Budget support to the health sector—The right choice for strong institutions? Evidence from panel data

• Published date: 01 May 2023
• Author: Tim Röthel
• Date: 01 May 2023
• DOI: http://doi.org/10.1111/rode.12967

REGULAR ARTICLE

Budget support to the health sector—The right

choice for strong institutions? Evidence

from panel data

Tim Röthel

Faculty of Law, Business and Economics,

University of Bayreuth, Bayreuth,

Germany

Correspondence

Tim Röthel, Faculty of Law, Business and

Economics, University of Bayreuth,

Bayreuth, Germany.

Email: tim.roethel@uni-bayreuth.de

Abstract

This article examines the relationship of sector budget

support to the health sector and the infant mortality

rate for developing countries. Project-type interventions

have been widely used in developing countries in the

past decades. These smaller-scale interventions often

did not bring the results that the donors would have

wanted, at least on a macro level. At the beginning of

the millennium, forums on aid effectiveness proposed

new principles to increase the effectiveness of aid.

Many scholars agreed that one of the answers would be

budget support. This article tries to answer whether

budget support is the efficient aid modality in countries

with strong institutions. In the baseline scenario, a

panel data analysis is applied, which includes 113 coun-

tries between 2010 and 2018. This dynamic linear panel

model is estimated by using ordinary least squares

(OLS) and system generalized method of moments

(GMM). Health sector aid, in general, has a significant

and negative effect on the infant mortality rate in the

average country. Sector budget support is insignificant

in the baseline estimation and when interacted with a

governance variable. In contrast, project-type interven-

tions exhibit significant and negative effects on the out-

come variable. The results indicate that sector budget

Received: 29 March 2022 Revised: 1 December 2022 Accepted: 5 December 2022

DOI: 10.1111/rode.12967

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distrib ution and

reproduction in any medium, provided the original work is properly cited.

© 2022 The Author. Review of Development Economics published by John Wiley & Sons Ltd.

Rev Dev Econ. 2023;27:735–770. wileyonlinelibrary.com/journal/rode 735

support might not be the superior choice among the

aid modalities in the health sector, even in countries

with good governance.

KEYWORDS

development, development policy, health, mortality, official

development assistance

JEL CLASSIFICATION

F35, I15, I18, O15, O20

1|INTRODUCTION

Ever since the first dollar of development aid has been transferred to developing countries,

there has been a debate on the effectiveness of aid and the ideal way how it should be delivered.

In the early years, there was still a lack of quantitative analyses (Adelman & Chenery, 1966;

Baldwin, 1969), and the focus laid on the relationship between aid, savings, and investments,

inspired by the Harrod-Domar model (Arndt, Jones, & Tarp, 2010).

This changed by the end of the 1990s. Starting with a political debate, dominated by Jeffrey

Sachs and William Easterly, the question was if and how aid should be delivered.

1

Then, a large

number of cross-country empirical evidence was published. These studies mainly examined

the effect of aid on gross domestic product (GDP) growth. The ambiguous results did not

solve the political debate either. On the one hand, aid was perceived to positively influence

growth (e.g., Burnside & Dollar, 2000; Hansen & Tarp, 2001), and on the other hand, no signifi-

cant relationship between these two variables could be found (e.g., Easterly, Levine, &

Roodman, 2004; Rajan & Subramanian, 2008). This debate has not ended yet. While the meta-

study by Mekasha and Tarp (2013) finds an overall significant and positive effect of aid on

growth, more recent studies of aid on growth and aid on productivity find ambiguous or even

negative effects (e.g., Bird & Choi, 2020; Groß & Nowak–Lehmann Danzinger, 2022).

A third strand of literature focuses on aid on a more disaggregated level. This includes

experimental approaches or impact evaluations of single development activities

(e.g., Banerjee & Duflo, 2012) and studies researching the effect at the sub-national level. One

common approach is to use geocoded aid data and the nightlight density to estimate the effect

of aid on growth (e.g., Dreher et al., 2021; Khomba & Trew, 2022). Other studies only examine

the effect of aid on specific sectors. The idea of these approaches is that the contradicting results

regarding aid and GDP growth might be explained by the time it takes until aid flows are trans-

lated into economic growth and the complex link between these two variables. Therefore, some

authors look at more disaggregated data where potential outcomes are more closely related to

the aid flows, and a possible effect can be identified more easily. One example is aid to the

health sector. Similarly, some studies use cross-sectional data, while others look at sub-national

data only. The advantage of sub-national data and the focus on one country only is that a clear

picture of the effect of aid on health outcomes can be drawn. De and Becker (2015), for exam-

ple, find a positive effect of health-related aid on the decrease of disease severity in Malawi.

Kotsadam, Østby, Rustad, Tollefsen, and Urdal (2018) look at the effectiveness of aid on the

infant mortality rate in Nigeria. They find that a closer proximity to aid projects leads to lower

736 RÖTHEL

infant mortality in the area. However, the results from these studies might not necessarily be

valid in other countries. Thus, it might be interesting to look at panel data studies as well to get

a broader view of aid effectiveness in the health sector. Mishra and Newhouse (2009) find a sig-

nificant effect of health sector aid on infant mortality. Similarly, Yogo and Mallaye (2015) show

that a significant relationship exists between health sector aid and child mortality. Recent stud-

ies of health aid on the infant mortality rate support these positive effects on the decrease of the

outcome variable (e.g., Woode, Mortimer, & Sweeney, 2021). These results regarding health out-

comes

2

are in line with the development of several health indicators that show dramatic

improvements over the past 30 years in low- and middle-income countries (World Bank, 2021).

For example, life expectancy at birth improved from around 63 years in 1990 to around 71 in

2018. Similarly, the infant mortality rate fell from around 71 in 1990 to around 31 in 2019

(World Bank, 2021). However, apart from the previously cited studies, evidence with respect to

the effectiveness of health sector aid compared to official development assistance (ODA) is still

relatively rare.

In addition, most researchers look at general aid only. However, to improve policy recom-

mendations, it is essential to further distinguish between different kinds of total aid or even dif-

ferent kinds of health sector aid. The Development Assistance Committee (DAC) of the

Organisation for Economic Co-operation and Development (OECD) defines eight different

modalities on how aid can be delivered (OECD, 2021b),

3

with the most well-known forms of

aid being project-type interventions and budget support, which will be further examined in this

study.

To the best of the author's knowledge, no cross-section analysis has yet either tested the

effectiveness of budget support in the health sector or compared these two types of delivering

aid to recipient countries. This distinction might be the answer to the debate of Easterly and

Sachs since different institutional settings might simply require different aid modalities.

This is why, in this article, a panel study is conducted to estimate the effect of health sector

aid on infant mortality using 113 countries between 2010 and 2018 in the baseline scenario.

Similar to the article by Mishra and Newhouse (2009), the starting point is the aforementioned

relationship, to then further disaggregate the aid data into project-type interventions and budget

support. Eventually, an interaction term between a governance variable and budget support is

used to test if the modality performs any better in the presence of strong institutions.

The main findings of this article are that aid to the health sector does significantly decrease

the infant mortality rate. The same applies to project-type interventions where an even stronger

effect can be found. Sector budget support has no significant effect on infant mortality for the

standard specification. The effect remains the same when budget support is interacted with a

governance variable.

The course of this paper is as follows: Section 2provides an overview of the two aid modali-

ties, project-type interventions and budget support, and discusses their advantages and disad-

vantages. In Section 3, the empirical specification and identification strategies are introduced,

and the data are briefly described. Section 4presents the results with respect to the hypotheses

formulated and tests their robustness. Section 5summarizes the results and concludes.

2|AID MODALITIES

Project-type interventions, often simply known as project aid, are the most well-known and

most widely used aid modality. They are defined as a “set of inputs, activities and outputs …to

RÖTHEL 737