Since 2003, the Center for Global Development (CGD), a Washington-based think tank, has been ranking the aid programs of the world’s richest countries using its commitment to development index (CDI). The CGD produces an index of donor performance, which measures “aid quantity and quality”. This index tracks net aid transfer as a share of gross domestic product (GDP), while penalizing tied aid and project “proliferation” and rewarding aid to poor but non-corrupt recipients (this is referred to as aid “selectivity”).
In 2011, Japan was ranked second lowest at 21st of the 22 OECD members in the index of donor performance with a score of 1.5. Only the Republic of Korea ranked below Japan. Nordic countries, such as Sweden and Norway, ranked highest with scores of over 12 points.
However, some empirical studies, including my recent ADBI Working Paper, contradict the findings of the CGD. Developing Asia, the main recipient of Japanese aid, has seen strong growth in recent decades and this simple fact suggests that Japanese aid has not been as ineffective as its CGD rankings imply.
A pioneering work by David Dollar and Victoria Levin (2004) examined the selectivity of 41 donor organizations by looking into policy selectivity (targeting aid at countries with sound policies) and poverty selectivity (targeting aid at poor countries) in their overall aid allocations. They found that some donors, such as Denmark, had both high policy selectivity and high poverty selectivity. The policy selectivity of Japanese aid allocation was relatively high, although Japan allocated less aid to poor countries.
However, since many donor countries tend to focus their aid allocation in their prioritized regions or countries, it is more appropriate to estimate selectivity in the aid allocations of a donor by region than to estimate it world-wide. Moreover, long-term trends in selectivity need be assessed to ensure that the long-term impacts of aid are also taken into consideration.
What I have found is that, in the long term and from a regional perspective, Japanese aid is significantly policy selective, and the total selectivity (the sum of policy and poverty selectivities) of Japanese aid is not much different in value from that of Danish aid, although in total, more Japanese aid is given to middle-income countries than to poor countries.
Evaluating selectivity by looking at gross official development assistance (ODA) in total alone may be misleading because ODA loans are provided to countries with higher debt sustainability, which tend to be middle-income countries. Japanese aid is “poverty selective” when it is measured by the aid allocation of grants which are delivered to low income countries, particularly in Sub Saharan Africa.
One of the major challenges in recent aid trends is the “proliferation” of aid agencies which provided aid to numerous recipients and the “fragmentation” of aid receipts from numerous donors in each developing country, which may create unnecessary and wasteful administrative costs, making it difficult to target funds where they are needed most. The CGD indicates that Japan allows project proliferation (small average project size) and that this has led to the low ranking of Japanese aid in the index of donor performance.
Other analysts have concluded the opposite. Arnab Acharya, Ana Teresa Fuzzo de Lima and Mick Moore (2006) produced an index of donor proliferation of each donor’s ODA delivered in 1999–2001 to its potential aid-receiving countries. They found that the leading “proliferators” are Germany, Canada, the Netherlands, Switzerland, Norway, Belgium, the US, and Sweden. Japan, along with Australia and the UK, is among those labeled “others,” i.e., “non-proliferators.”
I used the same method to create a longer-term (1980–2006) index of donor proliferation for 22 bilateral donors. This index ranked Japan 19 out of 22 countries. In other words, Japan has concentrated its aid-giving on fewer countries over the long term.
I also developed an index of aid fragmentation for 1980–2006 for recipient countries in Sub-Saharan Africa and Asia. I found that the levels of aid fragmentation in Asia, and in East Asia in particular, have been consistently lower than those for Sub-Saharan Africa.
The performance of ODA is often assessed by its impact on GDP growth. I examined Spearman’s rank-order correlations betweenthe ranking of countries which achieved higher average GDP per capita growth rates in 160 developing countries during 1990–2007, and the ranking of countries which received higher amounts of ODA provided by major donors during the same period.
Out of the ODA provided by major donors, only Japanese ODA had a positive and statistically significant (at 5%) correlation in ranking with GDP per capita growth rates. While other major donors reduced the share of aid that is expected to have an impact on growth in their total ODA in the 1990s and the early 2000s, Japan maintained its share of such aid to sustain the growth of recipient countries.
While Japanese net ODA as a ratio of its GDP is among the smallest of major donor countries and Japanese aid has received a low ranking in the CDI assessments, the quality of Japanese aid is high when evaluated from results-oriented perspectives.
See Kihara (2012) for more detailed analysis.
Acharya, A., A. Teresa F de Lima, and M. Moore. 2006. Proliferation and Fragmentation: Transaction Costs and Value of Aid. Journal of Development Studies (42)1: 1-21.
Dollar, D., and V. Levin. 2004. The Increasing Selectivity of Foreign Aid, 1984–2002. Washington, DC: World Bank.
Kihara, T. 2012. Effective Development Aid :Selectivity, Proliferation and Fragmentation, and the Growth Impact of Development Assistance. ADBI Working Paper No. 342.
Thank you for your response to my post.
I appreciate your clarification and comments. It is my great honor to receive comments from the author of CDI, which I mentioned in my post.
Let me re-clarify the points you commented on.
1. Prioritized region and selectivity
Although I did not mention the reasons donors have prioritized regions due to space limitations of blog posts, it is true that many donor nations have their priorities in delivering aid. The priority may come from donors’ motivations to satisfy their altruism for the poorest people, or to share the burden to provide global or regional public goods, or to strengthen political and economic ties with certain countries and regions. These motivations back the delivery of aid. As ODA relies on the taxpayer money of donor countries, the aid should be of some value to the people of donor countries in addition to the people of recipient countries. For example, Japan’s ODA Charter states, “The objectives of Japan’s ODA are to contribute to the peace and development of the international community, and thereby to help ensure Japan’s own security and prosperity” (1. Objectives), and, “In light of the objectives stated above, Asia, a region with close relationship to Japan and which can have a major impact on Japan’s stability and prosperity, is a priority region for Japan” (4. Priority Regions).
I am not suggesting “ignoring” the “poverty selectivity” that indicates more aid should go to poorer countries. But what I am suggesting is the need to examine selectivity in more detail rather than evaluating it by looking at aid “in total” alone. Although Japanese ODA has been primarily directed at middle-income countries due to more ODA loans being allocated to them as they have higher debt sustainability, more Japanese “grant” aid has been allocated to lower-income countries, particularly in Sub-Saharan Africa.
2. Ranking of selectivity in CDI
We appreciate the high ranking of selectivity of Japanese aid in your recent publications. The point of my working paper and blog post, however, is to see this measure more closely with by taking different types of aid into account, and from the long-term and regional perspective, i.e. “one size may not fit all.” I appreciate your recent efforts to evaluate each donor by its regional contribution, but more elaboration may be needed to reflect the effectiveness of the aid delivered by each donor.
3. Proliferation in the size of the project
Yes, there is no contradiction in saying that Japan funds smaller projects in fewer countries. But there is also no contradiction in saying that smaller projects from a donor may not cause a heavier burden on the recipients than the burden that larger projects from numerous donors with different procedures may cause. Which is more burdensome? Having developing countries deal with a number of projects with the same procedures, or having them deal with fewer projects that have numerous different procedures because of the differences in donors? The transaction costs of aid recipients who handle more donors would be higher, as the “government effectiveness” (measured by a part of KKZ governance indexes) is negatively affected by the diversion of donors (measured by index of aid fragmentation, i.e. inverse Herfindahl Index of aid shares of donors in a recipient). Government effectiveness, in turn, negatively affects the growth performance of the recipient countries. (The 2SLS regression results are shown in the appendix of my working paper).
4. The impact of aid on growth
I understand you have pessimistic views on growth regression of aid (as is seen in the title of your working paper) Through the Looking-Glass (2008). However, your colleagues at CGD, Clemens, Radelet, and Bhavnani (Counting Chickens When They Hatch: The Short Term Effect of Aid on Growth. First version. 2004) identified ”short-impact aid” (SIA) that can have a growth-enhancing impact on a country within four years.
Although my blog post only mentioned the aid-growth nexus of Japanese aid in rank-order correlations, I also re-estimated in my working papers and Japanese articles the impact of SIA on growth for different sample periods and for Asia and Sub-Saharan Africa by following the method Clemens et al. (2004) employed (2SLS method with lagged explanatory variables to handle endogeneity). I found there the similar results as they found, in that the SIA has two to three times greater growth impact than using aggregate aid (net or gross ODA). Furthermore, as mentioned in the blog post, Japan has consistently provided its recipients with SIA, which has constituted more than 70% of Japan’s ODA since the early 1970s. While other major donors reduced their share of SIA in the 1990s and early 2000s, Japan maintained its share, which indicates Japan helped sustain the economic growth of recipient countries.
Counting Chickens When They Hatch: Timing and the Effect of Aid on Growth (Clemens et al. Revised version. 2011) is not so straightforward, but admits that “increases in aid have been followed on average by modest increases in investment and growth.”
5. Net ODA “minus” per GDP
It is true that the “very low net aid volume as a share of the economy” is the major contributor to the low ranking of Japanese aid in CDI. “Net aid” in CDI is the gross ODA minus ODA repayments and net interest payments, which is less than net ODA, which does not subtract the interest payment. “Net aid” does not count the forgiveness of the “other official flows “ (OOF) (e.g. untied loans from JBIC) debt and “offsetting entry for ODA debt relief,” which net ODA counts. As such, the loan-oriented ODA, which is a characteristic of Japanese aid, is penalized and the effort to reduce the debt servicing burden is not well valued. The regional Commitment to Development Index for Japanese “aid” to East Asia and Pacific scored “null” (zero), which may also be because of this “net aid” concept as substantial repayments have come from the aid recipients of this region.
But the real point is whether this “net aid/GDP of the donor” has differentiable impacts on growth, poverty reduction, or welfare enhancement of the recipient countries. Is the 10 times smaller (in GDP) donor’s $1 worth 10 times more for the recipients than the $1 of the 10 times larger (in GDP) donor? For the recipient, each $1 may be almost the same regardless if it was received from a small donor or a big donor. That is why I put “result-oriented perspectives” to evaluate the quality of aid. There has been a long standing commitment of major donors to provide ODA at a volume of more than 0.7% of their GDP, but the commitment is for securing ample aid and not for ensuring the effectiveness of aid.
As for the quality of aid, recent literature ranks Japanese aid as high as 8 out of 39 donors (Easterly and Pfutze. 2008), or 7 out of 31 donors in Fostering Institution and Maximizing Efficiency in Birdsall, Kharas and Perakis (2011), the first name of the authors is the president of CGD.
As one size or type of aid may not fit all regions or recipients, one size or measure may not fit when evaluating commitments, contributions or aid effectiveness of all donors.
Thank you for this analysis. As the person who runs the Commitment to Development Index, I would like to make several corrections:
1. The post says that “since many donor countries tend to focus their aid allocation in their prioritized regions or countries, it is more appropriate to estimate selectivity in the aid allocations of a donor by region than to estimate it world-wide.” But the text does not explain why. If a donor gives all its aid to relatively rich countries, why should we ignore that in measuring selectivity? Are poor people in Africa less deserving of aid because they are farther away?
2. Your results on selectivity do not contradict those in the CDI. The Japan page of the CDI decribes selectivity as a “Japan Strength” and ranks Japan #4 overall on selectivity.
3. The discussion of proliferation confuses two different concepts. The CDI measures project proliferation</a)—the tendency to fund small projects—while the analysis you cite measures country-level proliferation—the tendency to give aid to many countries. There is no contradiction in saying that Japan funds smaller projects in fewer countries.
4. It is not surprising that Japan’s aid recipients, predominantly in Asia, have grown faster. But that does not prove that Japanese aid caused faster growth. The cross-country statistical approach to assessing the impact of aid on growth is no longer taken that seriously.
5. The post rightly points out that Japan (like the United States) gives very little aid for the size of its economy. I think it is worth emphasizing that this is the major reason that Japan scores low on the aid component of the CDI, not the selectivity and proliferation issues focused on in the post. And, along with high barriers to foreign rice and foreign workers, this is why Japan scores low on the CDI overall.
I agree with many of the points brought out in this article on Japanese aid effectiveness. Many donor countries may have different perceptions and objectives about aid. However, aid is also an instrument to serve foreign policy objectives of the donor nation in many cases. Effectiveness of aid process largely depends on policy selectivity. Merely choosing a country based on its poverty especially soft loans as aid may not serve the purpose as the country chosen may not be having capacity, policies and institutional arrangements to effectively and efficiently utilize the same. Ineffective utilization of aid especially soft loan sets a bad example and both donor and recipient countries and other countries considering to replicate the same may be averse in transacting these types of aid agreements.
However, a number of small grants may be considered for countries on basis of poverty even though country may not be policy selective in order to build up capacity. So, if these aspects are taken into consideration, ranking of commitment to development index (CDI) of Japan and other countries in OECD may change.
Thank you very much for your response to my blog post, and I apologize for my belated answer, as I agree with almost all of your comments. Let me address the points you raised.
1. You wrote that “aid is also an instrument to serve foreign policy objectives of donor nation in many cases”, which I concur. Many donor nations have their priorities in delivering aid. The priority may come from donors’ motivations to satisfy their altruism to help the poorest people, or to share the burden to provide global or regional public goods, or to strengthen political and economic ties with certain countries and regions. These motivations back the delivery of aid. As ODA relies on the taxpayer money of donor countries, the aid should be of some value to the people of donor countries in addition to the people of recipient countries. For example, Japan’s ODA Charter states, “The objectives of Japan’s ODA are to contribute to the peace and development of the international community, and thereby to help ensure Japan’s own security and prosperity” (1. Objectives), and, “In light of the objectives stated above, Asia, a region with close relationship to Japan and which can have a major impact on Japan’s stability and prosperity, is a priority region for Japan” (4. Priority Regions).
2. You are correct that the “effectiveness of aid process largely depends on policy selectivity. Merely choosing a country based on its poverty especially soft loans as aid may not serve the purpose as the country chosen may not be having capacity, policies and institutional arrangements to effectively and efficiently utilize the same.” The panel estimates in Appendix I (pp. 41–42) of my ADBI working paper suggest that “policy selectivity” would make sense as a means of heightening the impact of aid on per capita growth by giving it to better-governed countries. But “poverty selectivity” could not be explained by the effect of aid on GDP growth because aid given to higher income countries resulted in a higher growth rate than aid given to lower income countries. “Poverty selectivity” may be better explained by the value of equity and poverty alleviation for which the poor should receive aid, but not by the growth impact of poverty selective aid.
3. I appreciate your pointing out that “a number of small grants may be considered for countries on basis of poverty even though country may not be policy selective in order to build up capacity.” This is the reason why a number of small projects are conducted by Japanese aid agencies, but it is also the reason why the “proliferation” of Japanese aid ranks badly in CDI scores. As you said, “if these aspects are taken into consideration, ranking of commitment to development index (CDI) of Japan and other countries in OECD may change,” which would happen by evaluating the quality of aid from a result-oriented perspective.
Not an specialist in the subject matter but it seems that evaluating ODA performace based on GDP growth, although interesting, it is sort of far fetched.
GDP growth might reflect policy decisions by governments, and not necessarily is a result of the volume of ODA contributions to that country per se. (Africa might be a case in point of large contributions but low GDP growth).
If the GDP approach is taken, then wouldn’t it be better to focus on contributions to GDP of sectors targeted by ODA (hard and soft) and the sustainability of such contributions over time?
Thank you for your comments on my blog post.
I agree with most of your comments, but we could think more.
1. You mentioned that “GDP growth might reflect policy decision by governments, and not necessarily is a result of the volume of ODA contributions to that country per se.” It may be true, from existing literature, that the contribution of the level (or change) of “policy and institution” or “governance” to the per capita growth is much larger than the level or change in aid volume. It may also be true from the literature, however, that the aid could enhance per capita growth in a better policy and institutional environment. Government policy is important for growth by itself, but it is also important for improving the impact of aid on growth, and eventually on poverty reduction. Aid cannot replace the recipients’ policy, but may help or “aid” their policies and development.
2. I believe you rightly noted that it would be “better to focus on contribution to GDP of sectors targeted by ODA (hard and soft) and the sustainability of such contributions over time.” In fact, it is becoming a common understanding that there are some aid “flows that are not intended or used to promote expansion in generalized productive capacity (such as humanitarian assistance or disaster relief) as well as flows whose effect on overall national growth, if it ever arrives, might come long after the time period under study (such as vaccination campaigns or school feeding projects)” (Clemens, Radelet, Bhavnani, and Bazzi. 2011. Counting Chickens When They Hatch: Timing and the Effects of Aid on Growth, CGD Working Paper 44. Revised 2011. pp.6–7). On the other hand, there are aid flows, called “short-impact” (or “early-impact”) aid (SIA), which includes budget support or program aid, project aid given for real sector investments for infrastructure, or direct support of production in transportation (including roads), communications, energy, banking, agriculture, and industry (Clemens et.al.(2011) p. 13). Clemens et al. (2004) and my working papers and articles found that SIA has two to three times greater growth impact than using aggregate aid (net or gross ODA). Furthermore, as mentioned in my blog post, Japan has consistently provided its recipients with SIA, which has constituted more than 70% of Japan’s ODA since the early 1970s. While other major donors reduced their share of SIA in the 1990s and early 2000s, Japan maintained its share (i.e. sustainability of its contribution), which indicates Japan helped sustain the economic growth of recipient countries.