019; Dell et al. 2012; Dellink et al. 2017). Specifically, in the
context of labour productivity, temperature shocks increase mortality rates and the
prevalence of various health issues, thus compromising labour supply and productivity
(Acevedo et al. 2020). Additionally, by lowering the rate of factor accumulation,
temperature shocks hinder investment growth (Acevedo et al. 2020). The literature
further demonstrates that weather-related shocks compromise trade, which is a
significant contributor to economic growth, by disrupting the functioning of transport,
distribution and supply chains, and altering trade patterns and comparative
advantages of countries (Dellink et al. 2017). Thus, trade-dependent economies (e.g.,
countries that primarily trade in agricultural commodities) tend to incur significant
economic loss resulting from lower exports and weakened trade positions (see, e.g.,
Dallmann 2019; Dellink et al. 2017; Papaioannou & de Haas 2017). With regards to
the production sectors, for example, the agricultural sector, extreme temperatures
reduce agricultural productivity and induce farmers to increase the use of their lands
and change their crop mix – leading to significant drops in yields (Aragón et al. 2018).
116
(ii) The ‘poor health condition’
Temperature shocks are associated with several health problems. The literature
suggests that extreme temperatures undermine health by causing dehydration, heat
exhaustion and heat- and cold-related stress (see, e.g., Al-Bouwarthan et al. 2019;
Cheshire Jr 2016; Kenny et al. 2018; Lim et al. 2015). On the one hand, higher
temperatures trigger dehydration by expanding blood vessels and causing excessive
sweating, whereas, on the other hand, cold temperatures shrink blood vessels, lower
peripheral circulation and trigger urinary voiding (Lim et al. 2015). Heat exhaustion
results in various health issues, including muscle cramps, mental fatigue, confusion,
severe headaches, weakness and poor muscle coordination (see, e.g., Armstrong et
al. 2007; Becker & Stewart 2011; Casa et al. 2015; Cheshire Jr 2016; Glazer 2005;
Jay & Kenny 2010; Kjellstrom et al. 2016), while heat- and cold-related stress impair
cognitive functioning by lowering thermal comfort and altering brain blood flow (Taylor
et al. 2016). Extreme temperatures have also been found to increase hospitalisation,
cause death, especially among vulnerable groups (e.g., the elderly, children and
economically disadvantaged), and worsen chronic health conditions, such a
cerebrovascular, respiratory and cardiovascular diseases, and diabetes and kidney-
related conditions (Kim 2016; Yang et al. 2019). Exposure to higher temperatures is
also associated with several infant health challenges. For instance, exposure to heat
waves during pregnancy decreases infant birthweight, while exposure to cold shocks
results in reduced infant birth length (Andalón et al. 2016).
(iii) The ‘conflict condition’
Temperature shocks are positively associated with conflict (see, e.g., Breckner &
Sunde 2019; Landis 2014; Maystadt et al. 2014) – with the effect larger in areas with
higher populations, lower levels of agricultural productivity and severe land
degradation (Breckner & Sunde 2019). Temperature shocks may lead to conflict by
diminishing agricultural output (Bollfrass & Shaver 2015), which in turn undermines
local labour markets and creates more income inequalities, resentment and tensions
in societies (Miguel et al. 2004). Undermined agricultural yields can cause migration –
a situation that can potentially kickstart and facilitate conflict and violence (Hsiang et
al. 2011). Specifically, evidence shows that migrants may facilitate the spread of civil
wars (Bollfrass & Shaver 2015) by altering the ethnic composition of states, worsening
117
economic competition, and enabling the spread of combatants, arms and ideologies
that are conducive to conflict (Salehyan & Gleditsch 2006). Further, a fall in agricultural
output tends to increase food prices, which may trigger social instability and increase
the incidence of violence (Bellemare 2015). In low-income countries, where economic
growth and temperature are negatively correlated, temperature shocks increase the
vulnerability of these countries to coups (Dell et al. 2012; Hsiang 2010). Evidence from
psychology also suggests that higher temperatures increase the propensity for
violence (Anderson et al. 2000).
(iv) The ‘poverty condition’
Weather-related shocks increase poverty and reduce food consumption per capita and
the number of meals consumed per day (see, e.g., Azzarri & Signorelli 2020; Baez et
al. 2020). As a result, some households tend to cope by increasing the supply of child
labour and selling assets (Baez et al. 2020). Extreme cold temperatures undermine
household budgets, which results in poor households reducing their food expenditures
while increasing fuel expenditures (Bhattacharya et al. 2003). By undermining
agricultural and labour productivity, temperature shocks cause a decline in household
income (Letta et al. 2017). People employed in sectors affected by shocks may lose
their jobs – thus, increasing the severity of poverty. Weather-related shocks also
exacerbate inequality and affect the poor more than the rich, given that poor people
lack the required resources, for example, financial resources and access to formal
credit, insurance and savings, to address the adverse impacts of these shocks.
Additionally, poor households tend to lose a substantial proportion of their wealth and
other possessions during extreme weather events (Busso & Messina 2020).
The preceding discussion suggests that temperature shocks undermine economic

growth and health and increase poverty and conflict. This creates an economic and
institutional environment consistent with those that attract foreign aid (Furuoka 2017).
Specifically, the literature on foreign aid allocation identifies at least three main
reasons for donors’ decisions to allocate aid, namely: 1) donor interest – here, foreign
aid is availed based on the donors’ economic and geopolitical reasons (e.g., trade
benefits involved and the presence of natural resources in aid-recipient countries), 2)
recipient needs – which involves development-related challenges that potential aid-
recipient countries are facing (e.g., poverty, poor health, natural disasters, conflict, low
118
economic growth, low literacy and higher mortality) and 3) recipient merit – here, aid
is availed to recipients that have good institutions and policies, which can transform
the allocated aid into pro-poor public expenditure (see, e.g., Alesina & Dollar 2000;
Furuoka 2017; Mahembe & Odhiambo 2019; Weck-Hannemann & Schneider 2019).
In trying to improve the livelihoods of the aid recipient countries, donors usually target
areas and sectors that are deprived and likely to benefit from aid. For example, as
classified by the OECD Creditor Reporting System (OECD 2022a), aid can be targeted
to areas such as agriculture, water and sanitation, education, transport, health, rural
development, the financial sector and conflict resolution, among others. In the case of
this study, due to the temperature shocks-induced conditions discussed previously, it
is expected that foreign aid would be allocated to relevant sectors based on the
recipient’s needs, which include the need to address lower economic growth, poor
health, conflict and poverty conditions. According to this view, temperature shocks
would encourage foreign aid flow.
On the contrary, evidence also shows that conflicts deter donors from providing foreign
aid (see, e.g., de Ree & Nillesen 2009; Lis 2013). In particular, conflicts are often
characterised by internal tensions that tend to make aid expenditure more vulnerable
to corruption and the destruction of relevant infrastructure that may disrupt the donors’
commercial interests. Additionally, donors may decide not to allocate aid, fearing they
might be accused of interfering in the internal politics of aid recipient countries (Lis
2013). In this regard, weather-induced conflicts would discourage foreign aid flow. In
summary, we argue that the effect of temperature shocks on foreign aid is ambiguous.
Therefore, either a positive or negative effect of temperature shocks on foreign aid
allocation should be expected.
Foreign aid can positively and negatively influence household borrowing. For instance,
on the one hand, aid can increase credit borrowing via financial development.
Evidence shows that foreign aid positively affects financial development and credit
access (see, e.g., Maruta 2019; Tsaurai 2018). Specifically, aid targeted at areas or
sectors associated with the financial sector tends to promote sound financial policy
reforms and improve the provision of banking and financial services. Additionally, aid
improves the performance of financial intermediaries and eliminates financial market
failures and inefficiencies, including weak institutions and poor administrative
119
procedures (Maruta 2019). In such an environment, financial intermediaries would be
expected to provide attractive and favourable lending deals to customers (e.g., lower
interest rates), which may entice households to borrow in times of crisis.
On the other hand, foreign aid may mitigate the impacts of shocks by augmenting
budgets for socio-economic development activities such as government transfers,
which a large proportion of households in the developing countries rely on (Hanna &
Olken 2018; Shepherd et al. 2011). By doing so, there is a higher probability that
households will not resort to credit borrowing when faced with weather-related shocks.
In this case, foreign aid would be expected to reduce household borrowing.
4.2.2 Temperature shocks and household borrowing
By lowering economic growth, triggering conflicts, raising poverty levels and

undermining health (see, e.g., Acevedo et al. 2020; Azzarri & Signorelli 2020; Bollfrass
& Shaver 2015; Yang et al. 2019), temperature shocks increase the probability of
household borrowing. Specifically, lower levels of economic growth are associated
with high unemployment rates and limited job opportunities (see, e.g., Aghion & Howitt
1994; Chand et al. 2017; Villaverde & Maza 2009), which can compel most households
with no other sources of income to resort to borrowing. By undermining the agricultural
sector, temperature shocks heighten food insecurity and loss of income sources
among people employed in this sector or other sectors that depend on the agricultural
sector (see, e.g., Mirzabaev 2015; Ngoma et al. 2019), which may result in increased
household borrowing. Conflicts destroy property, businesses and discourage
investment, hindering employment opportunities and causing more unemployment.
Faced with such situations, most affected households may have no other choice but
to borrow. Those with undermined health due to temperature shocks may face
difficulties in securing employment, leaving them with no alternatives to sustain
themselves except by borrowing. In summary, temperature shocks leave most
households in a worse-off position, which the households are likely to respond by
sourcing credit from various lenders.
We draw on data from three main sources. Household data for Malawi is obtained
from the World Bank Microdata Library – an online library that provides a wide range
120
of information concerning developing countries, including information on the people,
institutions, environment, communities and economic activities.8 The library contains
datasets produced by the World Bank (e.g., impact evaluation datasets) as well as
other international institutions and statistical agencies. From the library, the study
extracts the 2016 Integrated Household Panel Survey (IHPS), which is part of Malawi’s
fourth Integrated Household Survey field operation. The 2016 IHPS was funded by the
Malawian government, the World Bank and Millennium Challenge Corporation, and
targets 1,989 households, traced back to 2013 IHPS and 2010/11 IHPS. The IHPS
series is nationally representative and includes information on households (e.g., credit
access, education, health, housing, consumption, expenditure, income and assets),
agriculture (e.g., livestock, livestock products, extension services access and network
roster), fishery (e.g., fisheries: calendar, labour, inputs and output) and community
(e.g., economic activities, community activities, resource management and communal
organisation). We use household longitudinal data from the 2010/11 IHPS, 2013 IHPS
and 2016 IHPS and restrict the sample to households that participated in all survey
periods.
Historical station weather data are obtained from the University of Delaware’s publicly
available online climate database – a database compiled using various sources,
including the Global Historical Climatology Network Monthly Version 3 (Lawrimore et
al. 2011), the Daily Global Historical Climatology Network archive (Menne et al. 2012),
the Atmospheric Environment Service/Environment Canada archive, and the State
Hydrometeorological Institute records, among others.9 The database provides monthly
weather data for each station, interpolated to a 0.5 by 0.5-degree grid resolution and
centred on the 0.25 degree. For this study, the gridded monthly-mean air temperature
time series (Version 5.01) data for the period 1900–2017 are extracted (Matsuura &
Willmott 2018). The gridded fields were generated using various spatial interpolation
techniques, such as the traditional interpolation, climatologically aided interpolation,
and digital elevation-model assisted interpolation (Willmott & Matsuura 1995; Willmott
& Robeson 1995; Willmott et al. 1985).
8 https://microdata.worldbank.org/index.php/about
9 http://climate.geog.udel.edu/~climate/html_pages/download.html#T2017
121
We obtain foreign aid data from AidData – an online database run by the College of
William & Mary (USA), which seeks to make development-related finance more
accountable, effective and transparent (Tierney et al. 2011). The database is freely
available and reports on multilateral and bilateral aid funds, sourced from 96 donor
institutions and agencies, from 1945 to date.10 Data on multilateral and bilateral aid
projects are extracted from several sources, including 1) donor systems and published
documents and reports (i.e., data published by donor governments), 2) aid project
records, 3) aid-recipient systems (e.g., Aid Information Management Systems) and 4)
the OECD Creditor Reporting System, among others. The database also records
official non-OECD multilateral and bilateral project-level foreign aid data as well as the
Chinese Development Finance data. Additionally, the database specifies the type of
aid that is being availed, including production aid (e.g., agriculture, industry, fishery,
forestry), social infrastructure and services aid (e.g., education, health, WSS), and
economic infrastructure and services (energy, communications, transport and
storage), among others. The existing literature on the efficacy of sub-national aid is
largely based on data from this database (see, e.g., Anaxagorou et al. 2020; Civelli et
al. 2018; Marty et al. 2017; Odokonyero et al. 2018). To identify the locations of foreign
aid projects, the AidData and Uppsala University developed an approach for
geocoding these projects, allowing us to know their precise locations (e.g., country,
state, province, district, and exact location) and be able to merge this aid dataset with
any geocoded dataset. We extract geocoded data on foreign aid disbursed to Malawi
by 30 donor institutions since 2000 (Peratsakis et al. 2012).
Given that we are mainly interested in examining the mediating role of foreign aid at
the household level, following the existing literature on the efficacy of sub-national
foreign aid (see, e.g., Marty et al. 2017), we consider households in each district. We
geographically match each household with temperature and foreign aid information
from the district in which they live. Given that the household data are for the periods
2010/11, 2013 and 2016, for the main analysis, we focus on aid projects allocated
during these surveys years and do the same with temperature data.
122
4.3.1 Household Borrowing
We measure the main outcome variable, household borrowing, using three indicators.
Consistent with the existing literature, the first indicator of borrowing (CREDIT
BORROWING1) captures the household’s probability of borrowing from various
sources (see, e.g., Fungáčová & Weill 2015). Specifically, this indicator is based on
the IHPS question, which asks respondents: “Over the past 12 months, did you or
anyone else in this household borrow on credit from someone outside the household
or from an institution for business or farming purposes?”. We set a dummy variable
equal to 1 if the respondent answered ‘Yes’ to the question. The second indicator
(CREDIT BORROWING2) reflects the number of loans that the household acquired in
the previous periods (see, e.g., Kendall et al. 2010). The IHPS poses the question:
“Over the past 12 months, how many loans did you obtain?”. The third indicator of
borrowing (CREDIT BORROWING3) captures the amount of credit (in Malawian
Kwacha) the household obtained in the last 12 months (see, e.g., Kuri & Laha 2011).
4.3.2 Temperature shock
We derive the mean annual temperature of each district in a given year by taking the
total of the extracted monthly temperature values per year and dividing the result by
12 months. Consistent with the literature (see, e.g., Björkman-Nyqvist 2013; Feeny et
al. 2021; Maccini & Yang 2009; Rocha & Soares 2015), we construct the temperature
shock variable by observing temperature deviations from the mean temperature value.
In particular, the temperature shock variable is measured as the deviation of the
temperature (observed in a particular year for a given district) from the historical mean
temperature for that district. Mathematically, this is specified as follows:
𝑇𝑑𝑡 − ̅̅̅
𝑇𝑑
𝑇𝑆𝑑𝑡 = (1)
𝑇𝑠𝑑
where, 𝑇𝑆𝑑𝑡 represents temperature shock for district 𝑑 at year 𝑡. 𝑇𝑑𝑡 denotes the
observed mean temperature for district 𝑑 at year 𝑡. ̅̅̅
𝑇𝑑 represents the long-
run/historical mean temperature for district 𝑑, while 𝑇𝑠𝑑 is the long-run standard
deviation of temperature. In robustness checks, we create the temperature shock
variable using various cut-offs, including 0.5, 1, 1.5 and 2 standard deviations.
123
4.3.3 Mechanisms
We consider foreign aid as a potential channel through which temperature shocks may
influence household borrowing. The foreign aid dataset assigns a precision code to
the location of each aid project, which indicates how close the GPS point of the
reported aid project is to the exact location of the physical establishment of the foreign
aid project in question. A precision code of 1 corresponds to an exact location;
precision code 2 defines a location as being near or within 25km radius of an exact
location; precision code 3 represents a second-order administrative division (ADM2)
such as a district; and precision code 4 corresponds to a first-order administrative
division (ADM1), such as a region, province or state. Given that we examine the impact
of district-level temperature shocks on household borrowing, following the literature
(see, e.g., Anaxagorou et al. 2020; Marty et al. 2017), the indicator of foreign aid is
derived by considering all aid project locations assigned precision codes 1 to 3.
Specifically, to derive the foreign aid that was availed to a particular district 𝑑𝑖 in a
given year 𝑡, the study adds all foreign aid disbursement amounts corresponding to
precision codes 1 to 3 in that district and year. The mathematical representation is as
follows:
𝑇𝑜𝑡𝑎𝑙 𝐴𝑖𝑑 𝐹𝑙𝑜𝑤𝑠𝑑𝑖,𝑡 = ∑(𝑎𝑖𝑑𝑝1,𝑡 + 𝑎𝑖𝑑𝑝2,𝑡 + 𝑎𝑖𝑑𝑝3,𝑡 ) (2)
where 𝑎𝑖𝑑𝑝1,𝑡 corresponds to foreign aid flows of precision code 1 at time 𝑡; 𝑎𝑖𝑑𝑝2,𝑡
corresponds to foreign aid flows of precision code 2 at time 𝑡; and 𝑎𝑖𝑑𝑝3,𝑡 corresponds
to foreign aid flows of precision code 3 at time 𝑡. Following Anaxagorou et al. (2020),
we also assume that all foreign aid projects established at the district level equally
benefit the whole population. In testing whether foreign aid is a transmission channel,
we use different types of aid, including total aid, agricultural aid, education aid, health
aid, and rural development aid, among others.
4.3.4 Control variables
Consistent with the literature (see, e.g., Chandio & Jiang 2018; Fungáčová & Weill
2015; Ouattara et al. 2019), we control for several factors that are likely to influence
household borrowing, including gender, age, marital status, household income,
employment status, educational status and residence. We also control for district and
124
year fixed effects. A detailed description and summary statistics of these controls and
other variables are reported in Table A4.1 in the appendix.
4.4 Methodology
To examine the impact of temperature shocks on household borrowing, we estimate

the following equation:
𝐻𝐵𝑖𝑡 = 𝜑1 𝑇𝑆𝑑𝑡 + ∑ 𝜑𝑛 𝑋𝑛,𝑖𝑡 + 𝛾𝑑 + 𝛿𝑡 + 𝜌𝑖 + 𝜇𝑖𝑡

(3)
𝑛
where 𝐻𝐵𝑖𝑡 is the household borrowing indicator for household 𝑖 at year t. 𝑇𝑆𝑑𝑡
represents temperature shock for district 𝑑 at year 𝑡. In estimating the main regression,
we examine the effect of temperature shocks in the previous period (𝑡 − 1) for district
𝑑 in which household 𝑖 is residing. In robustness checks, we test the sensitivity of the
results using temperature shocks in the current period (𝑡) and other previous periods,
such as (𝑡 − 2) and (𝑡 − 3). 𝑋𝑛 is a vector of all the controls that are likely to influence
household borrowing. 𝜇 denotes the error term. The coefficient of interest is 𝜑1 , which
captures the effect of temperature shocks on household borrowing.
We estimate equation (3) using a panel fixed effect model with 𝛾𝑑 , 𝛿𝑡 and 𝜌𝑖 denoting
district, year and household fixed effects, respectively. District-level fixed effects
account for all unobservable and time-invariant factors, whereas year-fixed effects
take care of time/year characteristics. The household fixed effects control for time-
invariant and unobservable household differences.
4.5 Results
4.5.1 Main Results
Table 4.1 reports the main results for the link between temperature shocks and
household borrowing indicators: CREDIT BORROWING1, CREDIT BORROWING2
and CREDIT BORROWING3. Panels A and B display bivariate and multivariate
results, respectively. Columns 1, 2 and 3 report pooled OLS results, while Columns 4,
5 and 6 present panel fixed effects results.
125
Pooled OLS Panel Fixed Effects

(1) (2) (3) (4) (5) (6)
VARIABLES CB1 CB2 CB3 CB1 CB2 CB3
Panel A: Bivariate results
Temperature shocks (t-1) 0.048*** 0.063*** 0.534*** 0.043*** 0.058*** 0.492***
(0.012) (0.014) (0.110) (0.012) (0.014) (0.110)
Controls No No No No No No
Observations 2,802 2,802 2,802 2,802 2,802 2,802
R-squared 0.035 0.039 0.039 0.010 0.013 0.014
Panel B: Multivariate results
Temperature shocks (t-1) 0.044*** 0.059*** 0.485*** 0.034** 0.047*** 0.395***
(0.013) (0.014) (0.113) (0.013) (0.015) (0.119)
Controls Yes Yes Yes Yes Yes Yes

Observations 2,712 2,712 2,712 2,712 2,712 2,712
R-squared 0.054 0.059 0.060 0.029 0.038 0.037
Notes: CB1, CB2 and CB3 are abbreviated forms of CREDIT BORROWING1, CREDIT BORROWING2
and CREDIT BORROWING3, respectively. All regressions above include district and year fixed effects.
Controls include employment status, educational status, household income, residence, gender, age
and marital status. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Across all panels and columns, the results show that temperature shocks are positively
associated with household borrowing. Specifically, as shown in Columns 4 of Panel A,
the bivariate regression results report that temperature shocks increase the
household’s probability of borrowing by 4.3 percentage points. In Column 5, a unit
change in temperature shocks generates a 0.058-unit increase in household
borrowing. From Column 6, a one-unit change in temperature shocks generates a 49.2
percentage change in household borrowing.
The multivariate results presented in Panel B, which control for various household
characteristics, reinforce the bivariate regression findings.11 Particularly, in Columns
4, we find that temperature shocks increase the likelihood of household borrowing by
3.4 percentage points. Column 5 shows that a unit change in temperature shocks
leads to a 0.047-unit increase in household borrowing. In Column 6, the results show
that a unit change in temperature shocks creates a 39.5 percentage change in
borrowing among households. The effects of most of the control variables on
household borrowing are consistent with expectations. For instance, being employed
11 Full multivariate panel fixed effects results are provided in Table A4.2 in the appendix section.
126
and educated positively affect borrowing. Similarly, female and household income
variables are positively associated with borrowing.
4.5.2 Robustness Checks
In this section, we carry out a suite of tests to explore the robustness of our results.
First, we examine the sensitivity of our results to alternative measures of temperature
shocks. The main results in this study, presented in Table 4.1, are based on a measure
of temperature shocks that reflects the standard deviations from the long-term mean
temperature. In Table 4.2, Panels A to D, to verify the sensitivity of results, we create
cut-off points for the standard deviations.
Table 4.2 – Alternative measures of temperature shocks
(1) (2) (3)

VARIABLES CREDIT BORROWING1 CREDIT BORROWING2 CREDIT BORROWING3
Panel A: Using cut-off of 0.5 standard deviation

Temperature shocks (t-1) 0.068** 0.078** 0.720**
(0.034) (0.035) (0.316)
Observations 2,712 2,712 2,712

R-squared 0.026 0.032 0.032
Panel B: Using cut-off of 1 standard deviation
Temperature shocks (t-1) 0.077*** 0.100*** 0.870***
(0.021) (0.024) (0.190)

R-squared 0.034 0.043 0.044
Panel C: Using cut-off of 1.5 standard deviation
(0.023) (0.026) (0.204)

R-squared 0.031 0.041 0.041
Panel D: Using cut-off of 2 standard deviation
(0.026) (0.031) (0.234)

R-squared 0.031 0.042 0.041
Notes: All regressions above include district and year fixed effects, and the same control variables as
in Table 4.1. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Specifically, we generate alternative indicators of temperature shocks based on 0.5,

1, 1.5 and 2 standard deviations. Across all panels and columns, the results show that
127
temperature shocks have a positive and statistically significant effect on household
borrowing. Thus, the results are robust.
Second, in Table 4.3, we check the robustness of our results to alternative indicators
of household borrowing.
Table 4.3 – Alternative measures of credit access
(1) (2) (3)

Temperature shocks (t-1) 0.034** 0.047*** 0.056***

(0.013) (0.008) (0.008)

R-squared 0.029 0.047 0.053
In the main regression results, we use the ‘number of loans’ obtained by a household
(i.e., CREDIT BORROWING2) as one of the indicators for household borrowing. In
further checks, we set a binary variable (CREDIT BORROWING4) equal to 1 if the
household acquired a loan in the past 12 months. In Column 2, we capture household
borrowing using a binary variable (CREDIT BORROWING5) equal to 1 if any member
of the household has an account at a bank, credit union, microfinance institution,
village savings organisation, post office or another financial institution. In Column 3,
the measure for household borrowing (CREDIT BORROWING6) is a dummy variable
equal to 1 if 12 months before the survey interview, a member of the household used
an account at any of the financial institutions mentioned earlier. Evidence shows that
individuals who have an account at a financial institution are more likely to borrow than
those without – thus, in other studies, account ownership and the use of an account
(i.e., CREDIT BORROWING5 and CREDIT BORROWING6) are used to capture
household borrowing (see, e.g., Akeju, 2022; Yue et al., 2021). As shown across all
columns, the effect of temperature shocks on household borrowing remains robust.
Third, we verify the sensitivity of our results to different lags of the temperature shocks
variable. The main results explore the impact of temperature shocks in the previous
period (𝑡 − 1) on household borrowing. In further checks, as shown in Panels A to C
128
of Table 4.4, we consider the current period (𝑡) (i.e., no lags) and the second (𝑡 − 2)
and third (𝑡 − 3) lags of temperature shocks.
Table 4.4 – Lagging temperature shocks
(1) (2) (3)

Panel A: No lags
Temperature shocks (t) 0.008 0.013 0.126
(0.017) (0.018) (0.164)

R-squared 0.024 0.030 0.029
Panel B: 2-year lag
Temperature shocks (t-2) 0.005 0.019 0.059
(0.019) (0.021) (0.176)

R-squared 0.024 0.030 0.029
Panel C: 3-year lag
Temperature shocks (t-3) -0.009 0.007 -0.065
(0.021) (0.023) (0.190)

R-squared 0.024 0.030 0.029
Across all panels, we find that temperature shocks in the current, second and third
periods have no effect on household borrowing. Comparing these findings with the
main results, we observe that only temperature shocks of the previous period have a
statistically significant impact on household borrowing.
Fourth, given that most climate related variables tend to be correlated, and not
including them may lead to potential omitted variable bias, we test the robustness of
our results by controlling for rainfall. Table 4.5 reports the results. In Columns 1–3, we
find that temperature shocks have a positive effect on household borrowing. That is,
the results remain robust. We also find that across all columns, rainfall is negatively
associated with household borrowing.
129
Table 4.5 – Controlling for rainfall
(1) (2) (3)

Temperature shocks (t-1) 0.033** 0.046*** 0.390***

(0.013) (0.015) (0.119)
Rainfall -0.003** -0.003** -0.035***
(0.001) (0.001) (0.012)

R-squared 0.034 0.041 0.044
Fifth, we examine the sensitivity of our results to the type of financial institutions
households borrow from. The survey gives information on formal (e.g., commercial
banks) and informal financial sources (e.g., local merchants, employers and religious
institutions) that may provide credit to households. We investigate how temperature
shocks affect 1) households’ access to credit, 2) the number of loans acquired and 3)
the credit amounts granted by these financial sources. The results presented in Panels
A, B and C of Table 4.6 show that temperature shocks have a positive and statistically
significant impact on households’ access to credit, the number of loans borrowed, and
credit amounts obtained from informal institutions. This means that in times of crises
(e.g., weather-related shocks), only informal institutions are more likely to lend credit
to households. In this regard, the impact of temperature shocks on borrowing remains
robust.
130
Table 4.6 – Access to formal and informal financial institutions or credit
sources
(1) (2)
VARIABLES FORMAL INSTITUTIONS INFORMAL INSTITUTIONS
Panel A: Access to sources of credit
Temperature shocks (t-1) -0.022 0.418***
(0.036) (0.118)

R-squared 0.009 0.042
Panel B: Number of loans obtained
(0.003) (0.015)

R-squared 0.041 0.055
Panel C: Amount of credit obtained
(0.003) (0.118)

R-squared 0.008 0.042
Sixth, in Table 4.7, we explore the impact of temperature shocks on household

borrowing among male-and female-headed households and in rural and urban areas.
131
Table 4.7 – Heterogenous effects of temperature shocks
(1) (2) (3)
Panel A: Female-headed households
Temperature shocks (t-1) 0.030 0.037 0.319
(0.033) (0.035) (0.306)
Observations 580 580 580

R-squared 0.137 0.135 0.150
Panel B: Male-headed households
Temperature shocks (t-1) 0.033* 0.040** 0.404**
(0.017) (0.020) (0.157)
Controls Yes Yes Yes

R-squared 0.026 0.029 0.034
Panel C: Rural areas
Temperature shocks (t-1) 0.030* 0.037** 0.347***
(0.016) (0.017) (0.133)

R-squared 0.036 0.041 0.044
Panel D: Urban areas
Temperature shocks (t-1) 0.034 0.064* 0.388
(0.031) (0.037) (0.297)

Observations 806 806 806
R-squared 0.068 0.074 0.082
Notes: All regressions above include district and year fixed effects. Robust standard errors in
parentheses. *** p<0.01, ** p<0.05, * p<0.1
In Panels A and B, where we look at male-and female-headed households, we find

that temperature shocks only increase household borrowing among male-headed
households. This may be due to numerous women empowerment programmes that
have been initiated over the years that seek to improve the welfare of Malawian women
and eliminate gender inequalities.12 13 14 In this regard, as far as male-headed
households are concerned, the results are robust. Panel C results show that
temperature shocks have a positive and statistically significant effect on borrowing
12 https://www.un.org/africarenewal/news/african-union-project-malawian-women%E2%80%99s-
empowerment-agriculture-cooperation-norway-and-germany
13 https://www.usaid.gov/gender-equality-and-womens-empowerment
14 https://www.cpar.ca/Our-Programs/Where-We-Work/Malawi/Our-Stories/Microfinance-empowering-
women-in-Malawi
132
among households living in rural areas. In Panel D, where the study focuses on urban
areas, similar results to Panel C are observed, but the effect of temperature shocks
on household borrowing depends on the measure of credit borrowing used. Based on
Panel C and D, we may argue that the impact of temperature shocks on household
borrowing remains robust and is not affected by location.
4.5.3 Foreign aid as a potential mechanism
In Section 2, we discussed how foreign aid might mediate the relationship between
temperature shocks and household borrowing. To examine the mediating role of aid,
we follow the existing literature that uses a two-stage technique to validate variables
as mechanisms (see, e.g., Awaworyi Churchill et al. 2019; Prakash et al. 2020).
In the first step, for foreign aid to be qualified as a transmission channel, it has to be
correlated with temperature shocks. Next, as we include foreign aid as an additional
covariate in our main regression model (i.e., the model examining the effect of
temperature shocks on household borrowing) in the second step, the magnitude of the
coefficient on the temperature shocks variable should decrease or become statistically
insignificant.
Table 4.8 shows the association between temperature shocks and types of foreign aid
(i.e., Total Aid, Education Aid, Health Aid, Rural Development Aid, Agriculture Aid, and
Water, Sanitation and Irrigation Aid).
Table 4.8 – Correlation between temperature shocks and foreign aid
(1) (2) (3) (4) (5) (6)

VARIABLES Total Education Health Rural Agriculture Water, Sanitation
Aid Aid Aid Development Aid Aid and
Irrigation Aid
Temperature shocks (t-1) -8.581*** -1.482*** -5.643*** -3.192*** -0.136 -2.403***

(0.177) (0.163) (0.171) (0.202) (0.114) (0.172)
Observations 2,802 2,802 2,802 2,802 2,802 2,802

R-squared 0.384 0.034 0.318 0.153 0.000 0.079
parentheses. *** p<0.01, ** p<0.05, * p<0.1
As shown in Columns 1-6, temperature shocks are negatively and significantly

correlated with all types of foreign aid except Agriculture Aid. The results imply that
133
temperature shocks discourage foreign aid. Given that there is no correlation between
temperature shocks and Agriculture Aid, for the next step, we exclude Agriculture Aid.
In Table 4.9, we report the results obtained after separately adding foreign aid types
to the temperature shocks and household borrowing model. In Panels A, B and C, the
dependable variables are CREDIT BORROWING1, CREDIT BORROWING2 and
CREDIT BORROWING3, respectively. Column 1 shows the effects of temperature
shocks on household borrowing, with no extra covariates. In Panel 1, Columns 2-6,
the results show that all types of foreign aid have a negative and statistically significant
effect on household borrowing, and their inclusion either reduces the magnitude of the
coefficient on the temperature shocks variable or makes the effect of shocks on
household borrowing statistically insignificant. We also obtain similar results in Panels
B and C, where household borrowing is captured using CREDIT BORROWING2 and
CREDIT BORROWING3, respectively.
134
Table 4.9 – Foreign aid as a mediator between credit borrowing and
temperature shocks
VARIABLES (1) (2) (3) (4) (5) (6)
Panel A: CREDIT BORROWING1 as dependent variable
Temperature shocks (t-1) 0.043*** -0.001 0.038*** 0.013 0.020 0.028**
(0.012) (0.016) (0.013) (0.015) (0.013) (0.013)
Total Aid -0.005***
(0.001)
Education Aid -0.004**
(0.002)
Health Aid -0.005***
(0.002)
Rural Development Aid -0.007***
(0.002)
Water, Sanitation and Irrigation Aid -0.006***
(0.002)
Observations 2,802 2,802 2,802 2,802 2,802 2,802
R-squared 0.010 0.026 0.015 0.020 0.025 0.024
Panel B: CREDIT BORROWING2 as dependent variable
Temperature shocks (t-1) 0.058*** 0.005 0.051*** 0.022 0.028** 0.039***
(0.014) (0.017) (0.014) (0.017) (0.014) (0.014)
Total Aid -0.006***
(0.001)
Education Aid -0.004***
(0.002)
(0.002)
(0.002)
(0.002)
Observations 2,802 2,802 2,802 2,802 2,802 2,802

R-squared 0.013 0.031 0.018 0.025 0.032 0.029
Panel C: CREDIT BORROWING3 as dependent variable
Temperature shocks (t-1) 0.492*** -0.010 0.425*** 0.176 0.236** 0.329***
(0.110) (0.145) (0.113) (0.137) (0.118) (0.112)
Total Aid -0.059***
(0.011)
Education Aid -0.045***
(0.015)
(0.014)
(0.015)
(0.015)
Observations 2,802 2,802 2,802 2,802 2,802 2,802

R-squared 0.014 0.038 0.022 0.027 0.036 0.033
Notes: All regressions above include district and year fixed effects. Robust standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1
135
Based on these findings, we contend that Total Aid, Education Aid, Health Aid, Rural
Development Aid, and Water, Sanitation and Irrigation Aid are channels through which
temperature shocks influences credit borrowing among households.
4.6 Conclusion
Temperature shocks remain a global challenge. We provide the first study to examine
the impact of temperature shocks on household borrowing and the mediating role of
foreign aid. We obtain household data for Malawi from the World Bank Microdata
Library for the years 2010/11, 2013 and 2016. Next, we geographically match this data
with foreign aid and weather data from the AidData website and the University of
Delaware’s online database, respectively. Running a panel fixed effects model and
controlling for various household factors that may be correlated with household
borrowing, the results suggest that temperature shocks have a positive and statistically
significant impact on household borrowing. These results are robust to several checks,
including the 1) use of alternative measures of temperature shocks and credit
borrowing, 2) examination of the effect of temperature shocks on formal and informal
financial institutions, 3) control for rainfall and 4) investigation of the heterogeneous
impacts of temperature shocks across female-and male-headed households as well
as across rural and urban areas.
Policy-wise, these findings support government policies geared towards 1) eliminating

all activities behind these shocks (e.g., burning of fossil fuels), 2) promoting the use of
cleaner and renewable energies and 3) increasing awareness of factors contributing
to weather shocks and global warming. Additionally, policies that enable households
to adapt and become resilient to temperature shocks and also save for the future or
times of crisis should be established. The governments could also increase transfer
payments to lessen the burden of temperature shocks on these households, which
may end up forcing them to borrow.
The channel analysis results indicate that Total Aid, Education Aid, Health Aid, Rural
Development Aid, and Water, Sanitation and Irrigation Aid are the potential channels
through which temperature shocks may transmit to household borrowing. Moreover,
the analysis also shows that foreign aid reduces household borrowing. A major
implication of this finding is that governments and key policymakers need to put in
place policies that ensure the flow of foreign aid and support sectors targeted by
136
foreign aid (e.g., education, health, rural development, and water, sanitation and
irrigation). For instance, the government could invest heavily in human capital –
training people working in these sectors to be more effective, ensuring the foreign aid
availed to these sectors perform as intended. In the same context, given that foreign
aid is vulnerable to corruption and mismanagement, aid-recipient governments need
to increase surveillance over the usage of the allocated aid, increase accountability
and enact laws that discourage corrupt behaviours. By doing so, foreign aid may stand
a chance in lowering household borrowing and improving other outcomes.
This study suffers from two limitations. First, the main analysis in this study is based
on only three survey years. More years would have allowed the study to run several
analyses, including examining the time-varying relationship that may exist between
temperature shocks and household borrowing. Second, the study only examined
foreign aid as a possible transmission channel through which temperature shocks
transmit to household borrowing. Future studies could examine 1) the effect of weather
shocks on household borrowing over a period of more than three years, 2) the impact
of other types of weather shocks (e.g., rainfall, humidity), and 3) other channels that
mediate the relationship between weather shocks and household borrowing.
137
References
Acevedo, S, Mrkaic, M, Novta, N, Pugacheva, E & Topalova, P 2020, 'The Effects of Weather
Shocks on Economic Activity: What are the Channels of Impact?', Journal of Macroeconomics,
vol. 65, p. 103207.
Adjognon, GS, Liverpool-Tasie, LS & Shupp, R 2020, 'Productivity Shocks and Repayment
Behavior in Rural Credit Markets: A Framed Field Experiment', The Journal of Development
Studies, vol. 56, no. 10, pp. 1909-1926.
Aghion, P & Howitt, P 1994, 'Growth and unemployment', The Review of Economic Studies,
vol. 61, no. 3, pp. 477-494.
Akeju, KF 2022, 'Household financial behaviour: the role of financial inclusion instruments in
Nigeria', Journal of Sustainable Finance & Investment, pp. 1-13.
Al-Bouwarthan, M, Quinn, MM, Kriebel, D & Wegman, DH 2019, 'Assessment of heat stress
exposure among construction workers in the hot desert climate of Saudi Arabia', Annals of
Work Exposures and Health, vol. 63, no. 5, pp. 505-520.
Growth, vol. 5, no. 1, pp. 33-63.
Alter, A, Feng, AX & Valckx, N 2018, Understanding the macro-financial effects of household
debt: A global perspective, International Monetary Fund, Washington DC.
Amoo, G, Eboreime, MI, Adamu, Y & Belonwu, MC 2017, 'The impact of private sector credit
on economic growth in Nigeria', CBN Journal of Applied Statistics (JAS), vol. 8, no. 2, p. 1.
103430.
Andalón, M, Azevedo, JP, Rodríguez-Castelán, C, Sanfelice, V & Valderrama-González, D

2016, 'Weather shocks and health at birth in Colombia', World Development, vol. 82, pp. 69-
82.
Anderson, CA, Anderson, KB, Dorr, N, DeNeve, KM & Flanagan, M 2000, 'Temperature and
aggression', in Advances in Experimental Social Psychology, Elsevier, pp. 63-133.
Aragón, FM, Oteiza, F & Rud, JP 2018, Climate change and agriculture: farmer adaptation to
extreme heat, IFS Working Papers.
Armstrong, LE, Casa, DJ, Millard-Stafford, M, Moran, DS, Pyne, SW & Roberts, WO 2007,
'Exertional heat illness during training and competition', Medicine & Science in Sports &
Exercise, vol. 39, no. 3, pp. 556-572.
Azzarri, C & Signorelli, S 2020, 'Climate and poverty in Africa South of the Sahara', World
Development, vol. 125, p. 104691.
138
Baez, JE, Caruso, G & Niu, C 2020, 'Extreme weather and poverty risk: evidence from multiple
shocks in Mozambique', Economics of Disasters and Climate Change, vol. 4, no. 1, pp. 103-
127.
Becker, JA & Stewart, LK 2011, 'Heat-related illness', American Family Physician, vol. 83, no.
11, pp. 1325-1330.
Bellemare, MF 2015, 'Rising food prices, food price volatility, and social unrest', American
Journal of Agricultural Economics, vol. 97, no. 1, pp. 1-21.
Berger, LM, Collins, JM & Cuesta, L 2016, 'Household debt and adult depressive symptoms
in the United States', Journal of Family and Economic Issues, vol. 37, no. 1, pp. 42-57.
Bhattacharya, J, DeLeire, T, Haider, S & Currie, J 2003, 'Heat or eat? Cold-weather shocks
and nutrition in poor American families', American Journal of Public Health, vol. 93, no. 7, pp.
1149-1154.
Björkman-Nyqvist, M 2013, 'Income shocks and gender gaps in education: Evidence from
Uganda', Journal of development Economics, vol. 105, pp. 237-253.
Bollfrass, A & Shaver, A 2015, 'The effects of temperature on political violence: Global
evidence at the subnational level', PloS One, vol. 10, no. 5, p. e0123505.
Breckner, M & Sunde, U 2019, 'Temperature extremes, global warming, and armed conflict:
new insights from high resolution data', World Development, vol. 123, p. 104624.
Brewer, M 2020, 'Household debt and children’s risk of food insecurity', Social Problems, vol.
67, no. 3, pp. 565-584.
Burke, M, Gong, E & Jones, K 2015, 'Income shocks and HIV in Africa', The Economic Journal,
vol. 125, no. 585, pp. 1157-1189.
Busso, M & Messina, J 2020, The inequality crisis: Latin America and the Caribbean at the
Crossroads, Inter-American Development Bank, Washington D.C.
Cachon, GP, Gallino, S & Olivares, M 2012, 'Severe weather and automobile assembly
productivity', Columbia Business School Research Paper, vol. no. 12/37.
Calleja-Agiusa, J, England, K & Calleja, N 2020, 'The effect of global warming on mortality',
Early Human Development, vol., pp. 105222-105222.
Casa, DJ, DeMartini, JK, Bergeron, MF, Csillan, D, Eichner, ER, Lopez, RM, Ferrara, MS,
Miller, KC, O'Connor, F & Sawka, MN 2015, 'National Athletic Trainers' Association position
statement: exertional heat illnesses', Journal of Athletic Training, vol. 50, no. 9, pp. 986-1000.
CDC 2020, Climate and Health, Centers for Disease Control and Prevention, USA.
Chand, K, Tiwari, R & Phuyal, M 2017, 'Economic Growth and Unemployment Rate: An
Empirical Study of Indian Economy', Journal of Indian Economy, vol. 4, no. 2, pp. 130-137.
Chandio, AA & Jiang, Y 2018, 'Determinants of Credit Constraints: Evidence from Sindh,
Pakistan', Emerging Markets Finance and Trade, vol. 54, no. 15, pp. 3401-3410.
Cheshire Jr, WP 2016, 'Thermoregulatory disorders and illness related to heat and cold
stress', Autonomic Neuroscience, vol. 196, pp. 91-104.
139
Civelli, A, Horowitz, A & Teixeira, A 2018, 'Foreign aid and growth: A Sp P-VAR analysis using
satellite sub-national data for Uganda', Journal of Development Economics, vol. 134, pp. 50-
67.
Clayton, M, Liñares-Zegarra, J & Wilson, JO 2015, 'Does debt affect health? Cross country
evidence on the debt-health nexus', Social Science & Medicine, vol. 130, pp. 51-58.
Colacito, R, Hoffmann, B & Phan, T 2019, 'Temperature and growth: A panel analysis of the
United States', Journal of Money, Credit and Banking, vol. 51, no. 2-3, pp. 313-368.
Cui, Y, Sun, G, Siddik, MNA & Liu, X 2017, 'Analysis on determinants of rural household credit
in China', Journal of Interdisciplinary Mathematics, vol. 20, no. 5, pp. 1179-1201.
Dallmann, I 2019, 'Weather variations and international trade', Environmental and Resource
301-313.
Dell, M, Jones, BF & Olken, BA 2012, 'Temperature shocks and economic growth: Evidence
from the last half century', American Economic Journal: Macroeconomics, vol. 4, no. 3, pp.
66-95.
Dell, M, Jones, BF & Olken, BA 2014, 'What do we learn from the weather? The new climate-
economy literature', Journal of Economic Literature, vol. 52, no. 3, pp. 740-798.
Dellink, R, Hwang, H, Lanzi, E & Chateau, J 2017, International trade consequences of climate
change, OECD, Paris.
Duican, ER & Pop, A 2015, 'The implications of credit activity on economic growth in Romania',
Procedia Economics and Finance, vol. 30, pp. 195-201.
FAO 2021, Economic and Policy Analysis of Climate Change, Food and Agriculture
Organization of the United Nations, Rome, <http://www.fao.org/in-
action/epic/countries/mwi/en/>.
Feeny, S, Trinh, T-A & Zhu, A 2021, 'Temperature shocks and energy poverty: Findings from
Vietnam', Energy Economics, vol. 99, p. 105310.
Fungáčová, Z & Weill, L 2015, 'Understanding financial inclusion in China', China Economic
Review, vol. 34, pp. 196-206.
García-Gómez, P 2011, 'Institutions, health shocks and labour market outcomes across
Europe', Journal of Health Economics, vol. 30, no. 1, pp. 200-213.
Glazer, JL 2005, 'Management of heatstroke and heat exhaustion', American Family

Physician, vol. 71, no. 11, pp. 2133-2140.
140
Graff Zivin, J & Neidell, M 2014, 'Temperature and the allocation of time: Implications for
climate change', Journal of Labor Economics, vol. 32, no. 1, pp. 1-26.
Hajat, S, O'Connor, M & Kosatsky, T 2010, 'Health effects of hot weather: from awareness of
risk factors to effective health protection', The Lancet, vol. 375, no. 9717, pp. 856-863.
Hanna, R & Olken, BA 2018, 'Universal basic incomes versus targeted transfers: Anti-poverty
programs in developing countries', Journal of Economic Perspectives, vol. 32, no. 4, pp. 201-
226.
Hiilamo, A & Grundy, E 2020, 'Household debt and depressive symptoms among older adults
in three continental European countries', Ageing & Society, vol. 40, no. 2, pp. 412-438.
Hsiang, SM 2010, 'Temperatures and cyclones strongly associated with economic production
in the Caribbean and Central America', Proceedings of the National Academy of Sciences,
vol. 107, no. 35, pp. 15367-15372.
Hsiang, SM, Meng, KC & Cane, MA 2011, 'Civil conflicts are associated with the global
climate', Nature, vol. 476, no. 7361, pp. 438-441.
Jay, O & Kenny, GP 2010, 'Heat exposure in the Canadian workplace', American Journal of
Industrial Medicine, vol. 53, no. 8, pp. 842-853.
Keese, M & Schmitz, H 2011, 'Broke, ill, and obese: The effect of household debt on health',
Ruhr Economic Paper, vol. no. 234.
Kendall, J, Ponce, A & Mylenko, N 2010, Measuring financial access around the world, The
World Bank.
Kenny, GP, Wilson, TE, Flouris, AD & Fujii, N 2018, 'Heat exhaustion', in Handbook of Clinical
Neurology, Elsevier, pp. 505-529.
Kim, EJ 2016, 'The impacts of climate change on human health in the United States: A
scientific assessment, by us global change research program', Journal of the American
Planning Association, vol. 82, no. 4, pp. 418-419.
Kiplimo, JC, Ngenoh, E, Koech, W & Bett, JK 2015, 'Determinants of access to credit financial
services by smallholder farmers in Kenya', Journal of Development and Agricultural
Kjellstrom, T, Briggs, D, Freyberg, C, Lemke, B, Otto, M & Hyatt, O 2016, 'Heat, human
performance, and occupational health: a key issue for the assessment of global climate
change impacts', Annual Review of Public Health, vol. 37, pp. 97-112.
Kuri, PK & Laha, A 2011, 'Financial inclusion and human development in India: An inter-state
analysis', Indian Journal of Human Development, vol. 5, no. 1, pp. 61-77.
Landis, ST 2014, 'Temperature seasonality and violent conflict: The inconsistencies of a

warming planet', Journal of Peace Research, vol. 51, no. 5, pp. 603-618.
Lawrimore, JH, Menne, MJ, Gleason, BE, Williams, CN, Wuertz, DB, Vose, RS & Rennie, J
2011, 'An overview of the Global Historical Climatology Network monthly mean temperature
data set, version 3', Journal of Geophysical Research: Atmospheres, vol. 116, no. D19.
141
Lee, M, Gaspar, R & Villaruel, ML 2020, 'Effects of temperature shocks on economic growth
and welfare in Asia', Resources and Environmental Economics, vol. 2, no. 2, pp. 158-171.
Letta, M, Montalbano, P & Tol, RS 2017, Temperature shocks, growth and poverty thresholds:
evidence from rural Tanzania, Sapienza University of Rome, Rome.
Li, M, Ferreira, S & Smith, TA 2020, 'Temperature and self-reported mental health in the United
States', PloS One, vol. 15, no. 3, p. e0230316.
Lim, Y-H, Park, M-S, Kim, Y, Kim, H & Hong, Y-C 2015, 'Effects of cold and hot temperature
on dehydration: a mechanism of cardiovascular burden', International Journal of
Biometeorology, vol. 59, no. 8, pp. 1035-1043.
Peace and Security Journal, vol.110, pp. 283-294.
Maccini, S & Yang, D 2009, 'Under the weather: Health, schooling, and economic
consequences of early-life rainfall', American Economic Review, vol. 99, no. 3, pp. 1006-1026.
1, p. 1626321.
vol. 2, no. 1.
Maruta, AA 2019, 'Can aid for financial sector buy financial development?', Journal of
Macroeconomics, vol. 62, p. 103075.
Matsuura, K & Willmott, CJ 2018, 'Terrestrial air temperature: 1900–2017 gridded monthly
time series', University of Delaware, Newark, DE Retreived from. http://climate. geog. udel.
edu/~ climate/html_pages/Global2017/README. GlobalTsT2017. html, vol.
Matz, JA, Kalkuhl, M & Abegaz, GA 2015, 'The short-term impact of price shocks on food
security-Evidence from urban and rural Ethiopia', Food Security, vol. 7, no. 3, pp. 657-679.
Maystadt, J-F, Calderone, M & You, L 2014, 'Local warming and violent conflict in North and
South Sudan', Journal of Economic Geography, vol. 15, no. 3, pp. 649-671.
Mecenas, P, Bastos, RTdRM, Vallinoto, ACR & Normando, D 2020, 'Effects of temperature
and humidity on the spread of COVID-19: A systematic review', PloS One, vol. 15, no. 9, p.
e0238339.
Menne, MJ, Durre, I, Vose, RS, Gleason, BE & Houston, TG 2012, 'An overview of the global
historical climatology network-daily database', Journal of Atmospheric and Oceanic
Technology, vol. 29, no. 7, pp. 897-910.
Miguel, E, Satyanath, S & Sergenti, E 2004, 'Economic shocks and civil conflict: An
instrumental variables approach', Journal of Political Economy, vol. 112, no. 4, pp. 725-753.
Minale, L 2018, 'Agricultural productivity shocks, labour reallocation and rural–urban migration
in China', Journal of Economic Geography, vol. 18, no. 4, pp. 795-821.
142
Mirzabaev, A 2015, Impact of Agricultural Income Shocks due to Extreme Weather Events on
the Food Security of the Poor in Central Asia.
Mountain Research Initiative 2021, Addressing Climate Change, Poverty, and Flooding in
Malawi, Mountain Research Initiative, Bern,
<https://www.mountainresearchinitiative.org/news-content/africa/addressing-climate-change-
poverty-and-flooding-in-malawi>.
Newman, C & Tarp, F 2020, 'Shocks and agricultural investment decisions', Food Policy, vol.
94, p. 101810.
Ngoma, H, Mulenga, BP, Snyder, J, Banda, A & Chapoto, A 2019, Poverty and Weather
Shocks: A Panel Data Analysis of Structural and Stochastic Poverty in Zambia.
Ouattara, NB, Xueping, X, BI, TBAY, Traoré, L, Ahiakpa, J & Olounlade, OA 2019,
'Determinants of smallholder farmers’ access to microfinance credits: A case study in
Sassandra-Marahoué District, Côte d’Ivoire', Agricultural Finance Review, vol. 80, no. 3, pp.
401-419.
Page, S 2019, 'The Development Aid Situation in Malawi', in Development, Sexual Cultural
Practices and HIV/AIDS in Africa, Springer, pp. 43-60.
Papaioannou, KJ & de Haas, M 2017, 'Weather shocks and agricultural commercialization in

colonial tropical Africa: did cash crops alleviate social distress?', World Development, vol. 94,
pp. 346-365.
Pelka, N, Musshoff, O & Weber, R 2015, 'Does weather matter? How rainfall affects credit risk
in agricultural microfinance', Agricultural Finance Review, vol. 75, no. 2, pp. 194-212.
Peratsakis, C, Powell, J, Findley, M, Baker, J & Weaver, C 2012, 'Geocoded activity-level data
from the government of Malawi’s aid management platform', Washington, DC: AidData and
the Robert S. Strauss Center for International Security and Law, available at http://aiddata.
org/subnational-geospatial-research-datasets, vol.
Prakash, K, Churchill, SA & Smyth, R 2020, 'Petrol prices and subjective wellbeing', Energy
Economics, vol. 90, p. 104867.
Rocha, R & Soares, RR 2015, 'Water scarcity and birth outcomes in the Brazilian semiarid',
Journal of development Economics, vol. 112, pp. 72-91.
Salehyan, I & Gleditsch, KS 2006, 'Refugees and the spread of civil war', International
Organization, vol., pp. 335-366.
143
Seppanen, O, Fisk, WJ & Lei, Q 2006, Effect of temperature on task performance in office
environment, Lawrence Berkeley National Laboratory, Berkeley.
Shepherd, A, Wadugodapitiya, D & Evans, A 2011, 'Social assistance and the'dependency

syndrome'', Chronic Poverty Research Centre Policy Brief, vol. no. 22.
Taylor, L, Watkins, SL, Marshall, H, Dascombe, BJ & Foster, J 2016, 'The impact of different
environmental conditions on cognitive function: a focused review', Frontiers in Physiology, vol.
6, p. 372.
Tsaurai, K 2018, 'Complementarity Between Foreign Aid and Financial Development as a

Driver of Economic Growth in Selected Emerging Markets', Comparative Economic Research,
vol. 21, no. 4, pp. 45-61.
Ujiie, M, Tsuzuki, S & Ohmagari, N 2020, 'Effect of temperature on the infectivity of COVID-
19', International Journal of Infectious Diseases, vol. 95, pp. 301-303.
van Beek, G, de Vogel, V & van de Mheen, D 2021, 'The relationship between debt and crime:
A systematic and scoping review', European Journal of Probation, vol. 13, no. 1, pp. 41-71.
Villaverde, J & Maza, A 2009, 'The robustness of Okun's law in Spain, 1980–2004: Regional
evidence', Journal of Policy Modeling, vol. 31, no. 2, pp. 289-297.
Warnatzsch, EA & Reay, DS 2019, 'Temperature and precipitation change in Malawi:

Evaluation of CORDEX-Africa climate simulations for climate change impact assessments and
adaptation planning', Science of the Total Environment, vol. 654, pp. 378-392.
Weck-Hannemann, H & Schneider, F 2019, Determinants of foreign aid under alternative

institutional arrangements, Routledge, England.
Willmott, CJ & Matsuura, K 1995, 'Smart interpolation of annually averaged air temperature in
the United States', Journal of Applied Meteorology and Climatology, vol. 34, no. 12, pp. 2577-
2586.
Willmott, CJ & Robeson, SM 1995, 'Climatologically aided interpolation (CAI) of terrestrial air
temperature', International Journal of Climatology, vol. 15, no. 2, pp. 221-229.
Willmott, CJ, Rowe, CM & Philpot, WD 1985, 'Small-scale climate maps: A sensitivity analysis
of some common assumptions associated with grid-point interpolation and contouring', The
American Cartographer, vol. 12, no. 1, pp. 5-16.
World Bank 2020a, DataBank: World Development Indicators,

World Bank 2020b, World Bank Approves $86 Million to Improve Access to Financial Services
for Micro, Small and Medium Enterprises in Malawi, World Bank, Washington D.C,
<https://www.worldbank.org/en/news/press-release/2020/09/03/world-bank-approves-86-
million-to-improve-access-to-financial-services-for-micro-small-and-medium-enterprises-in-
malawi>.
144
World Bank 2021, Innovation and Digital Transformation in the Financial Sector in Malawi,
World Bank, Washington D.C, <https://www.worldbank.org/en/results/2021/01/14/innovation-
and-digital-transformation-in-the-financial-sector-in-malawi>.
Yang, Z, Wang, Q & Liu, P 2019, 'Extreme temperature and mortality: evidence from China',
International journal of biometeorology, vol. 63, no. 1, pp. 29-50.
Yue, P, Korkmaz, AG, Yin, Z & Zhou, H 2021, 'The rise of digital finance: Financial inclusion
or debt trap? ', Finance Research Letters, p. 102604.
145
Appendix 4.A
Variable Description Mean Std. Dev.

CREDIT BORROWING1 Dummy variable equal to 1 if in the past 12 months 0.222 0.416
the household borrowed on credit from banks,
microfinances or other sources, among others
CREDIT BORROWING2 Number of loans obtained in the past 12 months 0.245 0.497
CREDIT BORROWING3 Log of credit amount obtained in the last 12 months 2.034 3.873
CREDIT BORROWING4 Dummy variable equal to 1 if a household obtained a 0.222 0.416
loan in the past 12 months
CREDIT BORROWING5 Dummy variable equal to 1 if any member of the 0.081 0.272
household has an account at a bank, credit union,
microfinance institution, village savings organisation,
post office or another financial institution
CREDIT BORROWING6 Dummy variable equal to 1 if 12 months before the 0.077 0.267
survey interview, a member of the household used
an account at any financial institutions
Temperature District mean temperature 21.694 2.151
Total Aid Log of total foreign aid 4.321 7.624
Agricultural Aid Log of foreign aid allocated to the agricultural sector 1.304 3.880
Education Aid Log of foreign aid allocated to the education sector 1.680 4.823
Health Aid Log of foreign aid allocated to the health sector 2.467 5.887
Rural Development Aid Log of foreign aid allocated for rural development 1.642 4.868
Water, Sanitation & Irrigation Aid Log of foreign aid allocated for water, sanitation and 1.698 5.223
irrigation purposes
Employed Dummy variable equal to 1 if the household head was 0.092 0.290
employed at any time over the last 12 months.
Educated Dummy variable equal to 1 if the household head 0.867 0.339
ever attended school.
Household size Number of household members 4.943 2.297
Household income Log of household income 6.073 5.720
Rural Dummy variable equal to 1 if the household stays in 0.700 0.458
a rural area
Female Dummy variable equal to 1 if the household head is 0.220 0.414
a female
Age Age of the household head 42.466 15.276
Separated Dummy variable equal to 1 if the respondent reports 0.050 0.218
their marital status as separated
Divorced Dummy variable equal to 1 if the respondent reports 0.039 0.195
their marital status as divorced
Widowed Dummy variable equal to 1 if the respondent reports 0.100 0.300
their marital status as widowed
Single Dummy variable equal to 1 if the respondent reports 0.026 0.160
their marital status as single
146
Appendix 4.B
Table A4.2 – Full results with controls
(1) (2) (3)

Temperature (t-1) 0.034** 0.047*** 0.395***

(0.013) (0.015) (0.119)
Educated 0.097*** 0.102*** 0.873***
(0.035) (0.036) (0.311)
Employed 0.057 0.053 0.755**
(0.039) (0.041) (0.361)
Female 0.079 0.111** 0.736*
(0.048) (0.051) (0.443)
Age -0.011** -0.012*** -0.107***
(0.004) (0.005) (0.039)
Age squared 0.251*** 0.291*** 2.656***
(0.097) (0.105) (0.874)
Household income 0.004 0.006** 0.026
(0.002) (0.003) (0.022)
Rural -0.023 -0.072 -0.336
(0.060) (0.065) (0.590)
Separated -0.094 -0.085 -0.886
(0.064) (0.067) (0.585)
Divorced -0.060 -0.055 -0.610
(0.076) (0.086) (0.685)
Widowed -0.012 -0.004 -0.262
(0.058) (0.064) (0.560)
Single 0.060 0.060 0.845
(0.090) (0.091) (0.884)

R-squared 0.029 0.038 0.037
parentheses. *** p<0.01, ** p<0.05, * p<0.1
147
Chapter 5
Generosity or Exploitation? A Sub-National Analysis of Natural Resources and

Foreign Aid Allocation
5.1 Introduction
The discussion on the determinants of foreign aid allocation has a long tradition.
Several studies in the literature, mostly at the macro level, have examined factors
influencing aid allocated to developing countries, including donor’s interests in the
recipient country, recipient’s needs and recipient’s merit or features that are appealing
to donors (see, e.g., Alesina & Dollar 2000; Furuoka 2017; Kim & Oh 2012; Mahembe
& Odhiambo 2019; Rahman & Giessen 2017). At the same time, there is a growing
literature that has taken the discussion to the sub-national level, focusing on how
factors such as poverty, electoral motives, institutional quality, and governance
systems, among others, influence the donor’s decisions to allocate aid (see, e.g.,
Anaxagorou et al. 2020; Bayale 2020; Bei 2019; Dipendra 2020; Yang et al. 2018).
Despite the voluminous studies in this aid allocation literature, not much is known
about the impact of natural resources on aid allocation at the sub-national level –
specifically, at the district level. Thus, we shed light on this relationship by examining
how natural resources influence the allocation of aid in Africa. We focus on aid and
natural resources given that these are important sources of revenue in developing
countries (see, e.g., Munyanyi et al. 2020; OECD 2011; Venables 2016), and are
associated with various development-related outcomes such as poverty, economic
growth, conflict, gender equality and education, among others (see, e.g., Odokonyero
et al. 2018; Pickbourn & Ndikumana 2016; Riddell & Niño-Zarazúa 2016; Vesco et al.
2020; Woldekidan 2015; Zallé 2019).
We examine the impact of natural resources on foreign aid allocation at the district
level using annual data for a panel of three countries, namely Nigeria, Somalia, and
Uganda, for the period 2000 to 2014. We use Satellite data and the PRIO-GRID
dataset to extract data on natural resources, while foreign aid data are obtained from
the AidData website. We measure natural resources using the Normalized Difference
Vegetation Index (NDVI) (an indicator that captures the vegetation resource) and the
total number of mineral deposits in a district. We capture foreign aid using two dummy
148
variables. The first dummy variable takes the value of 1 if a district received aid (i.e.,
aid disbursements) in a particular year. The second dummy variable equals 1 if the
donor community made aid commitments to a particular district in a given year,
otherwise 0.
Applying a panel fixed effects model and controlling for relevant factors, we find that
natural resources positively affect foreign aid. These results are robust to several
sensitivity checks. Further, we explore channels through which natural resources may
influence aid allocation. Our findings indicate that economic growth and conflict are
important channels through which natural resources may transmit to the receipt of aid.
Given the availability of matching data on foreign aid and natural resources at the sub-
national level, we focus on Nigeria, Somalia, and Uganda. However, these countries
also make for important case studies for a number of reasons. First, these countries
are heavily dependent on foreign aid. For instance, Nigeria has been an aid recipient
country since the 1960s (USAID 2021a; World Bank 2020a). Due to its undermined
tax system and heightened corruptions levels that compromise its national revenues,
Nigeria has been receiving huge amounts of foreign aid to meet its development goals
(Amusa et al. 2020). Between 1980 and 2013, Nigeria received US$30.06 billion in
grants and US$7.62 billion in loans (Amusa et al. 2020) and US$2.7 billion in ODA for
the period 2012 to 2018 (Adebayo & Beton Kalmaz 2020). Like Nigeria, Somalia has
been an aid-dependent country since the 1960s, receiving large amounts of aid to
support its development projects (Muhumed 2020). According to Muhumed (2020), for
the periods between 2012 and 2016, and 2017 and 2018, the country received a yearly
average of US$1.3 billion and US$2 billion in foreign aid, respectively. In 2020,
Somalia again received US$2 billion in ODA (United Nations 2021a). Following the
civil unrest and economic mismanagement that took place between the 1970s and
1980s in Uganda, the country has been receiving substantial amounts of aid to revamp
and stabilise the economy and supplement the national budget (Hisali & Ddumba-
Ssentamu 2013). Notably, aid contributes a significant percentage to the national
budget. Evidence suggests that approximately 30% of the Ugandan budget is financed
by foreign aid (Findley, Harris, et al. 2017).
Second, these countries are highly endowed with natural resources. Nigeria’s natural
resources include oil, natural gas, iron ore, coal, gold, limestone, tin, arable land, and
149
vegetation, among others (see, e.g., Gutti et al. 2012; Ogunleye 2008; OPEC 2021).
About 10% of Nigeria’s GDP is attributed to the gas and oil sector (OPEC 2021). In
terms of vegetation, Nigeria has a well-diversified vegetation landscape, which
includes savanna lands, rain forests, woody forests, fresh-water swamp forests and
mangrove forests – making it a biodiversity-rich nation (Nwaogu et al. 2017). In the
same context, Uganda exhibits the following resources: agricultural land, vegetation
(e.g., woodlands, forests, and grasslands), and minerals such as iron ore, petroleum,
gold, limestone, phosphate, cobalt, and columbite-tantalite, among others (Abraham
et al. 2020; Abraham et al. 2021). Despite Somalia being regarded as one of the
poorest nations globally (see, e.g., Doboš 2016; Essa 2021; Florence et al. 2021) and
some of its natural resources still remain untapped (ITC 2021), the country is endowed
with resources such as natural gas, iron ore, uranium, coal, tin, and vegetation (e.g.,
pastures and forests), among others (ITC 2021; UNEP 2005).
We make significant contributions to the literature in at least two distinct ways. First,
we add to the literature on the impacts of natural resources (see, e.g., Kotsadam &
Tolonen 2016; López-Feldman et al. 2007; Sebri & Dachraoui 2021; Zallé 2019).
Although this literature has explored the impact of natural resources on outcomes such
as economic growth, conflict, quality of institutions, employment, income inequality,
and poverty, among others, our study extends the literature by examining how natural
resources influence foreign aid allocation.
Second, we contribute to the broad literature on the determinants of foreign aid

allocation (see, e.g., Anaxagorou et al. 2020; Furuoka 2017; Kim & Oh 2012;
Mahembe & Odhiambo 2019). This literature advances at least three reasons behind
donor’s aid allocation decisions, namely 1) ‘donor’s interests’ (e.g., to acquire new
markets, enjoy trade benefits, influence politics, create political alliances, and shape
good governance, paving way for investment from donor countries, among others)
(Rahman & Giessen 2017); 2) ‘recipient’s merit’ (e.g., attributes that can ensure aid
effectiveness such as good macroeconomic policies and high-quality institutions)
(Alesina & Dollar 2000); and 3) ‘recipient’s needs’ (e.g., economic growth,
employment, health, peace and improved living standards, among others). We make
additions to this literature by showing that natural resources are an important
determinant of aid allocation. Further, in this aid-allocation literature, our study is
closely related to studies that have sought to understand the allocation of aid at the
150
sub-national level (see, e.g., Anaxagorou et al. 2020; Bayale 2020; Bei 2019; Dipendra
2020; Marty et al. 2017; Yang et al. 2018). Many of these studies have investigated
factors that influence aid allocation at the regional level. Some of these factors include
electoral motives, poverty, health status, governance systems and institutional
arrangements, among others. However, we differ from these studies by examining aid
allocation at the district level and focusing on natural resources.
We organise the remainder of the chapter as follows: Section 5.2 discusses the
potential channels through which natural resources may influence foreign aid
allocation. Sections 5.3, 5.4 and 5.5 describe the data and variables incorporated,
explain the methodology applied, and interpret the results, respectively, while Section
5.6 concludes.
5.2 Why Should Natural Resources Influence Aid Allocation?
A number of reasons can be proposed for why we can expect natural resources to
influence foreign aid allocation. In this section, our discussions focus on two potential
mediators – economic growth and conflict – through which natural resources are likely
to influence the allocation of aid. Later in the chapter, we examine how these mediators
link natural resources to aid.
5.2.1 Economic growth
A number of studies have examined the impact of natural resources on economic

growth. Despite their efforts, the precise association between natural resources and
growth still remains unsettled (see, e.g., Brunnschweiler & Bulte 2008; Gerelmaa &
Kotani 2016; Sachs & Warner 1995; Sachs & Warner 2001; Zallé 2019). On the one
hand, a large proportion of studies in the literature argues that natural resources have
a negative effect on economic growth (see, e.g., Karl 1997; Mehlum et al. 2006; Sachs
& Warner 1995; Sala-i-Martin & Subramanian 2013) – an effect, which according to
the literature, is often referred to as the ‘curse of natural resources’ (see, e.g.,
Gerelmaa & Kotani 2016; Sachs & Warner 2001; Zallé 2019).
The literature has provided many hypotheses to explain this curse of natural
resources. For instance, evidence suggests that natural resources lead to a poor
economic performance by undermining the quality of institutions (see, e.g., Mehlum et
al. 2006; Sala-i-Martin & Subramanian 2013; Zallé 2019). In particular, according to
151
this evidence, natural resources undermine the development of quality institutions
conducive to economic growth by compromising the rule of law, hindering the
functioning of democratic institutions, and reducing civil liberties (Isham et al. 2005;
Jensen & Wantchekon 2004; Leite & Weidmann 1999; Sala-i-Martin & Subramanian
2013).
Some studies argue that natural resources promote rampant rent-seeking behaviour,
resulting in low economic growth levels. Notably, the rents generated by natural
resources tend to heighten the number of agents engaged in rent-seeking, as opposed
to pursuing productive activities, destroying the rents generated by natural resources
(Raggl 2017), weakening property rights, and lowering the attractiveness of productive
activities – undermining growth in the long term (Hodler 2006).
Other studies link natural resources to increased corruption (see, e.g., Badinger &
Nindl 2014; Bhattacharyya & Hodler 2010; Leite & Weidmann 1999). According to
these studies, in countries characterised by the abundance of natural resources and
comparably low polity-scores, natural resources are more likely to facilitate corruption,
which lowers economic growth (see, e.g., Anh et al. 2016; Gründler & Potrafke 2019).
In the same context, a huge strand of the literature regards this negative relationship
between natural resources and growth is due to so-called ‘Dutch disease’ (see, e.g.,
Gerelmaa & Kotani 2016; Sachs & Warner 1995; Sachs & Warner 2001). This
literature strand describes the Dutch disease as when highly profitable natural
resources sectors crowd out capital and labour from other economic sectors that could
produce positive externalities such as economic growth (Vesco et al. 2020), and
trigger inflation and appreciation of the exchange rate, harming export-based
manufactures and compromising economic growth (see, e.g., Mehlum et al. 2006;
Sachs & Warner 2001). Particularly, the abundance of natural resources tends to
crowd out several forms of capital, including human capital, physical capital, and social
capital, which leaves countries with low levels of human capital (see, e.g., Gylfason
2001; Gylfason & Zoega 2006). With inadequate human capital, countries face
complexities in getting rid of their dependence on primary products and experience
little diversification within their economies, transmitting to lower growth in the long run
(Zallé 2019).
152
Lower levels of economic growth tend to attract foreign aid. The aid literature suggests
that one of the many reasons behind aid allocation is the recipient’s needs, which
include the need to improve economic growth levels (see, e.g., Alesina & Dollar 2000;
Furuoka 2017; Mahembe & Odhiambo 2019). Put differently, donors are more willing
to disburse aid to countries with lower economic performance levels.
On the contrary, studies link natural resource abundance to improved economic

growth (see, e.g., Ben-Salha et al. 2018; Hinh & Russel 2016; Kwakwa et al. 2021;
Lyatuu et al. 2021). These studies contend that natural resources boost growth by 1)
generating economic rents that can be used for various development purposes in the
country, 2) increasing public spending and government revenues (e.g., from mineral
exports), and 3) raising the demand for locally produced goods and services. For a
more pronounced effect of natural resources on economic growth, evidence suggests
that countries should ensure democracy and political stability – important factors for
the efficient and effective utilisation of natural resources for long term growth (Kwakwa
et al. 2021). Assuming the donors’ criteria of allocating aid to countries with poorer
economic growth holds, we would expect high levels of economic growth to discourage
aid. In this regard, the impact of natural resources on foreign aid is ambiguous – it can
either be positive or negative and depends on how natural resources influence
economic growth. We hypothesise that economic growth mediates the relationship
between natural resources and foreign aid.
5.2.2 Conflict
Natural resources are associated with conflict (see, e.g., Kahl 2018; Mildner et al.
2011; Nillesen & Bulte 2014; Vesco et al. 2020). Evidence suggests that natural
resources tend to encourage profit-seeking strategies targeted at mobilizing and
fighting to seize the resources and take advantage of their rents (Vesco et al. 2020).
Particularly, when resources are abundant, people are more likely to have a higher
incentive to seize them to satisfy their needs (e.g., reduce poverty) and not only their
greed (Mildner et al. 2011; Vesco et al. 2020).
Regardless of the huge abundance of natural resources, some poor and marginalised
groups may be subject to constraints in the use or the access to such resources due
to dynamics of material transformations and power distribution (Kahl 2018). In such
situations, natural resources tend to kick-start distributional conflicts, as people
153
challenge the systems of rules that are associated with the governance of the
resources (see, e.g., Agarwal 2001; Ostrom 1990). In addition to unequal access to
resources, conflicts related to natural resources are more likely to arise due to the
failure of the state to ensure equitable distribution of the resources (see, e.g., Bullard
1993; Guha & Alier 2013; Martinez-Alier 2003). To rebels, for example, natural
resources may represent a prize worth fighting for and potential resources to finance
a rebellion or ongoing conflicts (see, e.g., Collier & Hoeffler 1998; Collier & Hoeffler
2004; Nillesen & Bulte 2014).
Conflict can both encourage and deter foreign aid. A strand of the existing literature
argues that conflict undermines aid flows by destroying important infrastructure and
exposing aid to corruption, which disturbs the donors’ interests (see, e.g., de Ree &
Nillesen 2009; Lis 2013). On the contrary, another strand of the literature regards
conflict as one of the many development challenges aid recipient countries face, which
the donor community has often responded to by giving more aid (Furuoka 2017). Put
differently, resolving conflict, and promoting peace in developing countries are some
of the major goals the donor community seeks to achieve (OECD 2022a). Against this
background, we argue that the relationship between natural resources and foreign aid
is mediated by conflict.
We collect data for a panel of three countries (i.e., Nigeria, Somalia, and Uganda) from
a number of sources. Data on foreign aid are sourced from AidData – an online
database run by the College of William & Mary.15 The database provides detailed and
freely available information on geo-referenced bilateral and multilateral foreign aid
projects from 96 donor organisations (Tierney et al. 2011).16 Additionally, it reports on
several development-related areas targeted by these aid projects, including
agriculture, education, water and sanitation, governance, and rural development,
among others. To compile the database, the College of William & Mary uses a wide
variety of sources such as the OECD Creditor Reporting System, projects records and
reports published by the donors, among others. Given that we are interested in
15Detailed information on the AidData database is found here: https://www.aiddata.org/datasets

16Aid projects have geo-codes that show their precise locations (e.g., GPS coordinates, district, state,
and province).
154
exploring the impacts of natural resources on aid allocation at the sub-national level,
we conduct our analysis at the district level.
We use Satellite data and the PRIO-GRID 2.0 dataset to extract information on natural
resources. Satellite data, also known as satellite imagery, typically refers to
information about planets in space, including Earth, gathered by man-made satellites
in their orbits.17 This type of data has been widely used in the literature (see, e.g.,
Amare et al. 2021; Awaworyi Churchill et al. 2021; Bujalský et al. 2018; Marbuah et al.
2021) mostly for exploring the Earth’s surface, weather changes and other incidents.
To generate this data, remote sensing technologies are used, which provide reliable
and timely data stream opportunities for a myriad of applications at land and sea. In
the case of our study, by using satellite data, we are able to locate and even visualise
areas endowed with various types of natural resources such as mineral deposits (e.g.,
iron ore and coal deposits) and vegetation (e.g., forests and savanna lands).
Our second data source for natural resources – the ‘PRIO-GRID dataset’, consists of
quadratic grid cells that jointly cover all terrestrial areas of the world. 18 Each grid cell
has a 0.5 × 0.5 decimal degree resolution and contains cell-specific information on
natural resources (e.g., mineral deposits, and various types of vegetation covering the
land), population, armed conflicts, climatic conditions, physical attributes, socio-
economic conditions, and more (Tollefsen et al. 2012). From this dataset, we also draw
on information on most of our control variables as well as other variables used to
perform our robustness checks.
Given that all our datasets show the names of the districts that have received foreign
aid and those that have not, and those endowed with natural resources and those
without overtime, we geographically match these datasets using district name and
year as our matching variables. We extract data for the period 2000 to 2014 from the
data sources to conduct our analyses.
5.3.1 Foreign aid
Consistent with the literature (see, e.g., Anaxagorou et al. 2020; Bezerra & Braithwaite
2016; Hicks et al. 2016; Marineau & Findley 2020), we capture foreign aid using foreign
17 More on satellite data here: https://www.iceye.com/satellite-data

18 PRIO-GRID here: https://grid.prio.org/#/
155
aid commitments and disbursements. The AidData database reports information on
aid commitments and disbursements made by the donor community annually. Aid
commitments refer to firm obligations made by donors, usually expressed in writing
and backed by the necessary financial resources to provide assistance to recipient
countries (OECD 2022b). On the other hand, aid disbursements are actual transfers
of financial resources by donors to the recipient countries. Despite aid commitments
being the commonly used indicator in aid allocation studies, to shed more light on the
exact amounts of aid actually allocated, we also consider aid disbursements. Evidence
shows that in the poorest and most aid dependent countries, disbursed aid volumes
tend to differ widely from aid commitments (Hudson, 2013).
We use dummy variables to capture foreign aid allocated at the district level for our
main analysis. The first measure is a dummy variable equal to 1 if foreign aid was
committed to a particular district in a given year. Similarly, the second indicator is a
dummy variable that equals 1 if aid was disbursed to a particular district in a given
year. For additional checks, we use foreign aid intensity (i.e., aid in US$ values) and
consider different types of foreign aid (e.g., bilateral and multilateral aid).
5.3.2 Natural resources
We capture natural resources using two indicators. Our first indicator considers
vegetation a natural resource and captures the resource using the NDVI derived from
satellite data. The NDVI is a measure of vegetation density and condition – designed
to capture live green vegetation on the ground globally (Goldblatt et al. 2020) and is
defined as:
𝑁𝐼𝑅 − 𝑅𝑒𝑑 (1)

𝑁𝐷𝑉𝐼 =
𝑁𝐼𝑅 + 𝑅𝑒𝑑
where, 𝑁𝐼𝑅 and 𝑅𝑒𝑑 represent the amounts of near-infrared and red light, respectively,
reflected by the vegetation and captured by the sensor of the satellite. The NDVI value
ranges between -1 and +1, with negative values representing the absence of
vegetation (Pettorelli et al. 2005). Several studies in the literature have broadly used
this index (see, e.g., Ayana et al. 2016; Brown 2010; Bujalský et al. 2018) given that it
is easy to calculate and can be replicated with other satellite products that capture
spectral light (Goldblatt et al. 2020). As further checks, we capture the vegetation
156
resource using various vegetation types, including ‘Savanna’ and ‘Forest’ lands. We
obtain this information from the PRIO-GRID dataset.
Our second indicator captures natural resources using the total number of mineral
deposits found in a district. These mineral deposits range from iron ore to coal
deposits. Annually, as indicated in the dataset, new mineral deposits have been
discovered, making it possible to examine how variations in the total number of mineral
deposits influence a district’s likelihood of receiving foreign aid.
5.3.3 Other variables
We control for a number of factors that may influence foreign aid allocation. For our
main analysis, we control for district population and economic activity. District
population is measured by the total number of people living in the given district, while
district economic activity is captured using satellite night-time light data, which has
increasingly become the standard of capturing economic growth at the sub-national
level (see, e.g., Ackermann et al. 2021; Appau et al. 2021; Awaworyi Churchill et al.
2021; Hodler & Raschky 2014).
For our transmission channel analysis and additional robustness checks, we also
control for other factors that are associated with foreign aid allocation, including
conflict, distance to the capital city (in km), district area (in square kilometres), and
climatic conditions (for a review of these factors see, Anaxagorou et al. 2020). By
controlling for district area, we account for the effect of geographical size. At the same
time, controlling for the distance to the capital city allows us to capture the role played
by the country’s capital city in aid allocation. Table A5.1 in the appendix provides a
description and summary statistics of variables incorporated.
5.4 Methodology
To investigate the impact of natural resources endowment on aid allocation, we

estimate the following equation:
𝐹𝐴𝑖𝑐𝑡 = 𝜑𝑁𝑅𝑖𝑐𝑡 + 𝜗𝑿′𝑖𝑐 + 𝜔𝑖𝑐𝑡 + 𝜇𝑖𝑐𝑡 (2)
where 𝐹𝐴𝑖𝑐𝑡 is foreign aid committed or disbursed to district 𝑖 in country 𝑐 at year 𝑡; 𝑁𝑅

represents the natural resources indicator (e.g., the NDVI and total number of mineral
157
deposits); 𝑋 is a vector of control variables (e.g., district economic activity and
population); and 𝜇𝑖𝑐𝑡 is an error term. The key parameter of interest is 𝜑, which
captures the effect of natural resources on the allocation of foreign aid.
We estimate equation (2) using a panel fixed effects model, with 𝜔𝑐𝑡 denoting country-
district-year fixed effects. To facilitate the interpretation of our results, we estimate the
equation as a linear probability model. Including country-district-year fixed effects
enables us to control for observed and unobserved country-and district-specific time-
varying factors that could be correlated with foreign aid (Anaxagorou et al. 2020).
5.5 Empirical Results
5.5.1 Main results
We begin by reporting results on the relationship between natural resources and

foreign aid. The results are presented in Table 5.1. Panels A and B report the results,
where we capture natural resources using the NDVI and total number of mineral
deposits, respectively. Foreign aid proxies are dummy variables equal to 1 if aid was
disbursed or committed to a given district in a particular year. Columns 1 and 2 show
the bivariate results, while Columns 3 and 4 report the full regression results – where
we control for covariates such as population and economic activity of the district.
158
Table 5.1 – Main Results
Bivariate results Multivariate results
(1) (2) (3) (4)
VARIABLES Total Aid Total Aid Total Aid Total Aid
Commitments Disbursements Commitments Disbursements
Panel A: Normalized Difference Vegetation Index
NDVI 0.010*** 0.009*** 0.005** 0.004*
(0.002) (0.002) (0.002) (0.002)
Population 0.042*** 0.042***
(0.004) (0.004)
Night-time lights 0.002** 0.001
(0.001) (0.001)
Observations 5,475 5,475 5,475 5,475

R-squared 0.120 0.113 0.166 0.162
Panel B: Total mineral deposits
Minerals 0.099** 0.097*** 0.070* 0.068*
(0.038) (0.038) (0.039) (0.038)
Population 0.042*** 0.042***
(0.004) (0.004)
Night-time lights 0.002** 0.001
(0.001) (0.001)
Observations 5,475 5,475 5,475 5,475

R-squared 0.119 0.113 0.166 0.162
Notes: All regressions include country, district and year fixed effects.
*** p<0.01, ** p<0.05, * p<0.1
Across all panels and columns, the results indicate a positive and statistically
significant relationship between natural resources and foreign aid. Specifically, the
bivariate results presented in Columns 1 and 2 of Panel A show that having natural
resources increases a district’s probability of receiving foreign aid by 1 and 0.9
percentage points, respectively. In Panel B – Columns 1 and 2, where we measure
natural resources using ‘total number of mineral deposits’, the bivariate results indicate
that having natural resources is associated with 9.9 and 9.7 percentage points
increase, respectively, in the district’s likelihood of receiving aid.
The multivariate results, shown in Columns 3 and 4, support the bivariate regression
results. From Panel A, we find that natural resources, as measured using NDVI,
increase the district’s probability of receiving foreign aid by 0.5 and 0.4 percentage
points, respectively. In Panel B, the results show that natural resources as captured
by the total number of mineral deposits in a given district, increases the district’s
likelihood of receiving aid by 7 and 6.8 percentage points, respectively.
159
The results for the control variables (i.e., population and economic activity)
incorporated into our main regression model are generally consistent with
expectations and the literature. Notably, we find that both population and economic
activity positively affect aid. Evidence suggests that high populous areas and those
characterised by increasing economic growth rates tend to be more aid eligible than
less populous areas and with poor economic growth rates (see, e.g., Briggs 2017;
Briggs 2021; Neumayer 2003; Watanabe 2006). An interesting observation here,
across all panels and columns, is that the effect (as measured by the coefficient
magnitude) of natural resources on aid is slightly lower on total aid disbursements than
on total aid commitments. This may be due to the fact that aid disbursements may
actually take time (e.g., months to years) to happen (Bearce & Tirone 2010).
5.5.2 Robustness checks
In this section, we conduct several checks to examine the robustness of our main
regression results. First, we examine the robustness of our results to the use of
alternative measures of foreign aid. In our main results, we capture foreign aid using
dummy variables. Given that natural resources may have an effect on the aid amount
committed and disbursed by donors, which we may not see when using dummy
variables, in Table 5.2, we unmask the nuanced effect of natural resources on total
foreign aid amounts (also known as foreign aid intensity).
Table 5.2 – Aid intensity and natural resources

(1) (2)
NDVI 0.096** 0.088***
(0.037) (0.034)

R-squared 0.165 0.159
Minerals 1.069* 1.009*
(0.606) (0.567)

R-squared 0.165 0.159
Notes: All regressions include controls as in Table 5.1.
*** p<0.01, ** p<0.05, * p<0.1
160
Columns 1 and 2 show foreign aid amounts in commitments and disbursements,
respectively. Across all panels and columns, we find that the effect of natural
resources endowment on aid remains positive and significant.
Second, we check the sensitivity of our results to the use of alternative indicators of
natural resources. Our main results capture the ‘vegetation resource’ using the NDVI.
In exploring the sensitivity of our results, we use other indicators that capture the
percentage of ‘type of vegetation covering the land’ available in a district. We use
‘Forest land’ as our first indicator – an indicator, which indicates the percentage of
area covered by forests. Our second indicator is ‘Savanna land’, which shows the
proportion of area covered by savanna lands. Forest-and savanna lands provide
water, grazing and browsing, food and fuel for millions of people, and have a unique
biodiversity that supports wildlife tourism. Table 5.3 reports the results, indicating that
forest and savanna lands are positively associated with foreign aid. Thus, our results
remain robust.
Table 5.3 – Aid and alternative indicators of vegetation

(1) (2)
Panel A: Forest lands
Forest lands 0.018*** 0.016***
(0.006) (0.006)

R-squared 0.167 0.163
Panel B: Savanna lands
Savanna lands 0.009*** 0.005**
(0.002) (0.002)

R-squared 0.168 0.162
*** p<0.01, ** p<0.05, * p<0.1
Third, we verify the robustness of our results to the use of alternative measures of
minerals. Our main results show the impact of the ‘total number of mineral deposits’
on aid. In Table 5.4, we examine the effect of the type of minerals on aid. We consider
total iron ore deposits and total coal deposits – the two main minerals found across
our sample countries. Panel A and B of Table 5.4 present the results, which show a
positive relationship between these types of minerals and aid. Based on these
161
findings, we contend that our results remain robust.
Table 5.4 – Aid and types of minerals

(1) (2)
Panel A: Total iron ore deposits
Iron ore 0.067*** 0.053***
(0.007) (0.006)

R-squared 0.165 0.161
Panel B: Total coal deposits
Coal 0.075* 0.073*
(0.043) (0.042)

R-squared 0.166 0.162
*** p<0.01, ** p<0.05, * p<0.1
Fourth, we examine the sensitivity of our findings to foreign aid types. In our main
regression results, our indicators of foreign aid are total aid commitments and
disbursements. In Table 5.5, we capture foreign aid using bilateral and multilateral aid.
Bilateral aid generally refers to assistance given by a government of one country
directly to a government of another country, while multilateral aid represents
assistance given through international institutions (e.g., the World Bank, United
Nations, International Monetary Fund) rather than by one specific country.
Table 5.5 – Types of aid and natural resources
(1) (2)
VARIABLES Bilateral Aid Multilateral Aid
NDVI -0.003 0.006***
(0.002) (0.002)

R-squared 0.161 0.075
Minerals 0.048 0.048*
(0.037) (0.025)

R-squared 0.161 0.074
162
As reported in Table 5.5, our results show that natural resources significantly attract
multilateral aid but not bilateral aid. A potential explanation here may be that
multilateral aid is less politicized, highly selective and often targets areas with a higher
probability of bettering lives in aid recipient countries (Gulrajani 2016). In our case,
multilateral aid donors may decide to target areas endowed with natural resources for
several reasons, for example, to reduce natural resource conflicts, improve natural
resources management and establish the relevant infrastructure that ensures efficient
and sustainable use of natural resources – revamping the standard of living of the
locals in the process. Over the years, in Nigeria (i.e., one of the sample countries), for
instance, there have been severe conflicts over the allocation of grazing lands and
management of various land-related resources (see, e.g., Egbuta 2018; Muhammed
et al. 2015; Obioha 2008). As a result, huge amounts of multilateral aid have been
disbursed to Nigeria primarily to eliminate conflict and promote peace (OECD 2022a).
In such as case, from the multilateral aid donor’s perspective, it would make sense to
allocate aid to such conflict-inflicted areas to help promote peace.
On the contrary, bilateral aid is more politicised than multilateral aid (see, e.g., Ahsan
2005; Findley, Milner, et al. 2017; Gulrajani 2016). That is, if an area that has potential
to receive aid, proves not to satisfy the interests of the bilateral donors, that area may
not receive at all. In our case, the bilateral donors may not be interested in allocating
aid to areas endowed with natural resources, given that such areas already have many
potential sources of income; therefore, no need for additional funding. Alternatively,
maybe due to conflicts associated with natural resources (see, e.g., Mildner et al.
2011; Vesco et al. 2020), bilateral donors may decide not to disburse aid to such areas.
In the end, bilateral donors may prefer to allocate aid to areas that have nothing to do
with natural resources but that possess other attributes that tally with their interests.
Fifth, we investigate the sensitivity of our results by controlling for the distance to the
capital city. One of the many factors that have been advanced in the literature as
important determinants of aid allocation is the distance to the capital city (Blair et al.
2021). Here, we expect areas (in our case, districts) close to the capital city to have a
higher probability of receiving aid than those far away. The findings are presented in
Table 5.6, which show that the effect of natural resources on aid remains positive and
significant.
163
Table 5.6 – Aid and distance to the capital city
(1) (2)
NDVI 0.005** 0.004*
(0.002) (0.002)

R-squared 0.167 0.162
Minerals 0.069* 0.068*
(0.039) (0.039)

R-squared 0.167 0.163
*** p<0.01, ** p<0.05, * p<0.1
The sizes of the coefficients on natural resources are almost the same as our main
results, implying that the distance to the capital city does not have much effect on how
natural resources influence aid allocation in our sampled countries. Sixth, in Table 5.7,
we examine the robustness of our results by controlling for district area. Evidence
shows that the location’s area significantly affects aid allocation – with larger areas
having a higher likelihood of receiving aid (Anaxagorou et al. 2020). After conducting
this exercise, we find that our results remain robust.
Table 5.7 – Aid and district area

(1) (2)
NDVI 0.005** 0.004
(0.002) (0.002)

R-squared 0.166 0.162
Minerals 0.069* 0.068*
(0.039) (0.039)

R-squared 0.166 0.162
*** p<0.01, ** p<0.05, * p<0.1
As our final robustness check, we control for the climate. Specifically, we control for
164
district average temperature and rainfall. The results, presented in Table 5.8, show
that our results are robust.
Table 5.8 – Aid and climate

(1) (2)
NDVI 0.077* 0.069*
(0.039) (0.035)

R-squared 0.208 0.201
Minerals 0.072** 0.070**
(0.035) (0.035)

R-squared 0.208 0.205
*** p<0.01, ** p<0.05, * p<0.1
We also observe an increase in the effect of natural resources on aid (as shown by
the magnitude of coefficients) when we control for climate variables. This shows that
these climate variables cause an upward spiral on the impact of natural resources.
5.5.3 Channel analysis
Next, we present the potential channel analysis, explaining channels through which
natural resources may influence aid allocation. We explored economic growth/activity
and conflict as some of these channels in Section 2. To carry out the potential channel
analysis, we employ a two-step process that is widely used in the literature (see, e.g.,
Awaworyi Churchill et al. 2019; Prakash et al. 2020). As a first step, for the potential
channels to be qualified as transmission channels, they need to be correlated with the
indicators of natural resources. We present the results in Table 5.9, which shows that
both economic growth and conflict are correlated with natural resources. Specifically,
in Panel B – Column 1, we observe a negative and statistically significant nexus
between mineral deposits and economic activity.
165
Table 5.9 – Correlation between natural resources and mediators
(1) (2)
VARIABLES Night-time light Conflict

NDVI 0.040 0.057*
(0.084) (0.030)

R-squared 0.380 0.037
Minerals -0.535*** 0.089**
(0.131) (0.039)

R-squared 0.380 0.036
Notes: Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
This finding lends support to studies that link the availability of minerals with lower
levels of economic growth and development (see, e.g., Davis 1995; Mulwa 2017;
Mulwa & Mariara 2016). On the contrary, in Column 2 of both Panels A and B, we find
that natural resources have a positive and statistically significant impact on conflict.
This finding is consistent with the literature that associates natural resources
endowment with increased conflict (see, e.g., Le Billon 2013; Lessmann & Steinkraus
2019). According to Lessmann and Steinkraus (2019), for example, the
heterogeneous distribution of natural resources increases the likelihood of civil wars.
For the second step, the addition of the potential channels into the natural resources
– aid equation should reduce the size of the coefficient on the natural resources
indicators or make the impact of these indicators on foreign aid statistically
insignificant. We report the results for this exercise in Table 5.10.
166
Table 5.10 – Channel analysis
(1) (2) (3)

VARIABLES Foreign Aid Foreign Aid Foreign Aid
NDVI 0.009*** 0.004* 0.003
(0.002) (0.002) (0.002)
Night-time light 0.001
(0.001)
Conflict 0.007
(0.005)

R-squared 0.113 0.162 0.162
Minerals 0.097*** 0.068* 0.067*
(0.038) (0.038) (0.039)
Night-time light 0.001
(0.001)
Conflict 0.007
(0.005)

R-squared 0.113 0.162 0.163
Notes: We capture foreign aid using total aid disbursements.
*** p<0.01, ** p<0.05, * p<0.1
Column 1 shows the results before adding the potential channels. Column 2 reports
the results after adding the economic growth indicator, while Column 3 shows results
after incorporating the conflict variable. As shown in the table in Columns 2 and 3, as
we incorporate these potential channels, the magnitude of the coefficient on the
natural resources indicators either decreases or become statistically insignificant.
Based on the results, we argue that conflict and economic activity are channels
through which natural resources transmit to aid.
5.6 Conclusion
A large proportion of the literature has investigated the determinants of foreign aid
allocation at both national and sub-national levels. Despite this effort, little is known
about the effect of natural resources on the allocation of aid. We examine the effect of
natural resources endowment on foreign aid allocation using a panel of three
countries, namely Somalia, Nigeria, and Uganda, for the period covering 2000 to 2014.
We draw on foreign aid data from the AidData website and extract data on natural
167
resources from the Satellite and the PRIO-GRID dataset. We find that natural
resources have a positive and statistically significant impact on aid allocation. Our
results are robust to several sensitivity checks, including 1) the use of alternative
measures of natural resources and foreign aid, 2) types of mineral deposits found in
districts, 3) types of foreign aid and 4) controlling for conflict, distance to the capital
city, district area and climate variables (e.g., rainfall and temperature). Moreover, we
explore channels through which natural resources may transmit to foreign aid. We find
that conflict and economic growth are the channels through which natural resources
may influence aid allocation.
The policy implications of our findings are straightforward. Given that our results
indicate that natural resources positively affect aid allocation, aid recipient
governments should ensure efficient and effective utilisation of natural resources to
attract aid. Policies that discourage corruptive conduct and any form of rent-seeking
behaviour, which may undermine the efficient use of natural resources, should be
established.
Future studies should explore how factors such as institutional quality mediate the
relationship between natural resources and aid allocation at the sub-national level
when data on these factors become available.
168
References
Abraham, MJ, Gift, R, Hennery, S, Asuman, G, Rita, A, Dexter, D & Joseph, M 2020, 'The
Prospects of Uganda’s Iron Ore Deposits in Developing the Iron and Steel Industry', Journal
of Minerals and Materials Characterization and Engineering, vol. 8, no. 04, p. 316.
Abraham, MJ, Ronald, K, Aaron, W, Rita, A, Jonan, K, Asuman, G & Joseph, M 2021,
'Evaluating the Energy Requirements for Uganda: Case for Natural Gas', International Journal
of Energy and Environmental Science, vol. 6, no. 4, p. 68.
Adebayo, TS & Beton Kalmaz, D 2020, 'Ongoing debate between foreign aid and economic
growth in Nigeria: a wavelet analysis', Social Science Quarterly, vol. 101, no. 5, pp. 2032-
2051.
Agarwal, B 2001, 'Participatory exclusions, community forestry, and gender: An analysis for
South Asia and a conceptual framework', World Development, vol. 29, no. 10, pp. 1623-1648.
Ahsan, M 2005, 'Politicization of bilateral aid and educational development in Pakistan',

Educational Studies, vol. 31, no. 3, pp. 235-250.
Growth, vol. 5, no. 1, pp. 33-63.
Amare, M, Abay, KA, Arndt, C & Shiferaw, B 2021, 'Youth Migration Decisions in Sub‐Saharan
Africa: Satellite‐Based Empirical Evidence from Nigeria', Population and Development
Review, vol. 47, no. 1, pp. 151-179.
Amusa, K, Monkam, N & Viegi, N 2020, 'Can foreign aid enhance domestic resource
mobilisation in Nigeria?', Journal of Contemporary African Studies, vol. 38, no. 2, pp. 294-309.
103430.
Anh, NN, Minh, NN & Tran-Nam, B 2016, 'Corruption and economic growth, with a focus on
Vietnam', Crime, Law and Social Change, vol. 65, no. 4, pp. 307-324.
pp. 506-518.
Ayana, EK, Ceccato, P, Fisher, JR & DeFries, R 2016, 'Examining the relationship between
environmental factors and conflict in pastoralist areas of East Africa', Science of the Total
Environment, vol. 557, pp. 601-611.
169
Badinger, H & Nindl, E 2014, 'Globalisation and corruption, revisited', The World Economy,
vol. 37, no. 10, pp. 1424-1440.
Bayale, N 2020, 'Empirical investigation into the determinants of foreign Aid in Sahel countries:
A panel bayesian model averaging approach', Defence and Peace Economics, vol., pp. 1-21.
Bei, LJ 2019, 'Where does the dragon’s gift go?: Subnational distribution of China’s aid to Sub-
Saharan Africa from 2007 to 2012', LSE Undergraduate Political Review, vol., pp. 27-59.
Ben-Salha, O, Dachraoui, H & Sebri, M 2018, 'Natural resource rents and economic growth in
the top resource-abundant countries: a PMG estimation', Resources Policy, vol., p. 101229.
Bezerra, P & Braithwaite, A 2016, 'Locating foreign aid commitments in response to political
violence', Public Choice, vol. 169, no. 3, pp. 333-355.
Bhattacharyya, S & Hodler, R 2010, 'Natural resources, democracy and corruption', European
Economic Review, vol. 54, no. 4, pp. 608-621.
Blair, RA, Marty, R & Roessler, P 2021, 'Foreign Aid and Soft Power: Great Power Competition
in Africa in the Early Twenty-first Century', British Journal of Political Science, vol., pp. 1-22.
Briggs, RC 2017, 'Does foreign aid target the poorest?', International Organization, vol. 71,
no. 1, pp. 187-206.
Briggs, RC 2021, 'Why Does Aid Not Target the Poorest?', International Studies Quarterly,
vol. 65, no. 3, pp. 739-752.
Brown, IA 2010, 'Assessing eco-scarcity as a cause of the outbreak of conflict in Darfur: a

remote sensing approach', International Journal of Remote Sensing, vol. 31, no. 10, pp. 2513-
2520.
Brunnschweiler, CN & Bulte, EH 2008, 'The resource curse revisited and revised: A tale of
paradoxes and red herrings', Journal of Environmental Economics and Management, vol. 55,
no. 3, pp. 248-264.
Bujalský, L, Jirka, V, Zemek, F & Frouz, J 2018, 'Relationships between the normalised
difference vegetation index and temperature fluctuations in post-mining sites', International
Journal of Mining, Reclamation and Environment, vol. 32, no. 4, pp. 254-263.
Bullard, RD 1993, Confronting environmental racism: Voices from the grassroots, South End
Press, Boston.
Collier, P & Hoeffler, A 1998, 'On economic causes of civil war', Oxford Economic Papers, vol.
50, no. 4, pp. 563-573.
Collier, P & Hoeffler, A 2004, 'Greed and grievance in civil war', Oxford Economic Papers, vol.
56, no. 4, pp. 563-595.
Davis, GA 1995, 'Learning to love the Dutch disease: Evidence from the mineral economies',
World Development, vol. 23, no. 10, pp. 1765-1779.
170
301-313.
Dipendra, K 2020, 'Which aid targets poor at the sub-national level?', World Development
Perspectives, vol. 17, p. 100177.
Doboš, B 2016, 'Shapeshifter of Somalia: Evolution of the political territoriality of Al-Shabaab',

Small Wars & Insurgencies, vol. 27, no. 5, pp. 937-957.
Egbuta, U 2018, 'Understanding the herder-farmer conflict in Nigeria', Conflict Trends, vol.
2018, no. 3, pp. 40-48.
Essa, FO 2021, 'Post-War and Development of a Historical Curriculum in Somalia', Scholars

Journal of Arts, Humanities and Social Sciences, vol. 10, pp. 489-501.
Findley, MG, Harris, AS, Milner, HV & Nielson, DL 2017, 'Who controls foreign aid? Elite
versus public perceptions of donor influence in aid-dependent Uganda', International
Organization, vol. 71, no. 4, pp. 633-663.
Findley, MG, Milner, HV & Nielson, DL 2017, 'The choice among aid donors: The effects of
multilateral vs. bilateral aid on recipient behavioral support', Review of International
Organizations, vol. 12, no. 2, pp. 307-334.
Florence, K, Mohamed, F, Rithaa Gilbert, MH, Leila, A & Clementina, N 2021, 'Linking Social
Protection and Nutrition Outcomes: Best Practices and Opportunities in Somalia', Journal of
Food and Nutrition Research, vol. 9, no. 8, pp. 430-433.
Gerelmaa, L & Kotani, K 2016, 'Further investigation of natural resources and economic
growth: do natural resources depress economic growth?', Resources Policy, vol. 50, pp. 312-
321.
Goldblatt, R, Heilmann, K & Vaizman, Y 2020, 'Can medium-resolution satellite imagery

measure economic activity at small geographies? Evidence from Landsat in Vietnam', The
World Bank Economic Review, vol. 34, no. 3, pp. 635-653.
Gründler, K & Potrafke, N 2019, 'Corruption and economic growth: New empirical evidence',
European Journal of Political Economy, vol. 60, p. 101810.
Guha, R & Alier, JM 2013, Varieties of environmentalism: Essays North and South, Routledge,
England.
Gulrajani, N 2016, Bilateral versus multilateral aid channels, Overseas Development Institute,
London.
Gutti, B, Aji, MM & Magaji, G 2012, 'Environmental impact of natural resources exploitation in
Nigeria and the way forward', Journal of Applied technology in Environmental sanitation, vol.
2, no. 2, pp. 95-102.
Gylfason, T 2001, 'Nature, power and growth', Scottish Journal of Political Economy, vol. 48,
no. 5, pp. 558-588.
171
Gylfason, T & Zoega, G 2006, 'Natural resources and economic growth: The role of
investment', World Economy, vol. 29, no. 8, pp. 1091-1115.
Hicks, DL, Hicks, JH & Maldonado, B 2016, 'Women as policy makers and donors: Female
legislators and foreign aid', European Journal of Political Economy, vol. 41, pp. 46-60.
Hinh, D & Russel, D 2016, Managing natural resources for growth and prosperity in low income
countries, OCP Policy Center, Rabat, Morocco.
Hisali, E & Ddumba-Ssentamu, J 2013, 'Foreign aid and tax revenue in Uganda', Economic
Modelling, vol. 30, pp. 356-365.
Hodler, R 2006, 'The curse of natural resources in fractionalized countries', European

Economic Review, vol. 50, no. 6, pp. 1367-1386.
vol. 129, no. 2, pp. 995-1033.
Hudson, J 2013, 'Promises kept, promises broken? The relationship between aid
commitments and disbursements', Review of Development Finance, vol. 3, no. 3, pp. 109-
120.
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', The World
Bank Economic Review, vol. 19, no. 2, pp. 141-174.
ITC 2021, Somalia: Country brief, International Trade Centre (ITC), Geneva,
<https://www.intracen.org/country/somalia/>.
Jensen, N & Wantchekon, L 2004, 'Resource wealth and political regimes in Africa',
Comparative Political Studies, vol. 37, no. 7, pp. 816-841.
Kahl, CH 2018, States, scarcity, and civil strife in the developing world, Princeton University
Press, New Jersey.
Karl, TL 1997, The paradox of plenty, University of California Press, Berkeley.
Kim, EM & Oh, J 2012, 'Determinants of foreign aid: The case of South Korea', Journal of East
Asian Studies, vol. 12, no. 2, pp. 251-274.
Kotsadam, A & Tolonen, A 2016, 'African mining, gender, and local employment', World
Kwakwa, PA, Adzawla, W, Alhassan, H & Achaamah, A 2021, 'Natural resources and
economic growth: Does political regime matter for Tunisia?', Journal of Public Affairs, vol., p.
e2707.
Le Billon, P 2013, Fuelling war: Natural resources and armed conflicts, Routledge, England.
Leite, MC & Weidmann, J 1999, Does mother nature corrupt? Natural resources, corruption,
and economic growth, International Monetary Fund, Washington DC.
Lessmann, C & Steinkraus, A 2019, 'The geography of natural resources, ethnic inequality
and civil conflicts', European journal of political economy, vol. 59, pp. 33-51.
172
López-Feldman, A, Mora, J & Taylor, JE 2007, 'Does natural resource extraction mitigate
poverty and inequality? Evidence from rural Mexico and a Lacandona Rainforest Community',
Environment and Development Economics, vol. 12, no. 2, pp. 251-269.
Lyatuu, I, Loss, G, Farnham, A, Winkler, MS & Fink, G 2021, 'Short-term effects of national-
level natural resource rents on life expectancy: A cross-country panel data analysis', PloS
One, vol. 16, no. 5, p. e0252336.
1, p. 1626321.
Marbuah, G, Gren, M, Mckie, BG & Buisson, L 2021, 'Economic activity and distribution of an
invasive species: Evidence from night-time lights satellite imagery data', Ecological
Marineau, JF & Findley, MG 2020, '‘The Swarm Principle’: A Sub-National Spatial Analysis of
Aid Targeting and Donor Coordination in Sub-Saharan Africa', Stability: International Journal
of Security and Development, vol. 9, no. 1.
Martinez-Alier, J 2003, The Environmentalism of the poor: a study of ecological conflicts and
valuation, Edward Elgar Publishing, Massachusetts.
vol. 2, no. 1.
Mehlum, H, Moene, K & Torvik, R 2006, 'Cursed by resources or institutions?', World

Economy, vol. 29, no. 8, pp. 1117-1131.
Mildner, S-A, Lauster, G & Wodni, W 2011, 'Scarcity and abundance revisited: A literature
review on natural resources and conflict', International Journal of Conflict and Violence (IJCV),
vol. 5, no. 1, pp. 155-172.
Muhammed, I, Ismaila, AB & Bibi, UM 2015, 'An assessment of farmer-pastoralist conflict in

Nigeria using GIS', International Journal of Engineering Science Invention, vol. 4, no. 7, pp.
23-33.
Muhumed, MM 2020, 'Somalia’s Aid Dependency in 1960-1990', Somali Studies: A Peer

Reviewed Academic Journal for Somali Studies, vol. 5, pp. 11-29.
Mulwa, R 2017, 'Natural resource curse and its causation channels in Africa', African Journal
of Economic and Sustainable Development, vol. 6, no. 4, pp. 244-261.
Mulwa, R & Mariara, J 2016, Natural resource curse in Africa: Dutch disease and institutional
explanations, International Food Policy Research Institute, Washington DC.
173
Neumayer, E 2003, 'What factors determine the allocation of aid by Arab countries and
multilateral agencies?', Journal of Development Studies, vol. 39, no. 4, pp. 134-147.
Nillesen, E & Bulte, E 2014, 'Natural resources and violent conflict', Annual Review of
Resource Economics, vol. 6, no. 1, pp. 69-83.
Nwaogu, C, Okeke, OJ, Fadipe, OO, Bashiru, KA & Pechanec, V 2017, 'Is Nigeria losing its
natural vegetation and landscape? Assessing the landuse-landcover change trajectories and
effects in Onitsha using remote sensing and GIS', Open Geosciences, vol. 9, no. 1, pp. 707-
718.
Obioha, EE 2008, 'Climate change, population drift and violent conflict over land resources in
northeastern Nigeria', Journal of Human Ecology, vol. 23, no. 4, pp. 311-324.
OECD 2011, The economic significance of natural resources: Key points for reformers in
Eastern Europe, Caucasus and Central Asia, OECD, Paris.
OECD 2022a, OECD.Stat: Creditor Reporting System (CRS), Organisation for Economic
OECD 2022b, Overview: What is the difference between pledges, commitments and
disbursements?, OECD, Paris,
<https://www.oecd.org/aidfortrade/frequentlyaskedquestions/>.
Ogunleye, EK 2008, 'Natural resource abundance in Nigeria: From dependence to

development', Resources Policy, vol. 33, no. 3, pp. 168-174.
OPEC 2021, Nigeria facts and figures, Organization of the Petroleum Exporting Countries
(OPEC), Vienna, <https://www.opec.org/opec_web/en/about_us/167.htm>.
Ostrom, E 1990, Governing the commons: The evolution of institutions for collective action,
Cambridge University Press, England.
Pettorelli, N, Vik, JO, Mysterud, A, Gaillard, J-M, Tucker, CJ & Stenseth, NC 2005, 'Using the
satellite-derived NDVI to assess ecological responses to environmental change', Trends in
Ecology & Evolution, vol. 20, no. 9, pp. 503-510.
Prakash, K, Churchill, SA & Smyth, R 2020, 'Petrol prices and subjective wellbeing', Energy
Raggl, AK 2017, 'Natural resources, institutions, and economic growth: the case of Nigeria',
World Bank Policy Research Working Paper, vol. no. 8153.
Rahman, MS & Giessen, L 2017, 'Formal and informal interests of donors to allocate aid:
Spending patterns of USAID, GIZ, and EU forest development policy in Bangladesh', World
174
Sachs, JD & Warner, A 1995, Natural resource abundance and economic growth, National
Bureau of Economic Research Cambridge, Massachusetts.
Sachs, JD & Warner, AM 2001, 'The curse of natural resources', European Economic Review,
vol. 45, no. 4-6, pp. 827-838.
Sala-i-Martin, X & Subramanian, A 2013, 'Addressing the natural resource curse: An

illustration from Nigeria', Journal of African Economies, vol. 22, no. 4, pp. 570-615.
Sebri, M & Dachraoui, H 2021, 'Natural resources and income inequality: A meta-analytic
review', Resources Policy, vol. 74, p. 102315.
Tollefsen, AF, Strand, H & Buhaug, H 2012, 'PRIO-GRID: A unified spatial data structure',
Journal of Peace Research, vol. 49, no. 2, pp. 363-374.
UNEP 2005, The State of the Environment in Somalia: A Desk Study, United Nations
Environmental Programme Geneva.
United Nations 2021, Aid Flows in Somalia 2021, United Nations, Somalia,
<https://somalia.un.org/en/160002-aid-flows-somalia-2021>.
USAID 2021, Nigeria: History, USAID, Washington D.C,

<https://www.usaid.gov/nigeria/history>.
Venables, AJ 2016, 'Using natural resources for development: why has it proven so difficult?',
Journal of Economic Perspectives, vol. 30, no. 1, pp. 161-184.
Vesco, P, Dasgupta, S, De Cian, E & Carraro, C 2020, 'Natural resources and conflict: A meta-
analysis of the empirical literature', Ecological Economics, vol. 172, p. 106633.
Watanabe, Y 2006, 'What determines bilateral aid distribution? Evidence from major donors',
thesis, University of Oregon.
Woldekidan, H 2015, 'The role of foreign aid in reducing poverty: Time series evidence from
Ethiopia', Journal of Economics and International Finance, vol. 7, no. 3, pp. 59-71.

Yang, H-m, Liu, P-l & Guo, Y 2018, 'Determinants of China’s development assistance for
health at the sub-national level of African countries (2006–2015)', Infectious diseases of
poverty, vol. 7, no. 1, pp. 1-9.
Zallé, O 2019, 'Natural resources and economic growth in Africa: The role of institutional
quality and human capital', Resources Policy, vol. 62, pp. 616-624.
175
Chapter 5 – Appendix 5.A
VARIABLES Description Mean SD
Total aid commitments Logarithm of total amount of aid (in US$) committed 1.580 4.783
to a district in a given year
Total aid disbursements Logarithm of total amount aid (in US$) disbursed to a 1.405 4.468
district in a given year
Total aid commitments (dummy) =1) if aid was committed to a district in a given year 0.100 0.300
Total aid disbursements (dummy) =1) if aid was disbursed to a district in a given year 0.092 0.288
Bilateral aid =1) if a district received bilateral aid in a given year 0.062 0.241
Multilateral aid =1) if a district received multilateral aid in a given 0.055 0.228
year
NDVI Normalized Difference Vegetation Index 9.782 3.859
Total minerals Total number of mineral deposits found in a district 0.088 0.290
Iron ore Total number of iron ore deposits in a district 0.001 0.033
Coal Total number of coal deposits in a district 0.085 0.285
Population Logarithm of total number of people in a district 1,229 133.6
Night-time lights Mean of night-time light 5.789 5.012
Capital distance Distance to the capital city in kilometres 5.784 0.642
Area Area of district in square kilometres 7.974 0.143
Rainfall District average rainfall per year 5.335 0.740
Temperature District average temperature per year 27.23 2.619
176
Chapter 6
Conclusion to Thesis
6.1 Summary and Conclusions
Foreign aid is regarded as an important tool to achieve sustainable development in

developing nations. To evaluate the efficacy of aid, several studies, employing cross
country analyses at the macro level, have examined the impact of aid on various
development outcomes, and findings from these studies are mixed (see, e.g., Birchler
& Michaelowa 2016; Munyanyi et al. 2020; Thomson, M et al. 2017; Wayland 2019).
Some of these studies posit that aid improves development-related outcomes such as
economic growth, education, health, access to water and sanitation, gender equality,
and poverty, among others (see, e.g., Banchani & Swiss 2019; Galiani et al. 2017;
Riddell & Niño-Zarazúa 2016), while other studies claim the opposite (see, e.g., Maïga
2014; Moyo 2009; Narang 2014; Thomson, M et al. 2017). A potential reason
advanced in the literature for these mixed results is that aid is too small and localised
to affect aggregate outcomes, hence the need to find other approaches to examine
aid’s effectiveness, particularly at the sub-national or less aggregated levels (Dreher
& Lohmann 2015).
Examining aid impacts at the sub-national level is a better approach, given that it helps
to reduce the unobservable differences that may exist in cross-country settings (Rajan
& Subramanian 2008). If not controlled for, these unobservable differences may lead
to spurious relations between aid and outcomes. Additionally, sub-national level
analysis enables us to examine aid at the specific geographic location to which it was
allocated. By defining areas that did and did not receive aid, sub-national aid data
allows for quasi-experimental approaches to estimating aid impacts (Marty et al.
2017).
Given the limited number of studies examining aid efficacy at the sub-national level
(see, e.g., De & Becker 2015; Kotsadam et al. 2018; Marty et al. 2017; Odokonyero et
al. 2018), this thesis examines aid efficacy and allocation at the sub-national level. In
doing so, the thesis makes significant contributions to the literature in several ways.
First, the thesis extends and contributes to the growing literature on the impact of aid
at the sub-national level by examining the effect of sub-national foreign aid on other
177
important outcomes other than health (e.g., energy poverty). Second, the thesis adds
to the literature that seeks to understand factors that help to narrow gender gaps in
health outcomes by showing that aid is an important factor. Third, the thesis
demonstrates the mediating role of aid between outcomes, for instance, between
temperature shocks and household borrowing. Last, by investigating the effect of
natural resources impacts on the allocation of aid, the thesis contributes to the
literature on the determinants of aid allocation at the sub-national level.
Understanding the impact and allocation of aid at the sub-national level improves
policy regarding fiscal adjustments, supplementary budgets, and developmental
programs designed to improve human welfare. Further, such knowledge helps the
donor community assess the effectiveness of aid and allocate it in a smarter way, as
advocated by the 2030 Agenda for Sustainable Development (United Nations 2015).19
Chapter 2 investigates the impact of sub-national foreign aid on child health outcomes.
Matching geo-referenced sub-national foreign aid and DHS datasets for Senegal and
applying a DID approach, the results show that aid reduces the likelihood of energy
poverty among households living near aid projects. The chapter contributes to the
literature on energy poverty determinants and foreign aid effectiveness. In addition,
the chapter also identifies channels through which energy poverty could transmit to
energy poverty, including education, income poverty and economic growth.
Chapter 3 examines the effect of sub-national aid on three categories of child health
outcomes, including anaemia prevalence, vaccination coverage and mortality rates in
the Democratic DRC. Using data from the AidData projects and DHS, results from DID
estimations show that aid increases vaccination coverage, lowers mortality rates, and
reduces anaemia prevalence among children living close to aid project sites. This
chapter also examines the role of foreign aid in influencing the gender gap in child
health outcomes. The findings indicate that aid narrows the gender gap in the
prevalence of anaemia, mortality, and vaccinations against tuberculosis, DPT, polio,
measles, and vitamin A deficiency.
19The 2030 Agenda for Sustainable Development, adopted in 2015, commits to achieving sustainable
development by the year 2030 worldwide. According to the Agenda, one of the many ways that should
be used to achieve sustainable development in developing countries is investing in ‘smarter’ foreign
aid. Additionally, foreign aid should be carefully allocated and channelled to reach areas, societies and
individuals that require it the most.
178
Chapter 4 explores the impact of temperature shocks on household borrowing and the
mediating role of foreign aid at the sub-national level. Geographically matching
Malawian household data with weather and foreign aid data and applying a panel fixed
effects model, the results show that temperature shocks have a positive and
statistically significant effect on household borrowing. Moreover, results also show that
1) total aid, 2) education aid, 3) health aid, 4) rural development aid, and 5) water,
sanitation and irrigation aid are the channels through which temperature shocks
influence household borrowing.
Chapter 5 contributes to the literature on the determinants of foreign aid allocation by

examining the effect of natural resources on sub-national aid allocation. Using sub-
national data on natural resources and aid for a panel of three African countries and
estimating a fixed-effects model, the chapter finds that natural resources positively
affect aid. That is, areas endowed with natural resources have a higher probability of
receiving aid than those without resources. The chapter also investigates transmission
mechanisms through which natural resources influence aid allocation. The findings
show that conflict and economic growth are important mechanisms that mediate the
relationship between natural resources and aid allocation.
6.2 Policy Implications and Recommendations
This thesis presents several policy-relevant suggestions. First, based on the findings
from Chapter 2, the question of whether aid is effective in improving development
outcomes needs to be revisited. The findings show that aid lowers energy poverty
among households who live close to aid projects sites. This suggests that for
households to benefit from aid projects, these projects should be scaled-up or spread
out across communities. Results from the channel analysis conducted in this chapter
imply that policies aimed at promoting education, economic growth and household
wealth are crucial in lowering energy poverty among households. Lowering energy
poverty would also improve other outcomes such as health, economic growth, life
expectancy, employment, and the environment, among others.
Chapter 3 finds that sub-national aid improves child health outcomes, such as
vaccination against diseases, anaemia, and mortality. Further, aid is also found to
narrow the gender gaps in these outcomes. Based on the results and consistent with
the policy suggestions presented in Chapter 2, aid projects need to be established
179
close to households to ensure aid efficacy. That is, there is a need to establish
community-based aid projects. With regard to narrowing the gender gaps in health
outcomes, the results suggest that foreign aid can be an important policy instrument
when devising policies aimed at achieving the goal of gender equality.
Based on the results from Chapter 4, policymakers should engage policies aimed at
ensuring the smooth flow of foreign aid and support areas targeted by aid, such as
education, health, rural development, and water, sanitation and irrigation. Further, the
government could improve surveillance over the usage of the allocated foreign aid,
increase accountability and enact laws that discourage corrupt behaviour. By doing
so, foreign aid may stand a chance in lowering household borrowing and improving
other outcomes.
Last, results in Chapter 5 show that natural resources attract aid. In this regard, aid
recipient governments should manage and utilise natural resources effectively and
efficiently to attract aid. Policies that discourage rent-seeking and corrupt behaviour
that may compromise the efficient utilisation of resources should be established.
6.3 Limitations and Suggestions for Future Research
Despite the thesis achieving its aims and objectives, some unavoidable limitations
were encountered. Overall, the thesis employed sub-national foreign aid data from the
AidData database – the only database that currently reports on foreign aid at the sub-
national level, which understates the volumes of aid allocated to developing countries.
Future studies should use other data sources on sub-national foreign aid as and when
they become available.
Chapter 2, due to data availability issues, could not explore the mediating role of other
important channels such as energy efficiency and income inequality. In this regard,
future studies should examine how these channels mediate the relationship subject to
data availability. Similar to Chapter 2, Chapter 3 could not examine the mediating role
of institutional quality between sub-national aid and child health outcomes due to data
availability issues. Given that institutional quality is an important variable when
examining aid impacts, future studies could incorporate this variable in exploring the
moderating role of aid at the sub-national level.
180
Considering Chapter 4, future studies could 1) explore the effect of temperature
shocks on household borrowing over a period exceeding three years, 2) examine other
types of weather-related shocks such as rainfall, and 3) identify channels other than
aid that mediate the relationship between temperature shocks and household
borrowing. With regard to Chapter 5, future studies should examine how institutional
quality mediates the relationship between natural resources and aid allocation in the
sub-national context as data become available.
181
References
Birchler, K & Michaelowa, K 2016, 'Making aid work for education in developing countries: An
analysis of aid effectiveness for primary education coverage and quality', International Journal
of Educational Development, vol. 48, pp. 37-52.
AidData, USA.
vol. 2, no. 1.
Macmillan, UK.
Narang, N 2014, 'Humanitarian assistance and the duration of peace after civil war', The
Journal of Politics, vol. 76, no. 2, pp. 446-460.
Rajan, RG & Subramanian, A 2008, 'Aid and growth: What does the cross-country evidence
really show?', The Review of Economics and Statistics, vol. 90, no. 4, pp. 643-665.

United Nations 2015, Transforming our world: the 2030 Agenda for Sustainable Development,
United Nations, New York.
182
156.
183
Appendix A
Publications and awards originating from the present thesis
Peer-reviewed Journal Articles
╪ Published
Chapter 3: Foreign aid and energy poverty: Sub-national evidence from
Senegal (2022), Energy Economics (ABDC ranking: A*-ranked)
╪ Revised and resubmitted

Chapter 2: Foreign aid and child health outcomes: Micro-evidence from the
Democratic Republic of Congo (revised and resubmitted), Health Economics
(ABDC ranking: A*-ranked)
Awards
╪ Vice-Chancellor’s Prize for Research Impact – Higher Degree by Research,

RMIT University, Australia, October 2021.
╪ Best HDR Milestone Conference Paper Award for Second Milestone, The 2020
College of Business and Law Spring Conference, RMIT University, Australia,
November 2020.
184

Munyanyi

Uploaded by

Copyright:

Available Formats

Munyanyi

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Munyanyi

Uploaded by

Copyright:

Available Formats

Essays on the Allocation and Efficacy of Foreign Aid at the

A thesis submitted in fulfilment of the requirements for the degree of

Musharavati Ephraim Munyanyi

School of Economics, Finance and Marketing

I give permission for the digital version of my research submission to be made

Musharavati Ephraim Munyanyi

Musharavati Ephraim Munyanyi

The literature explores various determinants of aid allocation, including recipient’s

Understanding aid impacts and allocation at the sub-national level provides a

Baliamoune‐Lutz, M 2016, 'The Effectiveness of Foreign Aid to Women's Equality

Laokri, S, Soelaeman, R & Hotchkiss, DR 2018, 'Assessing out-of-pocket expenditures for

Lechthaler, F, Matthys, B, Lechthaler-Felber, G, Likwela, JL, Mavoko, HM, Rika, JM,

Thomson, M, Kentikelenis, A & Stubbs, T 2017, 'Structural adjustment programmes adversely

Warnatzsch, EA & Reay, DS 2019, 'Temperature and precipitation change in Malawi:

World Bank 2020, DataBank: World Development Indicators,

Foreign aid has emerged as an important policy intervention for achieving

2.2 A Brief Overview of Related Literature

2.3 Data and Variables

2.3.1 Health outcomes

2.3.2 Sub-national aid

We include a set of socio-demographic control variables consistent with the existing

2.4.1 Main empirical specification

𝐶𝐻𝑖,𝑗,𝑡 = 𝛼0 + 𝛼1 𝑡𝑖𝑚𝑒 + 𝛼2 (𝑡𝑖𝑚𝑒 ∗ 𝑎𝑖𝑑𝜃𝑘𝑚𝑗,𝑡−1 ) + 𝐸𝐺𝑗,𝑡 + 𝜗𝑋𝑖,𝑗,𝑡 + 𝜀𝑖,𝑗,𝑡 (1)

In addition to internal bureaucratic processes in donor countries, the existing literature

2.4.2 Oaxaca-Blinder decomposition model

2.5.1 Main results: Aid and child health outcomes

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

Panel A: Bivariate results

Controls Yes Yes Yes Yes Yes Yes Yes

2.5.2 Robustness checks

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

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 30,253 30,155 30,257 30,164 29,887 29,467 31,728

Table 2.3 – Alternative aid types

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

Panel A: Health aid

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Table 2.4 – Foreign aid intensity

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

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

Table 2.5 – Foreign aid and conflict

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

Observations 10,761 10,694 10,754 10,706 10,711 9,966 11,190

Observations 33,873 33,676 33,965 33,719 33,708 32,985 35,892

Observations 44,634 44,370 44,719 44,425 44,419 42,951 47,082

4 More information on UCDP can be found here: https://ucdp.uu.se/.

Another important consideration is the year-gap between the pre-implementation and

2.5.1 Gender differentials and child health outcomes

Controls Yes Yes Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes

Table 2.8 – Oaxaca-Blinder child health outcomes decomposition

Total 0.004*** 0.004 -0.0003

Total 0.003 0.015*** 0.013***

Total -0.0005*** -0.002 -0.002

Total 0.003 0.015* 0.013*