How Much Foreign Aid Reaches Foreign Governments?

The broad consensus in the international development community is that “country ownership” is a good thing. According to the Paris and Accra agreements on aid effectiveness, developing countries should set their own strategies for poverty reduction and donors should align behind these objectives and use local systems for aid delivery. Ownership is meant to reflect alignment between donor and partner, and help build capacity.

At the same time, many of the concerns which are raised around foreign aid are based on the premise that it provides an easy source of rents for political elites.  

But these critics will be relieved to hear that the Paris and Accra agreements have seemingly been ignored: even using a generous measure of ownership, we find that only about a third of assistance is actually managed by those it is intended to help.  

A good proxy for country ownership is who handles the funds

Donors often measure country ownership by the share of aid that is “country programmable”. This bit of jargon refers to overseas development assistance that isn’t humanitarian aid or debt relief, involves cross-border flows (excluding administrative expenses and refugee hosting costs), and is part of a cooperation agreement. For the average bilateral donor, about half of its aid is country programmable.

This is a very low bar for defining country ownership. It only requires that there will be some new goods and services provided in country X that the government of country X didn’t actively refuse. It doesn’t imply that said government had any role in choosing the project, handling the money, or implementing any of the activities.

A better measure of “ownership” is asking who manages the spending. We can answer that by using the OECD Development Assistance Committee’s Creditor Reporting System (CRS). This system provides data on the “channel of delivery” of projects financed by official flows including aid. The channel of delivery is “the first implementing partner” according to the official definition, “the entity that has implementing responsibility over the funds and is normally linked to the extending agency by a contract or other binding agreement, and is directly accountable to it.”

Our measure still probably errs on the optimistic side with regards to how much aid local recipients actually control. Granted, in some cases the institution listed in “channel of delivery” will pass along the funds to a local organization and this won’t be captured in the data (perhaps particularly when the channel of delivery is reported as a multilateral institution). And the numbers as reported might surprise some aid aficionados—for example US Millennium Challenge Corporation funds are all implemented by the US government according to the CRS report. On the other hand, the projects being financed may still be specified by the extending agency, which means partner-country implementation is an over-generous measure of true ownership. Aid provided as budget support is the acme of country ownership, but at least with partner-implemented aid we know the partner has operational control.

About a third of total aid goes to partner governments or other local organizations

With these caveats, the numbers give some indication of who is actually spending dollars committed to finance for development. And, particularly when it comes to foreign aid, the answer for many donors appears to be “institutions in the donor country.”

The first panel of figure 1 shows trends over time in the of channel of delivery for aid, including finance from bilateral and multilateral donors. The second panel shows trends over time in the of channel of delivery for total development finance, including finance from bilateral and multilateral donors. In short, the figure suggests that, for aid in 2020:

  • 22 percent is channeled through donor governments (as it might be, the US Department of Health and Human Services buys drugs for distribution in Tanzania or the Swedish government pays hosting costs for Ukrainian refugees),
  • 10 percent is executed by private firms or NGOs in donor countries (the U.S. Agency for International Development contracts US-based Chemonics to hire consultants to provide technical assistance in Chad),
  • 29 percent is executed by partner country governments (Ethiopia uses the proceeds of a loan from the World Bank to build a road),
  • 3 percent is channeled through private firms or NGOs in developing countries,
  • 20 percent of development finance is earmarked bilateral aid implemented by multilaterals agencies (UNICEF buying vaccines for example), with an additional 16 percent through channels including networks and international NGOs.

Only 32 percent of aid funds are managed by partner country governments, private sector firms and NGOs combined. The picture is slightly better when it comes to development finance as a whole (which includes non-concessional investments): 43 percent of all official flows are implemented by partner country institutions.

Figure 1. Most aid does not pass through recipient governments or any other local organization

Percent of development finance through different channels

Foreign aid percent through different channels

Note: before 2016 channel code separating donor/recipient private channel was not available. As well as international networks, public-private partnerships and international NGOs, “other” includes a small number of observations labelled “public sector institutions”, for which it wasn’t clear whether donor or partner country is meant. This last category accounts for less than two percent of overseas development assistance.

Source: Creditor Reporting System, OECD

Donors who entrust more aid to recipient governments also rely more on loans versus grants

Figure 2 reports on the channels used for aid from individual major donors, including core contributions to multilateral institutions like the IDA, the World Bank’s low-interest lending and grantmaking arm. That means core funding reported in Figure 1 that subsequently flows out of multilateral agencies or that they implement themselves (Ethiopia’s World Bank credit in the examples above) is reported here as a core multilateral. As 66 percent of core multilateral funding is implemented by recipient governments (compared to 20 percent of direct bilateral aid from major donor countries), they account for most of total recipient-executed aid despite channeling only 38 percent of total flows. The result is reported recipient shares in Figure Two for nearly all donors are considerably smaller than the average reported in Figure 1.

Nonetheless, the differences across donors are large: Japan, the EU and France finance partner-executed aid as a considerable proportion of their total support while the UK, US, Canada and the Netherlands are far more likely to channel aid through home-country governments and organizations.

In part that will reflect the greater use of lending by Japan and France: that partner governments are contractually involved means they can insist on managing the funds. The predominance of loans in non-aid official finance will also be a major factor behind higher recipient executed funding in donor finance as a whole.

Figure 2. The biggest bilateral donors give almost no aid directly to recipient governments or local organizations

Percent of bilateral aid through different channels, 2020

Graph - Percent of bilateral aid through different channels, 2020

Note: As well as international networks, public-private partnerships and international NGOs, “other” includes a small number of observations labelled “public sector institutions”, for which it wasn’t clear whether donor or partner country is meant.  Core multilateral funding is that for which donors relinquish control, whereas for “channeled through multilaterals”,  multilateral agencies are simply used as implementing agencies for donor-specified p
Creditor Reporting System, OECD and DAC table 1, OECD

Directing less aid to country governments also reflects political preferences in donor countries: those in which there is more skepticism about state-delivery and a higher preference for markets might be less willing to use governments to deliver services. However, it also appears that donors less likely to channel aid through partner governments are also considerably less likely to draw their country aid objectives from partner country-led frameworks (as measured by the Global Partnership for Effective Development Cooperation). Donors engaging less with partner country governments are not just demonstrating a preference for outsourcing, but a general de-prioritization of partner country ownership.

We’re not strongly confident in the trends presented in Figure 1. They may be an artifact of improved reporting as much as a changed reality. At least in the 2019-20 period, there does appear to have been a shift towards partner-executed aid, potentially related to the pandemic. But when partner governments still only implement about 29 percent of aid (and budget support to partner countries, the purest form of country-owned aid, still accounted for just 6 percent (even) in 2020), we are a long way from a dominant model of country ownership. For donors including the UK, Canada, and the US, effectively the only way they provide meaningful levels of recipient-executed aid assistance today is through the multilateral system, and there is little sign of that changing. (In the UK in particular it is probably getting worse: the country used to give as much as 20 percent of bilateral aid as budget support and was a comparatively generous backer of the multilateral system: times have long changed for budget support, and now they are changing for multilateral support).

This suggests the need to caveat concerns around aid dependency and the corrosive effect of aid on partner governments and institutions. It may be that foreign assistance is worth 75 percent of government expenditures in ten countries worldwide, but the extent to which that assistance is a substitute for domestic resources or an easy source of rents is surely reduced if less than a third of that assistance is actually controlled by partner governments (even then with significant conditions as to use and monitoring).

Channeling aid through domestic firms or NGOs might (arguably) make it more palatable to donor-country voters skeptical about the value of foreign aid. There are reasonable concerns with capacity and controls around partner-executed assistance, especially in some of the partner countries where aid flows are largest, and good reasons to channel some resources to non-government actors in partner countries directly. So, perhaps not all finance should be “country owned.” But we are an extremely long way from that.


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.