The Impact of Foreign Aid on Economic Growth in Africa: Empirical Evidence from Low Income Countries

Abstract

This article aims to shed some insights into the ongoing debate on the aid-growth nexus by examining whether sources of aid matter in explaining aid effectiveness. In doing so, we consider three main proxies for bilateral aid based on three sources of aid such as Total Aid (TA); Traditional Donors aid (TDA) and Non-Traditional Donors aid (NTDA) as independent variables in a dynamic panel growth model within a system GMM framework. The study uses a panel dataset from 25 Low-Income Countries (LICs) in Africa over the period 2000–2017. The main findings show that the impact of aid on economic growth appears to be negative and significant for TA and TDA proxies, while it is positive but insignificant when the aid proxy is NTDA. A relatively larger share of TA and TDA disbursement away from the direct growth-enhancing productive sectors towards the unproductive sectors seems to have contributed to their strong negative impact on growth. The key policy implication is that governments in LICs in Africa and donors should work in collaboration to design effective ways to ensure that TDA should target the direct growth-enhancing sectors.

Keywords:

• foreign aid
• growth
• LICs
• system GMM
• aid sources
• Africa

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 Broadly speaking, as shown in Greenhill et al. (2013), Traditional Donors (TDs) refer to donors which are commonly known as Development Assistant Committee (DAC) members while Non-Traditional Donors (NTDs) stand for those donors outside the DAC system. The list of TDs and NTDs is shown in the second section, Table 2.

2 Somalia appears to have an exceptionally higher degree of aid dependency even exceeded 85 percent in 2017 (86.4 percent): about a 9.1-fold increase from 9.5 percent in 2000 to 86.4 percent in 2017. It shows a substantial variation across aid sources where it significantly increased about 32-fold for NTD’s aid from 0.2 percent in 2000 to 6.1 percent in 2017 while it increased about 8.7-fold for TD’s aid between 2000 (9.3 percent) and 2017 (8.3 percent). When Somalia is excluded, the average aid dependency for the 26 LICs becomes 6.1 percent (total aid), 5.9 percent (TD’s aid) and 0.3 percent (NTD’s aid) during 2000–2017.

3 As the common practise, this study follows the OECD CRS sectoral classification of aid commitments.

4 Among 10 NTDs (Brazil, Chile, China, Colombia, Costa Rica, India, Indonesia, Mexico, Qatar, and South Africa) that do not report to the OECD system, nine of them are excluded from the study due to unavailability of the required dataset from any credible sources. Data on aid from China is extracted from AidData online database, which is increasingly becoming the most credible source of data for Chinese global aid flows since 2000. AidData has compiled ‘ODA-like’ (such as grants, interest-free loans and concessional loans) flows from China since 2000 and the latest to 2014 (constant 2014 USD). Unlike other studies that have used ‘ODA-like’ flows as foreign aid, we excluded interest-free loans as it does not qualify the DAC aid definition. Only the type of aid flows that qualify DAC ODA criteria (grants and concessional loans with a grant element of at least 25 percent) are extracted from AidData and included in this study. Moreover, concessional loans with no information on the grant element are excluded.

5 Consistent with recent literature (Herzer & Morrissey, 2013; Wamboye et al., 2013), we use the three macroeconomic policy variables separately rather than a policy index (or interaction term between aid and policy index) used in Burnside and Dollar (2000) at least for two reasons. First, it is hardly possible to empirically disentangle which variable drives the effects as the policy index may hide the different effects of each policy component. Second, available empirical evidence on the relationship between interactive terms and aid effectiveness is highly inconclusive.

6 The traditional IV approach that relies on ‘external’ instruments using OLS/2SLS and fixed-effects model has been highly criticized for failing to find a valid ‘external’ instrument to control aid endogeneity in the aid-growth regression. After carefully reviewing the existing standard external instruments for aid (such as lagged aid, population, rainfall, colonial legacy, primary exports, arms imports, policy, policy interactions, GDP per capita, Egypt dummy) commonly used among past studies, recent literature (Werek et al., 2008; Armah, 2010) has argued that these instruments have been hardly valid. This is because these instruments have either: failed to satisfy the main requirement for instrument validity which is exogeneity, or tended to be time-invariant.

Additional information

Notes on contributors

Mamo G. Tefera

Mamo G. Tefera is currently a PhD Student in the Department of Economics at the University of South Africa. Mr. Tefera' main interests are macroeconomics and development economics.

Nicholas M. Odhiambo

Nicholas M. Odhiambo is a seasoned academic whose research has made a significant contribution to the scholarly body of knowledge. He is currently one of the top economics researchers, not only in Africa, but also internationally. He is also one of the most-cited and most-read economics authors worldwide – according to RePEc/IDEAS. He is currently working as Professor of Economics and Head of Macroeconomic Policy Analysis research flagship programme at the University of South Africa (UNISA).