Abstract
We assess whether bilateral and multilateral donors of foreign aid specialized and coordinated their activities with other donors as agreed in the Paris Declaration of 2005. We account for donor heterogeneity, varying aid priorities and recipient characteristics in order to isolate changes in donor behaviour over time. Recent shifts in aid priorities, such as the rising importance of general budget support, have reduced the fragmentation of aid. Nevertheless, our results reveal that aid fragmentation persisted after the Paris Declaration and coordination among donors has even weakened.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In addition, the European Union passed the European Consensus on Development in November 2005, which “for the first time identified a framework of common principles for the member states and the commission in this policy field [development cooperation]” (Engel and Keijzer 2008).
According to Berthélemy (2006), the sign on aid from other donors switches from significantly positive to significantly negative once fixed effects are accounted for.
Similar to Frot and Santiso (2011), Brück et al. (2011) employ aggregate aid data to analyse how often aid accelerations coincided between donors in response to shocks and policy changes in the recipient countries during the period 1960-2007. Both studies do not assess co-movements of donors at the level of specific aid sectors (see below). Moreover, the herding measures of Frot and Santiso (2011) do not account for aid quantities as co-movements simply relate to the proportion of donors increasing (or decreasing) their aid allocation to recipient i in period t.
Note also that the Paris Declaration has been evaluated in considerable detail from the perspective of individual recipient countries. Most recently, Wood et al. (2011: 26) concluded, inter alia, that “aid fragmentation is still found to be high in at least half of the evaluations.” The summary report of Wood et al. (2011) provides mainly qualitative assessments on the basis of various case studies conducted by independent evaluation teams managed by the respective partner country.
See OECD (2011a, Table 1.1) for more detailed definitions and the gaps between targets and achievements. In some contrast to the general statement of ‘considerable progress’, the report concludes that “little progress has been made among donors to implement common arrangements or procedures and conduct joint missions and analytical works” (page 16).
The number of participating countries increased from just 32 in 2006 and 2008 to 78 in 2011.
The report provides the example of a Latin American country where, according to the country’s own assessment, program-based approaches did not exist, whereas its donors reported that 64 % of aid was delivered through PBAs.
See Appendix 1 for the list of aid sectors.
The Herfindahl index attaches disproportionately high weights to the largest aid shares of particular recipients and sectors in a donor’s overall aid budget.
The OECD defines another criterion from the recipient’s perspective that is omitted in the following, namely whether the donor belongs to the largest donors that cumulatively contribute at least 90 % of the recipient’s total aid inflows.
Note that this measure relates to the number of bilateral aid relations. By contrast, Birdsall and Kharas (2010) propose an indicator on the donor’s specialization (Indicator ME5) that relates to the amount of aid granted in significant and non-significant bilateral relations. As exemplified in Appendix 5, both measures may deviate substantially from each other, as well as from the Theil index.
See Aldasoro et al. (2010) for more details and the relevant literature.
We also estimated a random effects model, but the Hausman test rejected this model as inconsistent.
While we control for general budget support (DAC/CRS code 510), debt relief operations (code 600) and also aid classified as “multisector/cross-cutting” (code 400), we follow Frot and Santiso (2011) in excluding emergency food aid (code 710) from the analysis. Donors routinely react to natural disasters and famines by increasing food aid; Frot and Santiso use the term “beneficial” herding if aid overlaps increase as a result.
Recall that time-invariant level effects of economic and political clout would be captured by the donor fixed effects δ j .
See also Davies and Klasen (2011) who argue that increasing selectivity in terms of donors focussing on needy and deserving recipients could result in crowding-in effects of aid from other donors on aid from donor j.
In a robustness test, we use only the data since 2004 (see Sect. 4 below).
See the list in Appendix 2. The two multilateral donors have to be excluded when estimating extended specifications of our model (with donor characteristics related to trade and political interests included).
Note that the variation of these variables is limited over time. Moreover, aid overlaps may even increase among donors with similar trade and political interests.
Given the relatively low number of clusters (i.e. 19 donors) the cluster-robust standard errors are to some extent downwards biased. Therefore, we alternatively performed estimations with non-clustered standard errors. Results turned out to be robust to non-clustering.
HIPC stands for heavily indebted poor countries.
We also experimented with weighting these characteristics by (i) the recipient countries’ population and (ii) the amount of aid granted by donor j to these countries. Both weighting schemes appear to be problematic, however, so that results are not reported. Using population weights implies that China and India dominate all other recipient countries. Using aid amounts may lead to biased results especially for the Theil indices, the calculation of which is also based on the aid amounts granted by donor j.
Note that aid overlaps are not necessarily affected by a stronger focus on poorer countries as long as donors specialize in particular sectors within these countries (in-country division of labor).
See Davies and Klasen (2011) for a similar line of reasoning.
This is corroborated by regressions for the sub-period before the Paris Declaration (not shown), where the respective coefficients are found to be insignificant.
If dropping the control variables did not matter, the fact that we account for donor fixed effects would still provide a justification for pursuing regression analysis rather than relying on purely descriptive statistics.
Note that the fully specified estimation in columns (4) and (8) of Table 5 does not cover the year 2009 due to missing data on UN voting patterns.
For a detailed discussion on inequality and concentration measures in the aid context, see Dreher and Michaelowa (2010).
We are most grateful to an anonymous reviewer for having alerted us to this issue.
References
Acharya, A. K., Fuzzo da Lima, A. T., & Moore, M. (2006). Proliferation and fragmentation: Transaction costs and the value of aid. Journal of Development Studies, 42(1), 1–21.
Aldasoro, I., Nunnenkamp, P., & Thiele, R. (2010). Less aid proliferation and more donor coordination? The wide gap between words and deeds. Journal of International Development, 22(7), 920–940.
Berthélemy, J.-C. (2006). Bilateral donors’ interest vs. recipients’ development motives in aid allocation: Do all donors behave the same? Review of Development Economics, 10(2), 179–194.
Bigsten, A. (2006). Donor coordination and the uses of aid. Göteborg University (http://gupea.ub.gu.se/dspace/bitstream/2077/2723/1/gunwpe0196.pdf; accessed:November 2011); French version: Revue d’économie du développement 14 (2/3): 77–103
Birdsall, N., & Kharas, H. (with A. Mahgoub and R. Rerakis) (2010). Quality of Official Development Assistance Assessment. Washington, DC: Center for Global Development.
Brück, T. & Guo, X. (2011). Who gives aid to whom and when? Aid accelerations, shocks and policies. DIW Discussion Papers 1133. Berlin: German Institute for Economic Research.
Chandy, L. (2011). It’s complicated: The challenge of implementing the Paris Declaration on aid effectiveness. Washington, DC: Brookings Institution (http://www.brookings.edu/opinions/2011/0922_paris_declaration_chandy.aspx;accessed: November 2011).
Davies, R. B. & Klasen, S. (2011). Of donor coordination, free-riding, darlings, and orphans: The dependence of bilateral aid commitments on other bilateral giving. University College Dublin and University of Göttingen (http://econstor.eu/bitstream/10419/48353/1/47_klasen.pdf; accessed: November 2011).
Dreher, A. & Michaelowa, K. (2010). Final report: methodology to measure progress towards in-country division of labor. Study on behalf of GTZ. Goettingen and Zurich, mimeo.
Easterly, W. (2007). Are aid agencies improving? Economic Policy, 22(52), 633–678.
Engel, P., & Keijzer, N. (2008). Donor harmonisation and the EU: In risk of a two-speed Europe. D + C Development and Cooperation 2: 60–62. (http://www.inwent.org/ez/articles/065233/index.en.shtml, accessed: November 2011).
European Commission (2008). Budget support: The effective way to finance development? EuropeAid Cooperation Office, Brussels. (http://ec.europa.eu/europeaid/infopoint/publications/europeaid/documents/budgetsupport08_en.pdf).
Frot, E., & Santiso, J. (2011). Herding in aid allocation. Kyklos, 64(1), 54–74.
Knack, S., & Rahman, A. (2007). Donor fragmentation and bureaucratic quality in aid recipients. Journal of Development Economics, 83(1), 176–197.
Knack, S., Rogers, F. H., & Eubank, N. (2010). Aid quality and donor rankings. Policy Research Working Paper 5290. The World Bank. Washington, DC: Development Research Group.
Nunnenkamp, P., & Thiele, R. (2013). Financing for development: The gap between words and deeds since Monterrey. Development Policy Review, 31(1), 75–98.
OECD (2005). The Paris Declaration on Aid Effectiveness and the Accra Agenda for Action (http://www.oecd.org/dataoecd/11/41/34428351.pdf; accessed: November 2011).
OECD (2007). 2006 survey on monitoring the Paris Declaration: Overview of the results. Paris.
OECD (2008). 2008 survey on monitoring the Paris Declaration: Making aid more effective by 2010. Paris.
OECD (2009). 2009 OECD report on division of labour: Addressing fragmentation and concetration of aid across countries. Paris.
OECD (2011a). Aid effectiveness 2005–10: Progress in implementing the Paris Declaration. Paris.
OECD (2011b). 2011 OECD report on division of labour: Addressing cross-country fragmentation of aid. Paris.
OECD (2011c). Evidence on trends in fragmentation and proliferation and implementation of results of division of labour processes. Key messages for HLF 4 and beyond. Paris.
Severino, J.-M., & Ray, O. (2010). The end of ODA (II): The birth of hypercollective action. Working Paper 218. Washington, DC: Center for Global Development.
Wood, B., Betts, J., Etta, F., Gayfer, J., Kabell, D., Ngwira, N., Sagasti, F., & Samaranayake, M. (2011). The evaluation of the Paris Declaration. Phase 2: Final report. Copenhagen: Danish Institute for International Studies (http://www.mofep.gov.gh/documents/ParisPhase2_web1.pdf; accessed: November 2011).
World Bank. (1998). Assessing Aid: What works, what doesn’t, and why. Oxford: Oxford University Press.
Acknowledgments
Financial support from the German Research Foundation (DFG GZ: DR 640/2-1) is gratefully acknowledged. We also thank two anonymous reviewers for helpful suggestions.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1: list of aid sectors included in the calculations of Theil indices and overlaps
Education (DAC/CRS code: 110), health (120), population programs (130), water supply and sanitation (140), government and civil society (150), other social infrastructure and services (160), transport and storage (210), communications (220), energy (230), banking and financial services (240), business and other services (250), agriculture, forestry, fishing (310), industry, mining, construction (320), trade policy and regulations (331), tourism (332), general environmental protection (410), other multissector (430), general budget support (510), developmental food aid, food security assistance (520), other commodity assistance (530), action related to debt (600), other emergency and distress relief (720), reconstruction relief (730), disaster prevention and preparedness (740)
Appendix 2: list of donors
Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, Switzerland, the United Kingdom, the United States, EU institutions, International Development Agency (multilateral donors in italics only included in baseline estimations)
Appendix 3
See Table 6.
Appendix 4
See Table 7.
Appendix 5: Alternative indicators of fragmented aid
As noted in the main body of the text, inequality and concentration measures such as Theil and Herfindahl indices are widely used in the academic literature on aid fragmentation.Footnote 32 It is open to debate, however, whether and under which circumstances these measures are most appropriate to reflect the transaction costs associated with relatively small amounts of aid spread across various recipient countries and sectors.Footnote 33 For instance, a rather moderate increase in the Theil index may blur a steep increase in transaction costs in the hypothetical case of donor j changing its aid allocation from scenario A to scenario B in Table 8.
Table 8 considers eight arbitrarily selected scenarios of donor j’s aid allocation across an again arbitrary number of ten recipient countries. Hence, the Theil index varies between zero and 2.303. The alternative measure, developed in OECD (2009; 2011b), refers to the share of non-significant bilateral aid relations and ranges from zero to one. Note that we assume an equal distribution of aid by the sum of all other donors in all scenarios A–H, and the global aid share of donor j is assumed to be 0.1 throughout. Under these conditions, the number of non-significant aid relations resembles the pattern of the Theil index only when assuming an extremely unequal or equal allocation across recipient countries (scenarios A and H, respectively). Deviations between both measures can be substantial between these extremes. Notably, scenario B of retaining one major recipient but spreading a small amount of aid across all other recipients results in a large number of non-significant aid relations, while the Theil index reacts only modestly to such a change. On the other hand, a change from allocation patterns as in scenarios F or G towards patterns as in scenarios B or C could be more relevant for our analysis given that the pre-Paris period was characterized by relatively high starting levels of the Theil index. While such a change would be associated with a considerably lower Theil index according to Table 8, one might argue that fragmentation did not really decrease. This would imply that our conclusion, based on the evidence for the Theil index, that fragmentation did not decline after the Paris Declaration would be rather conservative in the sense of understating the ‘true’ degree of fragmentation after the Paris Declaration. This would fit with the 2011 OECD report, where the general conclusion is that “aid is becoming increasingly fragmented”.
For comparative purposes, we also calculate the measure proposed by Birdsall and Kharas (2010). In contrast to the number of non-significant aid relations, Birdsall and Kharas consider the amount of donor j’s aid that is spent in non-significant relations. This measure also ranges from zero to one. Not surprisingly, the degree of fragmentation differs not only between the Theil index and the OECD’s count measure, but also between these two measures and the measure of Birdsall and Kharas.
About this article
Cite this article
Nunnenkamp, P., Öhler, H. & Thiele, R. Donor coordination and specialization: did the Paris Declaration make a difference?. Rev World Econ 149, 537–563 (2013). https://doi.org/10.1007/s10290-013-0157-2
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10290-013-0157-2