Is the aid architecture an οrderly process?

Paper by Louka T. Katseli, Paraskevi Boufounou and Maria D. Argyrou, National and Kapodistrian University of Athens.

Published in “Sul futuro delle politiche di cooperazione al sviluppo” (CeSPI Report, December 2021).

English title “Next Cooperation: On the Future of International Development Cooperation Policy” (CeSPI, 2022)

(p. 48 – 93)

Table of contents

1.     Introduction.. 3

2.     Overall Trends. 6

3.     Sectoral allocations. 11

3.1       What evidence?. 11

3.2       The cases of Italy and Portugal. 18

4.     Grants and loans. 21

5.     Determinants of ODA allocations. 26

5.1 A recipient country’s perspective. 26

5.1       A donor country’s perspective. 27

6.     Proliferation of donors. 29

6.1       The role of financial developments and colonial ties. 29

6.2       The role of governance. 31

7.     Regionally-Based  Delegated Cooperation: a proposal to streamline the aid architecture.. 34

Related Bibliography & References. 39

Appendix 1 – ODA measures. 41

Appendix 2 – Allocations in Italy and Portugal by region and income level.. 44

Appendix 3 – Definitions of the different types of financial aid provided.. 46

Appendix 4 – Econometric analysis. 47

A.         Econometric analysis of ODA by recipient countries. 47

B.         Econometric analysis of ODA by donor countries. 48

1.     Introduction

Official Development Assistance (ODA) is the official and concessional part of resource flows to developing countries. It constitutes of Bilateral and Multilateral flows[4].

As argued in an earlier paper by Daniel Cohen and Louka Katseli (2007)[5]: “one would like to think that the international aid architecture is an orderly process guided by a few simple principles”. For instance, one would expect that when action across donors needs to be coordinated, Multilaterals would take the lead; when instead projects need to be single-handedly managed, Bilaterals would step in. One may also think that Bilaterals allocate their development assistance according to their own comparative advantage or their national interest: food exporting countries would support food aid; ex colonial powers channel their assistance mainly to their colonies; those that are leaders in health service provision allocate a larger share of their development assistance to build effective health and/or nursing delivery systems in partner countries and so on…”

The analysis provided in this paper suggests that the aid architecture is not as orderly a process as one would like to think, especially if considered from a recipient country’s perspective. However, the patterns that have recently emerged in ODA allocations and practices, as a consequence of global shocks and developments, are expected to continue and provide the basis for policy recommendations that could improve ownership and effectiveness of development assistance.

To investigate the patterns of ODA disbursements across Multinationals and Bilaterals in recent years we use the sectoral specialisation index developed by Cohen and Katseli (2007) and apply it to both Bilateral and Multilateral ODA allocations in 2007 and 2019. The analysis demonstrates that Multilateral donors tend to be much more specialized than Bilateral ones who appear more “political” in their approach to ODA disbursements. As expected, allocation patterns have been influenced by institutional competences in the context of global developments. Thus, after the eruption of the financial crisis in 2007, Multilaterals have taken the lead in “Social Infrastructure & Services”, “Economic Infrastructure & Services” and “Production Sectors”, extending mainly loans to partner countries to finance long-term investments. Bilateral donors on the other hand, appear to be more specialized in “Actions relating to Debt” and “Humanitarian Aid” as they probably extended assistance to partner countries for the latter to be able to repay past loans or implement projects which would be popular and draw support in their own domestic constituencies.

Following the signing of the UN Millennium Development Goals (MDGs) in 2000 and the Sustainable Development Goals (SDGs) in 2015, Bilaterals and Multilaterals have increasingly shared presence in a number of social sectors, most notably health and education.

The loan/grant theoretical debate casts some more light on the observed patterns. The early literature on this issue (Rodrik, 1995, Bulow and Rogoff, 1992) has argued that Multilaterals are better at making loans to the poorest countries than Bilaterals. The argument has been that Multilaterals are less politicised; they obey rules which are better suited for making loans and being repaid and they profit from superior negotiation and enforcement capacity. This is why Multilateral lending is better able to crowd in private capital than Bilateral loans, which are too politicized to generate positive externalities. The brief review provided in section 3 supports these arguments. Even though the agenda of the Multilaterals is often driven by key shareholders which are none other than the Bilaterals, Multilaterals have tended to behave more like bureaucracies that operate according to rules while Bilaterals tend to be more politicized and practicing discretion. Although more intuitively appealing than the logic of comparative advantage driven by sectoral or country preferences, this new divide is partially substantiated by empirical evidence.

The proliferation of donors and financial instruments in the international scene appears to have segmented further the aid-delivery system in recent years and to have made it less orderly. By looking at the evidence for the year 2019, we document that more than 54 donors are operating, on average, in each recipient country. The corresponding number in 2004 was 23.4 (Cohen and Katseli, 2007) This number falls dramatically to less than 3.7, when we consider only those donors which supply more than 0.5% of the recipient country’s GNI; it falls even further, to 2 donors per recipient country, when we limit the analysis to donors which supply more than 1% of the recipient’s GNI. As shown by the empirical analysis provided, the number of donors on the ground varies with respect to recipients’ specific characteristics, such as size, quality of governance etc.

The evidence of a limited specialisation and unclear allocation of roles between Multilaterals and Bilaterals, as well as the increasing proliferation of donors suggest that the main policy challenge is to reduce transaction costs on the ground and improve coordination and implementation capacities in recipient countries. With this aim in mind, we propose to streamline and increase the effectiveness of ODA allocations through a regionally-based delegated cooperation system, ie. by channelling assistance mainly through regional IFIs and delegating competences to independent regional capacity building organizations, such as the African Capacity Building Foundation (ACBF) to mature projects, coordinate stakeholders on the ground and oversee project- implementation processes, proper monitoring and evaluation of outcomes.

2.     Overall Trends

Since 1970, the rate of change of both Bilateral and Multilateral Assistance has been positive with its share of global GNI decreasing by approximately 10 percentage points. Overall, ODA disbursements rose from 6,949 million current USD in 1970 to 161,027 million current USD in 2020 ie from 0.26% of global GNI to 0.18%. ODA rose fast especially after 2000, probably as a consequence of the signing of the Millennium Development Goals. As can be seen in Figure 1 and Table 1, between 2000 and 2020, Bilateral Assistance increased by 120% and Multilateral Assistance by 87%, despite a short-lived decrease between 2004-2006.

Figure 1– ODA composition 1970-2020 (2019 USD millions – disbursements)

Source: OECD Stats [DAC1], 2021

The share of Bilaterals in total ODA has fluctuated since the 1970s between 60-70% (Figure 2). With the recent doubling of ODA, this share is expected to increase even further.

Figure 2 – Shares of ODA by different categories of donors % of total ODA (disbursements)

Source: OECD Stats [DAC1], 2021

As can be seen in Table 1, the majority of ODA over the past two decades has come as expected from Bilateral donors. Bilaterals have contributed on average almost 2/3 of total ODA, while Multilaterals have provided a little more than ¼; UN agencies have accounted around 5% (± 1%) of total ODA, and Regional Development Banks for 2.5%.

It should be kept in mind however, that DAC statistics tend to overestimate the importance of Bilaterals in ODA disbursements to the extent that they count as Bilateral that part of aid which is channelled via Multilaterals to specific countries and projects[6].Given that this practice enjoys widespread acceptance, we use DAC definitions and ODA statistics for the remainder of the paper.

Table 1 – Shares of ODA by different categories of donors % of total ODA

Source: OECD Stats [DAC1], 2021 Own calculations

It is interesting to note that after the crisis, the shares of both Bilaterals and Multilaterals in ODA disbursements increased, while the share of UN disbursements declined. As expected, disbursements by multilaterals rose faster in the post crisis period as compared to the pre-crisis one (44% against 20%) while disbursements by Bilaterals appear to have increased at a slower pace (30% against 51%).

Overviewing ODA from the recipients’ perspective, as shown in Figure 3, it is evident that since 2000, the share of net ODA received by Middle Income Countries (MICs) has sharply decreased from 58% of ODA in 2000 (32,089 USD million) to 38% in 2019 (64,365 USD million). The share of ODA to BRICS since 2003 has remained steady, around 5%. Low Income Countries (LICs) have seen a steep increase in both share and volume of ODA received (from 12% and 6,784 USD million in 2000 to 30% and 49,875 USD million in 2019 respectively). It appears therefore that following the signing of the MDGs, a reallocation of ODA has in fact taken place in favour of low-income countries with the aim to eradicate poverty and improve standards of living.

As shown in Figure 4, ODA per capita rose sharply after 2000, especially for the LICs. The rise was much slower for BRICs which experienced even a sudden decrease after 2015.Despite the reallocation of ODA towards the LICS, middle-income countries still receive the largest share of ODA relative to the LICs as evidenced in Figure 5.

Figure 3 – Net ODA and Official Aid Received for selected regions
1971 – 2019 (% of total ODA)

Source: The World Bank, World Development Indicators, 2021

Figure 4 – Volume of ODA received for selected regions 1971-2019
(current USD million)

Source: The World Bank, World Development Indicators, 2021

Figure 5 – Volume of ODA per capita received for selected regions 1971-2019 (current USD)

Source: The World Bank, World Development Indicators, 2021

3.     Sectoral allocations

To investigate the pattern of specialisation of Bilaterals and Multilaterals in particular sectors of economic activity and whether this pattern has been affected by global shocks such as the financial crisis of 2007, a sectoral specialisation index has been constructed covering both Bilateral and Multilateral ODA allocations.

3.1  What evidence?

The specialisation index originally developed by Cohen and Katseli (2007) measures each donor’s – Bilateral (b) or Multilateral (m) – contribution to a particular sector (s) as a fraction of the share of the sector in total ODA.

Given its construction, values of the index that exceed unity indicate relative specialisation; the higher the value of the index, the greater the relative specialisation of each donor in the particular sector.

To analyse potential changes in sectoral specialisation following the eruption of the financial crisis, changes in the specialization index for 2007 and 2019 are presented.

As presented in Table 2, Multilateral ODA disbursements in 2019 are mainly channeled in Production Sectors (1.47), Economic Infrastructure and Services (1.22), and Social Infrastructure and Services (1.18). This pattern of specialisation is robust over time with specialization in these sectors remaining prevalent. Such pattern is consistent with the principle that Multilaterlas are better suited to finance long-term investments mainly through debt-financing.

On the other hand, for the same period of reference, Bilaterals appear to specialize in Actions Relating to Debt and Humanitarian Aid. Following the financial crisis of 2007, a significant increase of the specialisation index related to Humanitarian Aid is recorded accompanied by a corresponding decrease in Actions Relating to Debt. The financial crisis thus appears to have changed the overall sectoral distribution of total ODA. Table 2 reveals that between 2007 and 2019, there was a significant increase in ODA directed to Humanitarian Aid by 8 percentage points. In the same period, the share of ODA channeled to Action Relating to Debt and to Social Infrastructure and Services decreased by 7 and 6 percentage points respectively.

Table 2 – Sectoral Specialisation by ODA main categories

Source: Own calculations using OECD Stats [DAC5], 2021

Decomposing further the sectoral specialization indices under each category to detect the direction of ODA allocations and the areas of specialization of Multilaterals and Bilaterals, the patterns that emerge are the following:

Under Social Infrastructure & Services which accounted for 37% of total ODA in 2019, four sectors, namely Government and Civil Society, Health, Education and Water Supply and Sanitation have accounted for the largest share of total ODA disbursements (30%). Multilaterals seem to have concentrated their efforts consistently in these areas, thus being more specialized than Bilaterals in their orientation. Instead, Bilaterals do not seem to specialize in any single area, with the exception perhaps of post-secondary education.

Under Economic Infrastructure and Services, which accounted for 17% of total ODA in 2019 as opposed to 15% before the crisis, the same pattern emerges: Multilaterals appear to concentrate their disbursements in a few areas as opposed to Bilaterals who do not exhibit any detectable pattern of specialization. While the sectors in which Multilateral donors appear to be specialized continue to be Transport & Storage, Communications and Energy, it is not surprising to find out that, following the financial crisis, a reallocation took place away from these more traditional sectors of specialization towards Banking and Financial Services.

Throughout the period under investigation, the Production Sectors received only 7-8% of total ODA. Agriculture, Fishing, Industry, Trade Policies and Regulations and Tourism seem to have remained preferred areas of specialization for Multilaterals in 2019 as opposed to the Mineral Resources and Mining sector which seems to have lost its attractiveness for Multilaterals. It is interesting to note that Bilaterals stepped in and reallocated disbursements towards that sector, with the index of sectoral specialization of Bilaterals increasing from 0.60 to 1.29 between 2007 and 2019. Apart from Mineral Resources and Mining, Bilaterals seem to specialize in construction services.

Table 3 also reveals that Bilaterals are more specialized than Multilaterals in “Action Relating to Debt” and in “Humanitarian Aid”, especially in “Emergency Response”. Multilaterals instead seem to specialize in “Disaster Prevention and Preparedness”. It  should also be noted that the share of ODA directed to “Humanitarian Aid” has increased by 9 percentage points between 2007 and 2019 (or almost 150%) mainly driven by the sharp increase in “Emergency Response” (from 6% to 14% of total ODA) This is probably  related to the climate -change crisis and the need to respond collectively to its adverse consequences world-wide as well as to conform to  commitments made under  the new policy agenda set by SDGs.

The indices presented in Table 3, confirms the following conclusions:

  • Humanitarian Aid and ODA disbursements towards Government and Civil Society Support, most notably general budget support, absorbed 20% of total ODA in 2007 and 27% of total ODA in 2019; Education and Health made up an additional 14%of total ODA disbursements.
  • While, Multilaterals and Bilaterals have been equally present in a number of sectors, the former are becoming increasingly more specialized in specific sectors.
  • By 2019, Multilaterals appear to be specialized relative to Bilaterals in a number of sectors under Social Infrastructure and Services (Education, Health, Water Supply and Sanitation), Economic Infrastructure and Services (Communications, Banking and Financial Services, Energy) and in almost all Production Sectors with the exception of Mineral Resources and Construction.
  • Multilaterals thus appear to have channeled their ODA disbursements to long-term infrastructure projects which could be debt-financed, while the post 2007 financial crisis seems to have prompted them to become more specialized in Banking and Financial Services and Actions Related to Debt. During the same period and possibly as a consequence of commitments towards the implementation of MDGs -especially the three goals related to Health (Goals 4,5 and 6)-, Multilaterals enhanced their specialization in Health.
  • Bilaterals seem to remain instead much less specialized across infrastructure or productive sectors while channeling their ODA disbursements increasingly towards Humanitarian Aid, Commodity and Development Food Assistance, Action Relating to Debt as well as Mineral Resources and Construction. These are sectors which are usually serviced by the NGO community of donor countries or donor-country companies.
  • Bilaterals appear to be more “politicized” than Multilaterals in deciding their sectoral ODA allocations catering to their own domestic constituencies’ interests, namely to local companies and banks in promoting natural resource extraction, the delivery of emergency assistance or debt cancellation in partner countries[7]. This conclusion confirms recent evidence by other authors (Gulrajani, 2016).

Table 3 – Sectoral specialization of ODA by Donor, 2007 & 2019

Source: Own calculations using OECD Stats [DAC5], 2021   

3.2  The cases of Italy and Portugal

Comparing ODA allocations by sector for Italy and Portugal, two similar (South European countries and ex colonial powers) countries such as Italy and Portugal, we can draw the following conclusions (see Table 4):

  • The two countries exhibit different trends as the financial crisis that hit Portugal the hardest seems to have affected ODA disbursements: Italy increased its ODA disbursements from 1,465 million USD in 2007 to 1,609 million USD in 2019 (an increase of 9.8%); Portugal instead reduced almost by half (by 45,8%) the ODA provided, from 277 million USD in 2007 to 150 million USD in 2019.
  • Portugal is strongly specialized in “Social Infrastructure and Services” with over 70% of its ODA provided to this sector. The same sector is the second most important in Italian ODA’s allocations. In particular, it counted for 22% of the country’s ODA disbursements in 2007 (second highest after “Action Relating to Debt”) and for 32% in 2019 (second highest after “Unallocated / Unspecified”). In absolute terms, the amount directed to “Social Infrastructure and Services” in Italy increased by 59.4% (from 325 million USD in 2007 to 518 million USD in 2019), while in Portugal it decreased by 47.8% (from 203 million USD in 2007 to 106 million USD in 2019), following the countries’ general ODA trends.
  • In 2007, 40% of Italian ODA was directed to “Action Related to Debt”. However, for 2019 the same percentage is almost 0. On the other hand, Portugal’s ODA directed to “Action Related to Debt” was close to 0 already since 2007, while in 2019 the country did not provide any amount for this purpose.

Table 4 – Sectoral allocation of ODA for Italy and Portugal, 2007 & 2019

Source: Own calculations using OECD Stats [DAC5], 2021   

4.     Grants and loans

While the ODA’s operationalization evolved over the decades, with minor and major changes in its definition and measurement, global economic conditions led to its modernization in 2019. One of the most important weaknesses of ODA measurement till then were the criteria and terms a loan needed to meet to qualify as ODA in a world of low interest rates. (Roodman, 2014). Hence, since 2019, the new ODA measuring system (grant equivalent) makes a clear distinction between grants and loans (see Appendix 1).

While Multilaterals appear to have become more specialised than Bilaterals over the last decade, there is still a large number of sectors where Multilaterals and Bilaterals share presence. One dimension that has changed dramatically, however, has to do with the way development finance is being provided.

Box 1 in Appendix 3 presents the definitions of the different types of financial assistance provided.

As presented in Figure 6 and Figure 7, debt finance by bilateral donors has been more than halved between 1970 and 2018 with bilateral commitments replaced by multilateral ones. More specifically, debt extended to both LICs’ and MICs’ has become predominantly Multilateral since 2005. In particular, between 1970 and 2005 Bilaterals’ share of outstanding debt to LICs was almost halved (from 70% to 38%), while Multilaterals’ share more than tripled (from 11% to 57%). During the same period, Bilaterals’ share of outstanding debt towards the MICs decreased from 66.4% to 24.4% while the Multilaterals’ share increased from 21% to 25%.

Figure 6 – Share of LICs debt % of PPG, total (DOD)

Source: World Bank, International Debt Statistics, 2021

Figure 7 –Share of MICs debt % of PPG, total (DOD)

Source: World Bank, International Debt Statistics, 2021

As presented in Figures 8 – 11, during the period 1996-2008, i.e. before the financial crisis, NFL Bilateral Loans (concessional or not) were negative towards all countries including towards the MICs and BRICs. For LICs, NFL Bilateral Loans were   negative for the period 1999-2006. In general, a slowdown in all types of assistance extended to LICs, MICs and BRICs took place by both Multilaterals and Bilaterals in the decade preceding the financial crisis. Following the financial crisis, loan commitments by both Bilaterals and Multilaterals increased rapidly, especially towards LICs. As expected, Net Financial Flows turned positive with the exception of the BRICs.

Figure 8 – World: Grants and Loans, USD million, 1978-2019

Source: World Bank, International Debt Statistics, 2021

Figure 9 – BRICS: Grants and Loans, USD million, 1978-2019

Source: World Bank, International Debt Statistics, 2021

Figure 10 – LICs: Grants and Loans, USD million, 1978-2019

Source: World Bank, International Debt Statistics, 2021

Figure 11 – MICs: Grants and Loans, USD million, 1978-2019

Source: World Bank, International Debt Statistics, 2021

The above evidence is consistent with the conventional wisdom that Multilaterals do have a comparative advantage in lending (Bulow and Rogoff,1992). According to Rodrik (1995), Multilateral Development Banks (MDBs) enjoy superior negotiation capacity due to political legitimacy and are more neutral relative to Bilaterals who are ridden with the legacy of Cold War and colonialism. While MDBs’ chief aim is to provide poor countries access to stable private lending (Rodrik, 1995), bilateral lending is widely perceived as disconnected from private creditors’ interests, ridden as it is by political considerations and visibility concerns (Alesina and Dollar, 2000).

A different argument advanced by Banerjee and He (2003), has to do with MBDs’ concentration of claims which facilitates debt restructuring and default management, and thus the prevention of debt crises. Bilateral creditors, even when they coordinate their commitments under the auspices of the Paris Club, have generally been too slow in restructuring outstanding debt, thus contributing to severe recessions (Cohen and Katseli, 2007).

Multilateral comparative advantage in lending is not as obvious as held by the proponents of conventional wisdom. First, there is no empirical evidence that Multilateral net loans crowd in private net loans. Furthermore, Multilaterals resort more systematically than Bilateral agencies or private creditors to ‘defensive lending’ so as to help countries service their debt. As shown by Cohen, Jacquet and Reisen (2006), when the debt service of a given country between 1980-2004 increased by one percentage point of GDP, private gross loans increased by 0.03 percentage points of GDP and Bilateral gross loans increased by three times that amount (0.09 percentage points of GDP); Multilateral gross loans, however, increased by three times the latter amount (0.30 points of GDP).

Secondly, as demonstrated by the failed debate over the Sovereign Debt Restructuring Mechanism (SDRM) launched by Anne Krueger at the IMF in 2002, institutionalizing restructuring mechanisms remains difficult for Multilateral creditors.

On the other hand, Bilateral donors have problems of their own. The debate on the “free-riding” behavior of new lenders such as China is not settled. If rich countries make grants and emerging donors make loans, a resolution mechanism will need to be found in case of debt crises, which is not yet obvious.

It thus appears that the loan/grant debate is not the critical dividing line for thinking of a new division of labour between Multilateral and Bilateral donors. New innovative lending strategies must be thought of, which merge grants and loans.

On the other hand, the evidence presented in the previous section on patterns of specialization and sectoral allocations confirms the hypothesis that Multilaterals do in fact perform more as bureaucracies driven by international commitments and rules, as opposed to Bilaterals who exercise more discretion in their aid and are more politicized relative to the former.

Taking a view from the ground, seems to support this conclusion.

5.     Determinants of ODA allocations

In this section, the role that specific exogenous characteristics of recipient countries (size, income, conflict, governance- IRAI[8]) play in determining the amount of ODA per capita received is explored. The econometric results, providing for heteroskedasticity and multicollinearity, are presented in Appendix 4.

5.1 A recipient country’s perspective

Annual data for the period 2005-2019 from 81 recipient countries are examined (due to data limitations) and separate analyses are conducted for the period before (2005-2007) and after the crisis (2008-2019). ODA per capita received in current USD (in log) is used as the dependent variable. The explanatory power (R2) of the model estimated for the above mentioned three periods is satisfactory. All independent variables are statistically significant and have the expected sign. The only exception is the sign of “conflict”, which is negative for the period before the crisis, but not statistically significant.

The main findings of the econometric analysis are the following:

  • Poverty of recipient country (log of GDP per capita in PPP terms): The results of the analysis indicate a strong negative relationship between the income per capita of the recipient country and the amount of ODA per capita it receives; in other words, the poorer the country, the more ODA it receives. The results are similar in all three periods examined.
  • Governance (IDA Resource Allocation Index – IRAI): Taking the IDA Resource Allocation Index (IRAI) as a measure of quality of governance[9] ,we find that the better the governance of the recipient country, the more ODA per capita it receives.
  • Size of recipient country (log of population): One of the striking findings of the analysis is that the smaller a country, the more ODA it receives in per capita terms! This surprising finding is possibly the result of geopolitics: a country is a country, whatever its size. This is consistent with the view presented in Alesina and Dollar (2000) according to which equal weighting of votes at the UN is a driver of ODA allocations. To put it more simply, development assistance at least for Bilateral donors remains an inter-governmental process.
  • Conflict (Dummy variable): As mentioned above, its sign is negative only for the period before the crisis, but it is not statistically significant. It turned positive and significant after the crisis. The same holds true for the whole examining period (2005-2019). This means that the presence of conflicts, tends to increase ODA per capita allocations

In conclusion, while the size of a country is negatively related to ODA per capita disbursed, poverty, good governance and conflict are positively related.

5.1  A donor country’s perspective

In this section, we investigate whether the characteristics of the donor country determine the amount of ODA per capita it provides. Annual data for the period 2005-2019 from 49 donor countries are studied taking into account the data limitations. Separate analyses were also conducted for the period before (2005-2007) and after the crisis (2008-2019). ODA per capita provided (current USD) in log terms was used as the dependent variable. Looking at ODA allocations from the perspective of the donor, a number of considerations seem to guide allocation practices. Following the methodology developed by Helen Milner (2005), a two-variables model has been used which is presented in Appendix 4. The explanatory power (R2) of the model estimated for the three periods is satisfactory and all independent variables are statistically significant and have the expected sign. The econometric results provide for heteroskedasticity and multicollinearity.

The conclusions of the econometric analysis can be summarised as follows:

  • Size of the donor country (log of population): Our results suggest that the bigger the country, the higher the ODA per capita it provides, with the result being robust over time.
  • Income per capita (log of GDP per capita in PPP terms): Similarly, our econometric analysis suggests that there is a positive and statistically significant relationship between income per capita of the donor country and the amount of ODA it provides in all examined periods.

The analysis supports the argument that the larger the size and the income per capita of the donor country, the more ODA per capita it provides.

6.     Proliferation of donors

To understand further the workings and the ensuing challenges of the aid-architecture system, we identify the number of donors in each country and the variation recorded in relation to recipients’ specific characteristics.

We focus on three categories of donors in any recipient country:

  1. all donors active in the country (total number);
  2. those who supply ODA equivalent to 0.5% of the recipient country GNI and
  3. those supplying at least 1% of GNI.

One would expect that the presence of donors in any given country is affected by economic and financial developments and prospects, political ties and geopolitical considerations and improvements in governance in the recipient country over time. The increasing globalisation of economic activity, the proliferation of financial instruments and financial stakeholders and the emergence of new actors and innovative modes of finance, both public and private (ODI et. al., 2015) are expected to lead to an increase in the average number of donors in any given country. Furthermore, ex colonial powers are expected to continue to have a relatively strong, even if diminishing, presence in their ex colonies, reflected, inter alia, in the extension of development assistance. Last but not least, improvements in governance are expected to attract a larger number of donors in any given recipient country The findings presented in the analysis below confirm these hypotheses.

6.1  The role of financial developments and colonial ties

Table 5 presents the number of Bilateral and Multilateral donors in each of the three categories by splitting the sample of 81 recipient countries according to their colonial past (88% of the countries in our sample are former colonies) for years 2004, 2007 and 2019, so as to highlight possible changes before and after the crisis.

What is striking in Table 5 is the proliferation of donors present in each recipient country over time! The number of donors present, both Bilaterals and Multilaterals, increased on average from 23 in 2004 to 41 in 2007 and 54 in 2019! The average number of donors present has more than doubled between 2004 and 2019 in all countries, regardless of their colonial or non-colonial past. The average number of Bilateral donors present in a recipient country is consistently higher than the number of Multilateral donors in all cases, rapidly increasing over time from 14 (13.6) donors in 2004, to 25 (24.9) in 2007, to 35 (34.7) in 2019 as an increasing number of bilateral donors extended their presence in countries with which they had no colonial ties (25 additional bilateral donors present in countries with no colonial past of recipient countries as opposed to 19 additional bilateral donors  in recipient countries with a colonial past). Table 5 also confirms the expected hypothesis that the number of Multilateral donors proliferated after the financial crisis, with their share increasing from 31.4% in 2007 to 49.2% in 2019.

The rapid proliferation of donors in any given country highlights the pressing challenges of ownership, coordination and effectiveness in an aid-architecture system that appears increasingly more segmented and unorderly…

Such needs become even more evident if one notes the dramatic decline in the number of donors who provide substantial amount of assistance in any given recipient country over time… As we move from the broad to more restrictive definitions, the number of donors who supply ODA equivalent to 0.5% of the recipient country GNI falls from 54 to 4 (3.7) in 2019 and to 2 if one limits the number of donors only to those who supply ODA equivalent to 1% of the recipient country’s GNI. Interestingly, at the 1% threshold, the ex-colonies do not seem to discourage other donors’ presence: there is approximately one more donor present in that case relative to the number present in countries with no colonial past.

Table 5 – Number of Donors by Colonial Past

Source: Own calculations using OECD Stats [DAC5], 2021

6.2  The role of governance

Due to the fact that data for IRAI that captures the governance effect (see footnote 8) start in 2005 we limit the analysis only to the post-crisis period, through a comparison of years 2007 and 2019.

Focusing on the 1% threshold of donors, we find that in 2007 the number of donors for small countries increased in countries with better governance; in the post-crisis period, the quality of governance does not seem to be a determining factor in the average number of donors present in any given recipient country. The number of donors present is in fact larger in the case of countries with bad as opposed to medium quality of governance and is larger for small as opposed to medium-size countries. (Table 6). Despite their enhanced needs and absorptive capacities, the larger recipient countries do not attract more donors, confirming the one country-one vote hypothesis expressed in the previous section.

Table 6 – Average Number of Donors (1% threshold) by recipient’s Size and Governance Quality

Source: Own calculations using OECD Stats [DAC5], 2021

When we now categorise recipient countries by income group as opposed to size (Table 7) we verify that before and after the crisis lower- income countries seem to have attracted a higher number of donors, with the number of donors increasing as governance improves.

Table 7 – Average Number of Donors (1% threshold) by Recipient’s Size and Income

Source: Own calculations using OECD Stats [DAC5], 2021

7.     Regionally-Based  Delegated Cooperation: a proposal to streamline the aid architecture

The evidence provided so far indicates that the aid architecture remains unorderly and entails high transaction costs for all recipient countries, even for those who have improved their governance and managed to channel more effectively development finance to high-priority projects and programmes.

More than sixteen years ago, on March 2, 2005, Ministers from both developed and developing countries signed the Paris Declaration Agreement, agreeing to increase alignment of aid with partner countries priorities and systems, to eliminate duplication of efforts, to rationalise donor activities and to reform and simplify donor policies and procedures, so as to encourage “collaborative behaviour”. The Millennium Development Goals signed in 2000 provided a useful framework for common priorities and action.

There is no question that much has changed over the past two decades. Many emerging economies and developing countries have experienced high rates of growth and improvements in their standards of living. Extreme poverty has been reduced and inter-country inequalities have been reduced. As globalization of commodity, services’ and capital markets has deepened and new investment opportunities have arisen, the composition of development finance has changed dramatically.  Domestic public   and international private flows have increased rapidly replacing ‘traditional sources of development finance such as ODA even for LICs

(ODI et al., 2015) Multilateral donors have proliferated and extended their presence in all continents. Innovative financial instruments have been created by public and/ or private stakeholders to meet the dual needs of mitigating risks and securing adequate returns. (Ibid, Chapter 3, pp. 89-127)  

The eruption of successive global shocks however during the preceding two decades- associated with the financial crisis, the climatic-change crisis, the refugee crisis and the pandemic crisis- and their devastating effects for millions of people across many countries, has brought to the surface the need for global governance reforms that would promote resilience of people and systems and underpin sustainable development. The pressing needs to address rising poverty and inequalities, secure a fair transition to meet environmental challenges, develop skills and capacities to adapt to technological changes, manage unprecedented refugee and migration flows and increase the effectiveness of public health and education systems can only be met if development assistance is properly and efficiently channelled towards the implementation of carefully selected projects by both donors and recipient countries. As private market participants play an increasingly important role especially in financial activities, a new partnership needs to be built, across all stakeholders, both public and private, aiming at Improved ownership, better coordination, lower transaction costs and proven capacity for project implementation.

Implementing the UN Sustainable Development Goals (SDGs) by 2030, to which 193 countries have committed, offers a credible and mutually-agreed framework for collective action. The so-called Agenda 2030, can encompass a broad set of projects and programmes that aim at economic transformation, social cohesion, environmental protection and improved governance.  

It is important for continental Organizations and bodies, such as the African Union or for that matter the European Union, in close collaboration with their members and technical support by their organs and specialized agencies, to work out a realistic road map for the effective implementation of Agenda 2030 at the country and regional levels.

The African Union’s Agenda 2063 for example, which is Africa’s strategic framework for inclusive and sustainable development, needs to be more closely aligned with Agenda 2030 and translated into concrete regional projects with detailed specifications of financial, regulatory, and capacity building actions as well as a clear delegation of managerial responsibilities. The AU’s 15 flagship programmes, that include inter alia the implementation of an integrated high-speed train network, the completion of the Grand Inga Dam Project, the formulation of an African Commodities Strategy and the establishment of an African Continental Free Trade Area. need to be matured and translated into concrete collective actions and sub-programmes that involve all of Africa’s regional bodies as well as the AU member states.

At a time when resources are limited, it is important to avoid proliferation of bodies and duplication of efforts. This is especially relevant for financial institutions where experience has shown that economies of scale enhance the capacity for resource mobilisation. The announced creation for example of a system of African Continental Financial Institutions – one of the flagship programmes of the AU- that aims to mobilise resources for the promotion of economic integration and development, provides for the establishment of an African Investment Bank, a Pan African Stock Exchange, an African Monetary Fund and an African Central Bank. The European experience and the challenges presented to set up an effective European financial architecture system can serve as a useful guide for the proper sequencing of decisions and actions. More importantly, the advantages derived from the creation of an African Investment Bank (AIB), analogous to the European Investment Bank, need to be evaluated against the costs of potential competition between the AIB and the African Development Bank (AfDB) in the continent.

Before launching the AIB project, the future delegated competencies and role of the AIB and the AfDB need to be clarified. How can the AU’s Agenda 2063 be aligned with priorities set up by the African Development Bank’s Strategy for 2013-2022? How can the African Development Bank and its African Development Fund be empowered to mobilise further resources, including ODA, from both bilateral and multilateral donors, to implement its strategy? What are the needs for capacity building across member states of the AU to make projects implementable and what is the role of specialised agencies of the AU such as the African Capacity Building Foundation or of the AFDB’s African Development Institute? How are priorities aligned across regional organisations and how are synergies developed across specialised institutions and multilateral donors so as to avoid duplication of efforts and high transaction costs on the ground? In general, how can the development – finance architecture, including the aid architecture, become more orderly and effective?

The need for greater coherence in regional economic and financial architecture is evident in all continents including Africa, Asia, Latin America and Europe. In view of the increasing integration of regional markets, increased “regionalisation” of development finance, including development assistance, can lower transaction costs at the individual country level, improve coordination, mobilize resources more effectively and avoid costly duplication of efforts. This could be promoted through delegated-cooperation between regional stakeholders to streamline finance, including ODA, in an orderly and efficient manner.

Delegated cooperation is defined by OECD/DAC as a working arrangement whereby “one donor (or a “lead donor”) acts with authority on behalf of one or more other donors, the so called “delegating” donors or “silent partners” (see Cohen and Katseli, 2007). The level and form of delegation may vary, ranging from responsibility for one element of the project cycle for a specific project to a complete sector programme or even a country programme. The “delegated cooperation principle” has in fact been used by the Nordic Plus countries to improve aid effectiveness (MFA, Norway, 2006).

The same principle can in fact be used at the regional level to help reduce the high transaction costs associated with efforts to coordinate a large number of donors at the country level and create incentives for greater division of labour among donors.

In Africa for example, delegated cooperation agreements could be structured to promote the implementation of Agenda 2063 in Africa’s geographic regions, ie in East Africa, Southern Africa, West Africa, Central Africa and North Africa. For each region, a financial institution, such as the AFDB, or a multilateral agency, such as the UNDP, supported by a Stakeholders’ Council and in close cooperation with the countries involved, can take a leading role in coordinating the mobilisation of resources and their channelling to selected projects and programmes, in building the needed capacities and in monitoring and evaluating the implementation process. The formation of Sectoral Councils (in education, infrastructures, health, etc) could provide technical assistance and highlight best- practices world-wide.

The above proposal derives from the key findings of our research regarding the need to enhance the effectiveness of ODA, taking into consideration the increased role of Multilaterals and private market participants in development finance and the differences in allocation principles between Multilaterals and Bilaterals. Multilaterals appear more willing to provide budget support and abide by rules and agreed criteria than Bilaterals, whose actions are more driven by political considerations, including domestic, public opinion, foreign policy and intergovernmental priorities. This might be one of the reasons why according to Gulrajani (2016) aid recipients in fact seem to prefer multilateral to bilateral channels of assistance with the former being more efficient than the latter.

Differences between countries and regions, time periods, aid objectives, and individual donor organizations are all factors that influence the effectiveness of aid whether delivered bilaterally or multilaterally (Biscaye, Reynolds and Anderson, 2017). However, recent evidence (Ezeaku,, 2019) suggests that whereas bilateral concessional debts induced economic growth to decline, multilateral concessional debts had a positive impact on growth. It is highly likely therefore, that regionally-based coordinated delegation led by multilaterals could in fact spur growth

Our proposal builds upon the conviction that a dramatic reduction of transaction costs is feasible and could raise the efficiency of the overall system. By suggesting a partnership scheme based on a regionally-based delegated cooperation system, we believe that not only regional integration would be promoted but more resources would be mobilised, governance would be simplified and coordination would be vastly improved. Furthermore, the business sector, NGOS and knowledge- organizations could gain a new status as full partners in development especially in sectors where there exists expertise or they have a demonstrable comparative advantage, e.g. in health, in education or humanitarian assistance.

Related Bibliography & References

Alesina, A., & Dollar, D. (2000). Who gives foreign aid to whom and why?. Journal of Economic Growth5(1), 33-63.

Arndt, C., & Oman, C. (2010). Uses and Abuses of Governance Indicators. Paris, France: Development Centre Studies.

Banerjee, A. V., & He, R. (2003). The World Bank of the future. American Economic Review93(2), 39-44.

Biscaye, P. E., Reynolds, T. W., & Anderson, C. L. (2017). Relative Effectiveness of Bilateral and Multilateral Aid on Development Outcomes. Review of Development Economics21(4), 1425-1447.

Brito, J. A. (2015). Defining Country Size: A descriptive Analysis of Small and Large States.

Bulow, J., Rogoff, K., Bevilaqua, A. S., Collins, S., & Bruno, M. (1992). Official Creditor seniority and burden-sharing in the former Soviet bloc. Brookings Papers on Economic Activity1992(1), 195-234.

Caballero, R. J. (2003). The Future of the IMF. American Economic Review93(2), 31-38.

Cohen, D., Jacquet, P., & Reisen, H. (2006). Beyond “Grants versus Loans”: How to Use Debt for Development. Financing Development: What are the Challenges in Expanding Aid Flows?.

Cohen, D., & Katseli, L. (2007). Multilaterals and Bilaterals on the Ground; Division of Labour or Coordination?. Paris: OECD Development Centre.

Ezeaku, H. C., Egbo, O. P., Nwakoby, I., & Onwumere, J. U. (2019). Effectiveness of Bilateral and Multilateral Concessional Debts on Economic Growth in Africa. International Journal of Emerging Markets.

Frot, E., & Santiso, J. (2009a). Crushed Aid: Fragmentation in Sectoral Aid. SITE Working Paper No. 6.

Frot, E., & Santiso, J. (2009b). Herding in Aid Allocation. OECD Working Paper No. 279.

Gulrajani, N. (2016). Bilateral versus Multilateral Aid Channels – Strategic Choices for Donors. Overseas Development Institute, London.

Hynes, W., & Scott, S. (2013). The Evolution of Official Development Assistance: Achievements, Criticisms and a Way Forward. European Center for Development Policy Management.

Marchesi, S., & Missale, A. (2004). What Motivates Lending and Aid to the HIPCs?. Centro Studi Luca d’agliano, (189).

Milner, H. V. (2006). Why Multilateralism? Foreign Aid and Domestic Principal-Agent Problems. In D. G., Hawkins, D. A., Lake, D. L., Nielson, & M.J., Tierney (Eds.), Delegation and agency in international organizations (pp 107-139). Cambridge University Press.

Overseas Development Institute (ODI), European Centre for Development Policy Management (ECDM), German Development Institute (GDI), National and Kapodistrian University of Athens and Southern Voice (2015). European Report on Development 2015: Combining Finance and Policies to Implement a Transformative Post-2015 Development Agenda. European Union, Belgium.

OECD (2021a) Official Development Assistance – Definition and Coverage. Retrieved October 21, 2021, from

OECD (2021b). Modernisation of the DAC Statistical System. Retrieved October 21, 2021, from

OECD (2021c). Official Development Assistance (ODA). Retrieved October 21, 2021, from

Rodríguez, J., & Santiso, J. (2007). Banking on Development: Private Banks and Aid Donors in Developing countries. OECD Working Paper No. 263.

Rodrik, D. (1995). Why is There Multilateral Lending?. NBER Working Papers 5160, National Bureau of Economic Research, Inc.

Roodman, D. (2014). Straightening the Measuring Stick: A 14-Point Plan for Reforming the Definition of Official Development Assistance (ODA). CGD Policy Paper44.

Vanheukelom, J., Migliorisi, S., Cangas, A. H., Keijzer, N., & Spierings, E. (2012). Reporting on Development: ODA and Financing for Development. Maastricht: ECDPM.

Appendix 1 – ODA measures

Official Development Assistance (ODA) is the official and concessional part of resource flows to developing countries. According to Hynes and Scott (2013) ODA is used to measure “donor efforts in supporting development cooperation objectives, providing the yardstick for documenting the volume and the terms of the concessional resources provided, assessing donor performance against their aid pledges and enabling partner countries, civil society and others to hold donors to account”. Vanheukelom, et. al (2012) suggest that it also allows for some sort of peer monitoring, by increasing transparency on donor inputs and comparability, which contributes to collective self-discipline and supporting benchmarking, information and advocacy activities by NGOs, media and accountability institutions.

The first ODA definition dates back in 1972, when DAC described ODA as “those flows to countries and territories on the  Development Assistance Committee (DAC) List of ODA Recipients and to multilateral institutions which are: a) provided by official agencies, including state and local governments, or by their executive agencies; and b) each transaction of which: i) is administered with the promotion of the economic development and welfare of developing countries as its main objective; and ii) is concessional in character and conveys a grant element of at least 25 per cent (calculated at a rate of discount of 10 per cent)” (OECD, 2021a). The same definition has been used for more than 40 years (up to 2017 data), even though there have always been debates regarding the appropriateness of the measure mainly from the side of the stakeholders in donor countries, who challenged ODA’s definition and reporting system (Vanheukelom, et. al 2012).

ODA’s operationalization evolved over decades, adopting both minor and strong changes, such as agreeing to include technical co-operation in development assistance and widening the list of activities to be considered as promoting development and welfare (Ibid). Nevertheless, failing to adjust to the times (Roodman, 2014) has led to its modernization in 2019. In particular, the Achilles’ heel of ODA was that it compromised between political expediency and statistical reality, as it has been based on interpretation and consensus and therefore has allowed flexibility (Hynes and Scott, 2013). A characteristic example of this loose definition is the dispute of France and Germany with other donors over the exact terms that a loan should meet to qualify as ODA in a world of low interest rates. (Roodman, 2014).

Today the ODA grant equivalent is a measure of donor effort (OECD, 2021a). Grants, loans and other flows entering the calculation of the ODA grant equivalent measure are referred to as ODA flows, that is, those flows to countries and territories on the DAC List of ODA Recipients and to multilateral development institutions which are: a) provided by official agencies, including state and local governments, or by their executive agencies, b) each transaction of which is i) administered with the promotion of the economic development and welfare of developing countries as its main objective; and ii) concessional in character[10]. Moreover, loans whose terms are not consistent with the IMF Debt Limits Policy and/or the World Bank’s Non-Concessional Borrowing Policy are not reportable as ODA (Ibid).

The new ODA measuring system makes a clear distinction between grants and loans, given that, for loans to the official sector which pass the above criteria for ODA scoring, the grant equivalent recorded as ODA is obtained by multiplying the annual disbursements on the loan by the loan’s grant element as calculated at the time of the commitment (Ibid). Hence, it better reflects the actual effort by donor countries – and their taxpayers, as only the “grant equivalent” of loans would now be recorded as ODA, that is, the more generous the loan, the higher the ODA value. Therefore, the new ODA measurement approach improved its reporting system by: a) providing a more realistic comparison of loans and grants, b) measuring ODA loans more accurately and credibly, ensuring comparability of data across providers, c) providing stronger incentives to use grants and highly concessional loans, which will continue to play a key role in mobilizing resources to support the Sustainable Development Goals (SDGs) (OECD 2021b; OECD 2021c).

Interesting conclusions regarding the difference between the old (net disbursements) and the new (grant equivalents) ODA measuring system are drawn if we compare the two measures for two countries, such as Italy and Portugal. Both countries are South European countries and ex colonial powers.

As shown in the following figures and table, the new measurement of ODA estimates higher values of ODA compared to the previous one; in terms of GNI however, there is no significant difference. In particular, according to the latest ODA measurement system, Italy’s ODA is higher than the old ODA measuring system by 1.8% for 2018 and by 2.7% for 2019. This difference is even higher in the case of Portugal, in which the change between the two measurements is 6.1% for 2018 and 7.6% for 2019. However, in terms of ODA as % of GNI the increase is similar in both countries and for both years and counts for about 0.01%.

Figure A. 1 – New and old ODA measuring system, Italy, USD million, 2018-2019

Source: OECD Stats [DAC1], 2021

Figure A. 2- New and old ODA measuring system, Portugal, USD million, 2018-2019

Source: OECD Stats [DAC1], 2021

Table A. 1 – ODA as % of GNI by measuring system, Italy & Portugal, 2018-2019

Source: Own calculations using OECD Stats [DAC1], 2021

Appendix 2 – Allocations in Italy and Portugal by region and income level

Table A. 2 – Italy’s Aid (ODA) disbursementsby region and income level

Source: OECD Stats [DAC2a], 2021

Table A. 3 – Portugal’s Aid (ODA) disbursements by region and income level

Source: OECD Stats [DAC2a], 2021

Appendix 3 – Definitions of the different types of financial aid provided

Box 1


Commitments, Bilateral creditors (COM, current US$) Bilateral commitments are the total amount of long-term loans for which contracts were signed in the year specified. Long-term external debt is defined as debt that has an original or extended maturity of more than one year and that is owed to nonresidents by residents of an economy and repayable in currency, goods, or services. Data are in current U.S. dollars.

Commitments, Multilateral creditors (COM, current US$)        Multirateral commitments are the total amount of long-term loans for which contracts were signed in the year specified. Long-term external debt is defined as debt that has an original or extended maturity of more than one year and that is owed to nonresidents by residents of an economy and repayable in currency, goods, or services. Data are in current U.S. dollars.

Net financial flows, Bilateral (NFL, current US$)         Bilateral debt includes loans from governments and their agencies (including central banks), loans from autonomous bodies, and direct loans from official export credit agencies. Net flows (or net lending or net disbursements) received by the borrower during the year are disbursements minus principal repayments. Data are in current U.S. dollars.

Net financial flows, Multilateral (NFL, current US$)     Public and publicly guaranteed Multilateral loans include loans and credits from the World Bank, regional development banks, and other Multilateral and intergovernmental agencies. Excluded are loans from funds administered by an international organization on behalf of a single donor government; these are classified as loans from governments. Net flows (or net lending or net disbursements) received by the borrower during the year are disbursements minus principal repayments. Data are in current U.S. dollars.

Source: World Bank, Debt Statistics (Metadata), 2021

Appendix 4 – Econometric analysis

A.    Econometric analysis of ODA by recipient countries

In Table A. 4, we analyse, for the period 2005 -2019 (pooled together) how the Log of ODA per capita (in current dollars) is explained by: the Log of GDP per capita (in PPP terms), the index IRAI (World Bank IDA Resource Allocation Index), the log of total population of the recipient country, and a dummy for conflict.

Table A. 4 – Econometric results for recipient countries

B.    Econometric analysis of ODA by donor countries

In Table A. 5, we explain, for the period 2005 -2019 (pooled together) total ODA per capita of the donor country as a function of the log of income per head (in PPP terms) and the log of population of the donor country.

Table A. 5 – Econometric results for donor countries

[1] Professor Emeritus, Department of Economics, National and Kapodistrian University of Athens

[2] Assistant Professor, Department of Economics, National and Kapodistrian University of Athens

[3] PhD Candidate, Department of Economics, National and Kapodistrian University of Athens

[4] For ODA definition and evolution of its measurement see Appendix 1.

[5]  The Cohen and Katseli (2007) paper was originally presented at the Bruno Kreisky forum for International Dialogue, Vienna, 7 October 2004, and at the Experts’ Workshop on Performance and Coherence in Multilateral Development Finance, Berlin, 29-30 January 2007. It remains unpublished. The present paper extends the analysis to recent years and adjusts its policy implications to take into account recent developments.

[6] When correcting for this bias, the balance in 2004 was about fifty-fifty (Cohen and Katseli, 2007)

[7] It should be noted that Actions Related to Debt were implemented, following the Multilateral Debt Relief Initiative (MDRI) in 2005-6.

[8] The World Bank’s IDA Resource Allocation Index (IRAI) is based on the results of the annual CPIA exercise that covers the IDA eligible countries. You can find it at the following website:

[9] For a critical review of governance indicators see Arndt and Oman (2006).

[10] 45 per cent in the case of bilateral loans to the official sector of LDCs and other LICs (calculated at a rate of discount of 9 per cent).

15 per cent in the case of bilateral loans to the official sector of LMICs (calculated at a rate of discount of 7 per cent).

10 per cent in the case of bilateral loans to the official sector of UMICs (calculated at a rate of discount of 6 per cent).

10 per cent in the case of loans to multilateral institutions (calculated at a rate of discount of 5 per cent for global institutions and multilateral development banks, and 6 per cent for other organisations, including sub-regional organisations)

About Author

Connect with Me:

Leave a Reply

  • Theme Settings