The effect of project aid fragmentation on economic growth

ABSTRACT

Aid fragmentation has been examined as a negative phenomenon to development outcomes based on aid volumes. However, aid fragmentation is a situation in which a large number of fragmented and non-coordinated donor projects carried out using procedures that vary from donor to donor are introduced in a recipient country, raising the transaction costs. Therefore, this article examines the effect of aid fragmentation based on the number of projects. It confirms that project aid concentration increases economic growth in poor countries with a higher degree of aid dependence. Under the SDGs, it is expected that aid fragmentation will become a more pressing concern.

KEYWORDS:

• Aid – Aid effectiveness, Development policies
• Governance and public policy
• Labour and livelihoods – economics

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Mitsuaki Furukawa is Professor, School of International Relations, University of Shizuoka. He was previously Director General of Security Management Department (2016–2019) at the Japan International Cooperation Agency (JICA), Japanese governments’ agency tasked with providing bilateral Official Development Assistance, Chief Representative for the South Sudan Office (2014–2016), Executive Senior Research Fellow of JICA Research Institute (2009–2014) and Resident Representative of JICA UK Office (2007–2009), among other posts. He holds a PhD in Social Sciences from Hitotsubashi University, Japan.

Notes

1 CRS contains detailed information on individual aid activities of most of the 23 member countries of the OECD’s Development Assistance Committee (DAC), and the multilateral development banks and UN agencies. The dataset is available at www.oecd.org/dataoecd/20/29/31753872.htm.

2 This is why General Budget Support was considered the preferred aid modality under the Paris Declaration.

3 http://www.oecd.org/dac/stats/usersguidetothecreditorreportingsystemcrsaidactivitiesdatabase.htm.

4 Hudson (2012, 1) believes that on a panel data basis, it is possible to conduct meaningful analysis using data from 2002 onwards.

5 Kharas (2007) considers CPA a better representation of the realities than ODA.

6 To count the number of projects, we used the same method as Frot and Santiso (2010). Aid activities are identified in the DAC CRS database by assigning an unique identification number to each activity, so different projects have different IDs, and we are able to count the number of projects for each donor. Moreover, projects with a transfer of zero or negative dollars have been removed. However, it must be noted that there are, for example, cases where an individual project coordinates aid in several locations, and the ensuing transaction costs cannot be captured in the number of projects. The analysis is limited by the present difficulty of obtaining these data.

7 Other than the HHI, the Theil Index has also been used as a method to measure concentration. Acharya, Fuzzo de Lima, and Moore (2006) used the inverse Theil Index to measure the Donor Proliferation Index, which represents the extent to which a donor proliferates a certain aid budget in a potential beneficiary country. According to the explanation in Acharya, Fuzzo de Lima, and Moore (2006, 37), the Theil Index can be found using the equation below. Supposing that aid to beneficiary country i accounts for ratio xi of the total amount of aid from a certain donor, we first seek H(x) below (assuming that potential beneficiary countries are n countries).

$H\left(x\right)\equiv \sum _{i=1}^n{x}_i\log \left({\displaystyle \frac{1}{x_i}}\right)$

When identical amounts of aid are distributed equally to all n countries, and all beneficiary countries receive 1/n each, the maximum value for H is $H\left(x\right)\equiv {\sum }_{i=1}^n{\displaystyle \frac{1}{n}\log \left(n\right)}$. The difference between this maximum value and the actual H (x) is the Theil Index (use ${\sum }_{i=1}^n{x}_i=1$).

$T=\log n-H\left(x\right)\equiv \sum _{i=1}^n{x}_i\log \left(n{x}_i\right)$

If all potential beneficiary countries receive the same amount of aid, the Theil Index T takes the minimum value of 0, but if the gaps widen, the value increases. If the potential beneficiary country is one country (n = 1), then T = 0. In this way, the Theil Index is a useful index when a donor provides aid only to some countries among potential beneficiary countries, but if attention is focused on aid concentrated with particular donors, HHI is more appropriate.

8 Panel data, which this article relies on, are more informative, provide more variability but less multicollinearity among the variables, and offer a greater degree of freedom and more efficiency (Baltagi 2005, 5; Kitamura 2005). Nevertheless, it is not possible to eliminate the problem of multicollinearity completely.

9 The Nordic Plus countries consist of Denmark, Finland, Norway, Sweden, Ireland, the Netherlands, and the United Kingdom. Until 2002, the group covered only four Nordic countries (Denmark, Finland, Norway, and Sweden). After 2002, the group expanded to include Ireland, the Netherlands and the UK and adopted the name Nordic Plus.

10 It is possible to estimate System GMM with the xtabond2 Stata command created by David Roodman.