Democratizing Foreign Aid

\\server05\productn\N\NYI\42-4\NYI403.
txt unknown Seq: 1 13-AUG-10 13:06
DEMOCRATIZING FOREIGN AID: ONLINE

PHILANTHROPY AND INTERNATIONAL
DEVELOPMENT ASSISTANCE
RAJ M. DESAI & HOMI KHARAS*
I. INTRODUCTION
The nature of development assistance is changing rapidly,
with new delivery mechanisms and new players becoming im-
portant parts of the aid system. Official funding is increasingly
being channeled through specialized agencies dedicated to
particular targets, like HIV/AIDS or malaria, instead of
through traditional agencies like the World Bank’s Interna-
tional Development Association, which provides support for
broad country development programs. A raft of new players
has emerged from the private sector, bypassing the traditional
channels. These new players include foundations, religious or-
ganizations, other NGOs and nonprofits, as well as individual
philanthropists.
Between 1998 and 2008, international private giving by
U.S.-based corporate and independent foundations and indi-
viduals doubled. Private aid, comprised of foundations, non-
governmental organizations, religious groups, and charities in
the United States contributed almost $37 billion to develop-
ment causes in 2007.1 The World Bank, in comparison, com-
mitted about $25 billion.2 This growth in private aid is seen at
all levels, from “mega-charities” such as the Gates, Ford, Mac-
* Raj M. Desai is a Nonresident Senior Fellow at the Wolfensohn

Center for Development at the Brookings Institution, and Associate Profes-
sor of International Development in the Edmund A. Walsh School of For-
eign Service at Georgetown University. Homi Kharas is a Senior Fellow for
global economy and development with the Wolfensohn Center.
1. See CTR. FOR GLOBAL PROSPERITY, HUDSON INST., THE INDEX OF
GLOBAL PHILANTHROPY AND REMITTANCES 2009 17 (2009), http://www.
hudson.org/files/documents/Index%20of%20Global%20Philanthropy%20
and%20Remittances%202009.pdf (citing figures from the OECD 2009 re-
port on development).
2. See THE WORLD BANK, ANNUAL REPORT 2007: FISCAL OVERVIEW, http:/
/web.worldbank.org/WBSITE/EXTERNAL/EXTABOUTUS/EXTANNRE
P/EXTANNREP2K7/0,,contentMDK:21489961~menuPK:4186949~pagePK:
64168445~piPK:64168309~theSitePK:4077916,00.html (tabulating amounts
1111
\\server05\productn\N\NYI\42-4\NYI403.txt unknown Seq: 2 13-AUG-10 13:06
1112 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111
Arthur, Rockefeller, and Hewlett foundations, to hundreds of

smaller foundations.3 Meanwhile, transnational nongovern-
mental organizations (NGOs) such as CARE, Oxfam,
Médecins Sans Frontières, and Save the Children, each with
annual budgets exceeding $500 million, now distribute more
development aid than the entire United Nations system.4
More recently, online philanthropy has proliferated with
the spread of global internet usage. Gifting aid over the in-
ternet is anonymous. There is no face-to-face contact with the
recipient and no expectation of material reward to the giver.
It is one of the purest of altruistic acts. It is also one of the
most rapidly expanding forms of private giving to interna-
tional causes. Although online giving still represents a small
fraction of private aid, several internet-based platforms have
been able to expand in scale, steadily increasing the number
of contributors, as well as the amounts disbursed to developing
countries.5
InterAction, an umbrella group of U.S. NGOs responsible
for giving $6 billion per year in foreign aid, estimates that its
members get contributions from 13 million groups, or about
50 million individuals. In addition, new online-giving tools
have vastly expanded the pool of potential donors. This “de-
mocratization” of foreign assistance raises several implications
for the relationships between aid givers and recipients. How
do private aid flows—especially when the ability to participate
in aid giving is open to all citizens—differ from official aid
flows? What are the factors that influence the character of
these private aid flows? For private aid that is facilitated by
internet social networking, what can be said about the ways in
which different online platforms influence private giving?
from loans, credits, guarantees, and grants; this amount was 4 percent
higher than total lending in 2006).
3. See Raj M. Desai & Homi Kharas, The California Consensus: Can Private
Aid End Global Poverty?, 50 SURVIVAL 155, 155-156 (2008).
4. See generally Organisation for Economic Co-operation & Develop-
ment, Creditor Reporting System, http://stats.oecd.org/Index.aspx?Dataset
Code=CRSNEW (last visited Apr. 12, 2010) [hereinafter OECD, Creditor Re-
porting System].
5. See, e.g., Kiva, Facts and Statistics, http://www.kiva.org/about/help/
stats (citing over 400,000 lenders distributing over $130 million to over
300,000 entrepreneurs) (last visited Apr. 12, 2010).
2010] DEMOCRATIZING FOREIGN AID 1113
We argue that private aid provided by newly enfranchised

individual citizen-donors and citizen-lenders via the internet
represents a unique and potentially consequential form of for-
eign assistance, in three respects. First, online-based private
aid mechanisms resolve some of the traditional constraints
that, historically, have limited the scale and scope of individual
charitable giving and given a rationale for aid to be channeled
through public agencies. Second, online-based private aid is
by definition a reflection of the preferences of philanthropic-
minded individuals who determine their own allocations
across countries and, within countries, across sectors, projects,
and individuals. Official aid, although funded by taxpayers,
gives their citizens little say over aid allocations. Third, this
new kind of private aid is strongly influenced by the specific
characteristics of the person or project receiving aid but not
sensitive to the characteristics of the country in which the re-
cipient lives.
Private internet-based aid represents a new modality for
providing development assistance. It is different from official
assistance by virtue of the fact that it reflects the views of multi-
ple small donors rather than a few experts. The new private
aid platforms do not seek to coordinate or plan aid in any way.
It remains to be seen whether this affects the development out-
comes it achieves to a significant degree.
Part II of this article examines the changing landscape of
development aid, focusing on the rise of online philanthropy
through the example of two well-known internet-based plat-
forms, GlobalGiving and Kiva. GlobalGiving facilitates matches
between community projects in developing countries seeking
funds and individuals who can contribute grants. Kiva allows
individual lenders the opportunity to lend to hand-picked
microfinance institutions at zero percent interest. These
microfinance institutions, in turn, screen local applicants and
lend money to entrepreneurs in developing countries. Both
are 501(c)(3) nonprofit organizations, and both operate pri-
marily though their internet portals, through which anyone
with a credit card or PayPal account can donate to develop-
ment projects or lend to micro-entrepreneurs who post re-
quests online. Part III considers constraints on the provision
of large-scale international aid, and explores how online phi-
lanthropy markets address these constraints. Part IV investi-
gates how democratization of private aid through the
GlobalGiving and Kiva online platforms results in different aid

flow distribution. This consists of three parts: first, a compari-
son of official aid flows to private aid flows from GlobalGiving
and Kiva; second, an analysis of the factors influencing fund-
ing decisions of individual GlobalGiving donors and Kiva lend-
ers; and third a discussion of the effects of donor fragmenta-
tion. In the conclusion we offer some tentative policy implica-
tions of the rise of private online aid for the current global
foreign-aid architecture.
II. THE CHANGING LANDSCAPE OF DEVELOPMENT AID

Warren Buffett made headlines in 2006 by announcing
that he was earmarking 85 percent of his wealth to six founda-
tions, an amount then valued at about $31 billion, the largest
philanthropic gift in history:6 “I know what I want to do, and it
makes sense to get going,” he said at the time.7 The record-
setting size of the declaration and the purposiveness in target-
ing it to international development highlights, in a somewhat
celebrity fashion, the increasing significance of private-sector
giving in the landscape of global aid. Private aid includes not
only generous gifts from philanthropists like Buffett, but also
funds from foundations, corporations, private voluntary orga-
nizations (PVOs), universities, and religious organizations.
Taken together, the private sector in OECD countries may
contribute around $60 billion per year in aid to developing
countries.8
That may seem small compared to official development
assistance (ODA) of $120 billion in 2008. But ODA includes
some artificial forms of “aid”: for example, administrative ex-
6. See All Things Considered: Buffett gift Sends $31 Billion to Gates Foundation
(Nat’l Pub. Radio broadcast June 26, 2006) (reporting on Buffett’s an-
nouncement of his donation).
7. Carol Loomis, Warren Buffett Gives Away his Fortune, CNNMONEY, June
25, 2006, http://money.cnn.com/2006/06/25/magazines/fortune/charity1
.fortune/index.htm.
8. See Homi Kharas, The New Reality of Aid, in GLOBAL DEVELOPMENT 2.0:
CAN PHILANTHROPISTS, THE PUBLIC, AND THE POOR MAKE POVERTY HISTORY?
53, 67 (Lael Brainard & Derek Chollet eds., 2007) (showing that the amount
of private aid from rich countries has sharply increased); Homi Kharas,
Trends and Issues in Development Aid 14 (Wolfensohn Cent. for Dev., Working
Paper No. 1, 2007) (citing statistics from the OECD estimating the amount
of total development aid from rich countries).
penses of bilateral aid agencies; relief on debts that were un-

likely to ever be repaid (along with the punitive accrued inter-
est on past late payments); inflated valuations of technical co-
operation provided by for-profit international consultants.
Considering only development projects and programs in de-
veloping countries, the global size of private aid is today
roughly equivalent to ODA.9 In the United States, private aid
is 50 percent larger than official aid.
This new source of funding, from large and small donors
alike, is changing the nature of private aid organizations.
They are no longer simply contractors for government-spon-
sored programs as in the past but rather have an independent
voice and strategy. InterAction estimates that two-thirds of its
members’ funds now come from private donations; ten years
ago, two-thirds came from the implementation of government
aid contracts. While NGOs depended on government con-
tracts, they simply followed official aid policies. Now that they
are financially independent, they have developed their own
voice.10
The sheer number of official aid players has exploded.11
There are significant benefits to this dynamism, including
more resources, more innovative solutions, and more direct
action. However, there have also been costs. The number of
development projects has grown while the average size of a
project has declined, burdening weak administrative structures
in recipient countries. There is overlap and waste because
each official donor needs to report independently on the
broad development landscape in which each project operates.
Accountability and sustainability are threatened. Mechanisms
for information sharing, coordination, planning, and scaling
9. See Kharas, Trends and Issues in Development Aid, supra note 8, at 1 (cit-
ing statistics from the OECD).
10. See INTERACTION VIEWPOINT, U.S. INTERNATIONAL AND HUMANITARIAN
ORGANIZATIONS SHOULD HAVE A STRONGER VOICE IN POLICYMAKING, Mar. 17,
2009, http://www.interaction.org/document/interaction-viewpoint-us-inter
national-development-and-humanitarian-organizations-should-ha (claiming
that U.S. NGOs are significant and independent actors in international de-
velopment).
11. Estimates suggest that there are 233 multilateral development agen-
cies; fifty-one bilateral donor countries (most with multiple official agen-
cies); several hundred international NGOs; and tens of thousands of na-
tional NGOs, not including community-based organizations, which could
number in the millions. Kharas, The New Reality of Aid, supra note 8, at 71.
up are breaking down. The key issues facing development aid

are those that arise from this fragmentation and the accompa-
nying volatility of aid disbursements. Seeing these problems
with official aid, many private aid agencies are stepping in to
try to do things better.
A. Internet Philanthropy—Global Giving and Kiva
Despite the growth of private aid in general, and internet-

based aid in particular, relatively little is known about the allo-
cation and effectiveness of private aid. Defenders of private
aid argue that private development assistance is more effective
than official development assistance (ODA) due to lower over-
head costs, less susceptibility to corrupt practices, and the fact
that aid is allocated according to need on the ground with very
little of the money funneled back to consultants and contrac-
tors in donor countries (as “technical assistance”), leaving
more for the beneficiaries in developing nations.12
GlobalGiving was launched in 2002 by two former World
Bank economists and is one of the oldest online philanthropy
markets. On the GlobalGiving website, prospective donors can
search projects by country/region, by “theme,” or according to
certain keywords. Donors can also search projects according
to the date they were posted on the website, by project spon-
sor, and according to the percentage of the project request
that has already been funded. Potential grant seekers must ei-
ther submit projects through one of GlobalGiving’s project
“partners” (mainly U.S. and international NGOs) or submit
projects directly after meeting certain due-diligence require-
ments. The primary goal of GlobalGiving’s website is to match
the interests of potential donors with the needs of individuals
seeking funds for projects; for this reason, it is sometimes re-
ferred to an “Ebay” for development assistance.
Kiva operates mainly by bundling together microloans to
entrepreneurs in developing countries. Prospective borrowers
post their projects through one of several affiliated
12. See, e.g., Axel Dreher et al., Acting Autonomously or Mimicking the State
and Peers: A Panel Tobit Analysis of Financial Dependence and Aid Allocation by
Swiss NGOs 20 (CESifo, Working Paper No. 2617, 2009), http://papers.ssrn
.com/sol3/papers.cfm?abstract_id=1392138 (claiming that private dona-
tions arguably respond more to visible success stories).
microfinance institutions (MFIs). The Kiva website, similarly,

allows project searches by geography, sector, and keyword.
Each site takes a different approach to the online “mar-
ket” for philanthropy or microcredit. GlobalGiving does not
restrict the size of donation requests, nor does it limit the
amount of time any given project may remain posted on the
website. By contrast, Kiva limits both loan size and time on the
website. Until the end of 2007, individual loan requests could
not exceed $1,200; that limit has since been raised to $2,000.
The maximum request for group loans remains $5,000. In ad-
dition, borrowers’ projects may list their requests on the web-
site for only thirty days, after which they are removed (the
maximum repayment period for all loans is twenty-four
months). Because of the smaller average size of individual
projects, Kiva has occasionally had to cap individual lenders’
contributions for lack of fundable projects.13
Both GlobalGiving and Kiva also make use of web-based,
interactive forms of communication—journals, blogs, or com-
ment forums. Those who have obtained grants through
GlobalGiving, for example, can post pictures or other informa-
tion documenting their progress and activities. Kiva’s field
partners may post “business journals” identifying how the loan
is being used, or what effect it has had on the business owner.
This reporting is not required by either organization, and thus
the flow of information from recipients can be erratic, and the
information provided is very rarely financially detailed. Never-
theless, Kiva and GlobalGiving platforms provide enough in-
formation to make a personal connection between the donor
and the recipient. A key problem for both organizations is to
decide on exactly what information (and how much informa-
tion) to provide to permit informed choices without over-
whelming an individual donor.
To assist prospective individual donors and lenders in
making decisions, both GlobalGiving and Kiva provide some
information about the projects listed. Once the projects are
posted on the websites, potential donors and lenders can
search projects by size (grant or loan amount), by recipient’s
region, and by sector. GlobalGiving donors may search, addi-
tionally, by project sponsor or the frontline organization that
will run the project, while Kiva lenders can search by gender of
13. Personal conversation with Matt Flannery, KIVA Founder and CEO.
the borrower (both GlobalGiving’s and Kiva’s websites also

have normal search capabilities, where they can query projects
by any term that appears in the project description). On both
websites, short narratives are included by the sponsoring or-
ganization (the charity that will use the money, or the MFI
through which the loan is being channeled) that describe the
purpose for which the funds will be used and give some brief
background information on the principal grantee(s) or bor-
rower(s). Once a project is selected, donor-lenders can con-
tribute funds in any amount up to the full amount re-
quested.14 Using a PayPal account (or in the case of
GlobalGiving, a direct payment from a credit card or a check),
donor-lenders then transfer funds in the pledged amount.
Projects accumulate funds from donors-lenders in this manner
until they are fully funded.
Besides the central difference between GlobalGiving and
Kiva, in that the former collects donations for grant requests
while the latter collects contributions to microloan requests,
each organization uses differently structured platforms to bun-
dle funds from individuals. In particular, GlobalGiving is less
restrictive, in that currently there are no limits in terms of pro-
ject size (and therefore contribution restrictions), listing time
allowed for grant requests on its website, and number of re-
quests allowed at any given time. Kiva, by contrast, limits pro-
ject size, listing time allowed, and the total number of listed
projects permitted on the site. All requests made to Kiva enter
a queue, and after a preliminary screening, they are posted on
the website for a maximum of thirty days, after which they are
pulled from the site.
These platform differences have created contrasting re-
sults for GlobalGiving and Kiva. Because GlobalGiving does
not limit grant requests, and because many of these grant re-
quests are for large amounts of money, the number of projects
is large, but the portion of fully funded projects is relatively
small. Kiva’s more controlled approach, on the other hand,
limits the number and variety of microloan requests that ap-
14. A minimum $10 contribution is required for GlobalGiving. See gener-

ally GlobalGiving, Frequently Asked Questions, http://www.globalgiving.
org/cb/vfpserv/help.html#4.5 (last visited Apr. 12, 2010). For Kiva, the
minimum contribution is $25. See generally Kiva, Help Center, http://
www.kiva.org/help (last visited Apr. 12, 2010).
pear on the site, but those that do appear have so far always
been 100 percent funded. As a result, GlobalGiving has an
abundance of projects, but most projects are listed for several
months before they are funded, while Kiva—facing no
shortage of individuals willing to lend relatively small
amounts—often is without an adequate number of loan re-
quests.
These differences can be seen in table 1, and in figures 1
and 2. Table 1 lists some basic indicators for both internet
platforms. Stark differences appear mainly in terms of project
amount, volume of activity, and funding rates. GlobalGiving’s
projects are, on average, eight times larger than Kiva’s. Activ-
ity on the Kiva website, on the other hand, is intense: 2,000
projects posted in an average month in 2008 (compared to
GlobalGiving’s twelve). The total dollar amounts requested
through Kiva’s site in an average month in 2008 was approxi-
mately $1.6 million, relative to GlobalGiving’s $71,000. As ex-
pected due to different average amount requests, Kiva lenders
fund a typical loan request in approximately two days while the
average GlobalGiving project requires over a year to be fully
funded. Finally, less than a tenth of all of GlobalGiving’s
projects posted since 2003 have been fully funded. All of
Kiva’s projects posted since 2006 have been funded 100 per-
cent.
TABLE 1: BASIC INDICATORS
Averages GlobalGiving Kiva

Amount requested (by project, 2008) $5,935 $725
Funding rate ($/hour) 1.19 93.68
Hours per grant/loan 11,000 52
U.S.-based donor share 0.78 0.70
New projects (by month, 2008) 12 2,229
Total requested (by month, 2008) $71,220 $1,616,025
Percent fully funded (by project, 2008) 9.5% 100%
Figure 1 shows monthly gross disbursements for GlobalGiving

and Kiva. GlobalGiving’s monthly disbursements reached
approximately $750,000 by the end of 2008, while Kiva was
disbursing about $3.5 million monthly. Kiva’s dramatic
growth of gross disbursement is also helped by the fact that by
the end of 2008 some original loans were being repaid and
could be re-lent.
FIGURE 1: GROSS DISBURSEMENTS VIA GLOBALGIVING AND KIVA

INTERNET PORTALS
Internet Portals
Figure 2 shows all grant and loan requests for GlobalGiv-

ing and Kiva, respectively, since their establishment. The first
graph plots all requested amounts (in U.S. dollars, log scale)
to GlobalGiving over a period of seven years, showing both
projects that are fully funded (solid circles) and those that are
as of yet not fully funded (hollow circles). The funded re-
quests tend to be scattered along the bottom of the graph, sug-
gesting that grant requests for large projects are more likely to
remain unfunded for a longer period of time. The second
graph shows a similar scatterplot for Kiva (over three years).
As all Kiva loan requests are fully funded, there is no distinc-
tion between funded and unfunded loan requests. The graph
also shows cut-offs at the $1,200 level (which is the maximum
loan that individual entrepreneur can request) and $5,000
(the maximum request for groups since late 2007).
FIGURE 2: PROJECTS (GRANTS AND LOANS) FUNDED
Kiva
Global Giving
III. ISSUES IN PRIVATE AID PROVISION

Foreign aid delivered through official channels does not
provide direct connections between citizens and recipients.
Citizens pay their taxes to the government, which in turn allo-
cates resources to other governments to fund myriad public
programs, among them programs that benefit poor individuals
around the world. There is no face-to-face contact between an
individual taxpayer and the final recipient, and, insofar as tax-
payers have inaccurate perceptions of how their government
spends development aid, there are few concrete expectations
of impact, return, or reward.
Many international development charities operate in a
similar manner. Private donors direct resources to an organi-
zation (with which the donor identifies with, agrees with, or
otherwise trusts), and the organization in turn allocates re-
sources to various programs and operational expenses. Some
organizations do allow varying forms of “sponsorship.” This
usually involves donors receiving updates from recipients (e.g.,
updates from a child recipient), or selecting a level of dona-
tion that corresponds to different types of the organizations’
activities (e.g., a donation of X dollars allows the charity to
purchase Y vaccines). Donors are not typically able to select
which child they sponsor, and funds received are not usually
earmarked.
These traditional modes of aid delivery are designed to
cope with the Samaritan’s Dilemma.15 Effective giving re-
quires an understanding of the structural causes of poverty,
and programs must be designed to address these causes, not
just to treat the symptoms. Government agencies and organ-
ized NGOs provide an institutional basis for making these de-
cisions. Private aid often operates without such institutional-
ized structures, and individuals giving aid must answer a set of
questions. Why should I give (and not someone else)? Will
my help be effective? How should I help? These questions
can be framed in economic terms. Is there a collective action
problem to solve? How high are transaction costs? Are there
agency costs?
15. See generally DEBORAH STONE, THE SAMARITAN’S DILEMMA: SHOULD

GOVERNMENT HELP YOUR NEIGHBOR? (2008) (discussing the question of the
role of government in caring for the poor).
The new forms of internet-based giving offer different an-

swers to the traditional models of charity. First, taxpayers who
support foreign aid in principle may be insufficiently mobil-
ized relative to particular interest groups.16 The internet of-
fers an opportunity for individual action rather than collective
action. Second, the route that official aid takes—from taxpay-
ers to government coffers, to aid agencies to governments in
developing nations to public agencies or private organizations
in the field—is long and winding and, whether accurate or
not, is often perceived by taxpayers in donor countries to be
costly and susceptible to corruption and leakage as funds move
from donor countries to beneficiaries in recipient countries.
Internet-based aid offers a more direct connection between
giver and recipient. Third, donor-country individuals may
want to help, but recognize that they need to act through one
or more intermediaries (usually governments or NGOs)—
parts of a global foreign-aid apparatus that may be simply too
insulated or centralized to incorporate the individual prefer-
ences.17 Internet-based giving offers many more opportunities
for choosing the kind of intermediation platform that donors
feel most comfortable with.
A. Collective Action Problems

The public economics literature suggests that collective
action problems may block private giving for worthy causes be-
16. See Robert K. Fleck & Christopher Kilby, Foreign Aid and Domestic Polit-
ics: Voting in Congress and the Allocation of USAID Contracts Across Congressional
Districts, 67 S. ECON. J. 598, 612 (2001) (using empirical data to show voting
patterns and voter preferences); Helen Milner & Dustin Tingley, Explaining
the Internationalist Orientation in American Foreign Economic Policy: Theories of
Legislative Politics in Trade and Aid Policy 8-9 (2008), http://www.princeton.
edu/~hmilner/working%20papers/Trading%20and%20Aiding_092408.pdf
(discussing the differences between general support for foreign aid and sup-
port among special interest groups).
17. See William Easterly, How to Assess Needs for Aid? The Answer: Don’t Ask
21 (N.Y.U. Dev. Res. Inst., Working Paper No. 18, 2005), available at http://
papers.ssrn.com/sol3/papers.cfm?abstract_id=876982 (emphasizing that in-
dividual preferences motivate searches for what works in development);
David Roodman & Scott Standley, Tax Policies to Promote Charitable Giving in
DAC Countries 3 (Ctr. for Global Dev., Working Paper No. 82, 2006), available
at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=98402 (stressing
that individual preferences are rarely incorporated into governmental aid
plans).
cause each individual, behaving rationally, tries to free-ride on

others’ generosity. Governments can overcome the collective
action problem by taxing everyone and providing grants to the
causes to which individuals would want to give. One common
empirical approach is to test whether individual donations are
smaller in areas where government grants are larger. Such
crowding-out would be evidence of collective action problems
at work.
A review of the literature finds evidence that crowding-out
in domestic charities is significant.18 Bekkers and Wiepking’s
summary mostly looks at cross-section studies. In the long run,
cointegration tests show that increased government spending
crowds out charitable giving, especially in the education sec-
tor. In the short run, however, the effects are weaker and not
significant. The authors suggest that it may take time for indi-
viduals to get full information on what the government is do-
ing.
Andreoni and Payne also confirm the crowding-out effect
in a large sample of charities.19 They demonstrate that crowd-
ing out occurs through two channels: classic crowding out
(where donors feel less willing to give) and fundraising crowd-
ing out (where organizations receiving grants reduce activity
to collect donations). The evidence suggests that fundraising
crowding out accounts for 68 percent of the observed crowd-
ing-out effect.
This evidence, however, relates to giving through NGOs
compared to government tax-and-spend programs. It does not
directly look at new forms of internet-based giving. For these
18. See, e.g., Rene Bekkers & Pamela Wiepking, Generosity and Philan-
thropy: A Literature Review 27-28 (Neth. Org. for Sci. Res. Working Paper Se-
ries, 2007), available at http://papers.ssrn.com/sol3/papers.cfm?abstract
_id=1015507 (surveying the literature on crowding-out effects, which have a
different impact on different sectors receiving donations). Garrett and
Rhine analyze time series for 1965-2003 and reach the same conclusion.
Thomas A. Garrett & Russell M. Rhine, Government Growth and Private Contri-
butions to Charity 6, 14-15 (Fed. Res. Bank of St. Louis, Working Paper No.
2007-012, 2007), available at http://research.stlouisfed.org/wp/2007/2007-
012.pdf (concluding after reviewing studies that crowding commonly results
from government spending).
19. See James Andreoni & A. Abigail Payne, Do Government Grants to Pri-
vate Charities Crowd Out Giving or Fundraising? 93 Am. Econ. Rev. 792, 793
(2003) (showing research that government grants have a demonstrable and
measurable crowding effect).
platforms, the premise is that the act of individual giving is

pleasurable to the donor—that is, assuming that giving is altru-
istic. If so, it would suggest that the collective-action constraint
to international aid is minimal.20
What seems unquestionable is that both official aid and
private international giving have risen strongly in the United
States over the last decade, suggesting limited crowding out.
U.S. official development assistance has doubled since 2000,
from $12 billion to $25 billion in 2008 (in constant 2007 dol-
lars). Private aid has also grown fast. In the United States,
private giving for international development totaled $36.9 bil-
lion in 2007. While time series data on private giving volumes
is not available, the number of private foundations has grown
from 40,100 in 1995 to 71,000 in 2005 (with more than 650
U.S. foundations making grants for international affairs).21
Meanwhile, international NGOs quadrupled from 6,000 to
26,000 in the 1990s; today there are thought to be some 40,000
20. There is considerable evidence of “psychic” rewards to charitable giv-

ing. Experiments show that giving charitable donations triggers brain activ-
ity in the subgenual cortex/septal region, areas related to social attachment
and bonding in other species. They conclude that altruistic behavior may be
hard-wired into humans. Trust in other humans is a biologically based part
of human nature, and experiments show that the presence of oxytocin, a
hormone that reduces social anxiety and helps people meet and bond with
each other, is also linked with a greater degree of trust that good behavior
will be reciprocated. The conclusion is that people have an in-built desire
and tendency to respond when they see someone in need. Internet-based
giving provides the connection with the needy. These biological studies sug-
gest there is no collective action problem when individuals have the oppor-
tunity to help others directly. See Jorge Moll et al., Human Fronto-Mesolimbic
Networks Guide Decisions About Charitable Donation, 103 PROC. OF THE NAT’L
ACAD. OF SCI. OF THE U.S., 15623, 15626 (2006) (showing brain scans that
demonstrate differences in brain activity when making altruistic decisions);
Michael Kosfeld, Brain Trust, 5 GREATER GOOD MAG. 18 (2008) (citing results
of a study which show increased trust from people with higher levels of oxy-
tocin in their brains).
21. STEVEN LAWRENCE ET AL., FOUND. CTR., FOUNDATION GROWTH AND
GIVING ESTIMATES: CURRENT OUTLOOK 1 (2007), available at http://founda-
tioncenter.org/gainknowledge/research/pdf/fgge07.pdf (citing more than
71,000 grantmaking foundations in the United States); LOREN RENZ & JSE
ATIENZA, FOUND. CTR., INTERNATIONAL GRANTMAKING UPDATE: A SNAPSHOT
OF U.S. FOUNDATION TRENDS (2006), available at http://www.cof.org/files/
Documents/International_Programs/2006%20Publications/IntlUpdateOct
06.pdf (showing rises in the numbers of organizations and dollars granted
internationally by U.S. foundations).
such organizations.22 The growth in private aid is all the more

striking because survey data suggests that public opinion over-
estimates the size of official aid by significant amounts. In the
United States, 69 percent of people think the U.S. government
gives more than other countries in international aid, as a share
of their GDP.23 That would tend to bias private giving down-
wards if crowding out is indeed serious.
B. Transaction Costs
Official aid is perceived to have low transaction costs be-
cause it operates at large scale. But official aid travels a long
route, with costs at each stage. The first stage is the cost of tax
collection when money is transferred from individuals to the
treasury. In this stage, costs consist of the direct administrative
costs of tax collection as well as deadweight losses from taxa-
tion. These costs can be substantial.24
In the second stage, official donor agencies transfer funds
to recipient country governments to support specific develop-
ment projects and programs. The administrative costs of these
agencies have averaged between 4 to 5 percent, according to
statistics reported by the OECD Development Assistance Com-
mittee.25
The third stage involves costs associated with transferring
the money from the recipient government to final benefi-
ciaries through project implementation. Administrative costs
of the project, corruption, and other leakages mean that only
about half the funds actually reach their stated end purpose.26
22. See Robert O. Keohane & Joseph S. Nye, Introduction, in GOVERNANCE

IN A GLOBALIZING WORLD 29 (Joseph S. Nye & John Donohue eds., 2000)
(claiming a four-fold increase in the number of NGOs since the 1990s).
23. CLAY RAMSAY ET AL., WORLD PUBLIC OPINION, AMERICAN PUBLIC OPIN-
ION AND GLOBAL HEALTH 6 (2009), available at http://www.worldpublic
opinion.org/pipa/pdf/may09/WPO_IOM_May09_rpt.pdf.
24. For example, Alm found deadweight losses on U.S. taxes could ap-
proach 9 percent. See James Alm, The Welfare Cost of the Underground Economy,
24 ECON. INQUIRY 243, 259 (1985) (citing conclusions from empirical data
on deadweight losses).
25. See generally OECD, Creditor Reporting System, supra note 4.
26. One World Bank study, based on surveys in Ghana, Tanzania, and
Rwanda concludes that, “approximately half of the overall amount allocated
to clinics and hospitals did not actually reach them.” Magnus Lindelow et al.,
Measuring Corruption in the Health Sector, in TRANSPARENCY INTERNATIONAL,
GLOBAL CORRUPTION REPORT 30 (2007). The GAO, in its April 11, 2008 Let-
In all, transaction costs on official aid could amount to 60 per-

cent or more.
Private aid, particularly internet-based, offers a more di-
rect connection between donors and recipients and potentially
reduces transaction costs. At both GlobalGiving or Kiva, the
flow of funds route is short: money goes from an individual to
the online platform, where it is pooled and transferred to a
financial or project intermediary in a recipient country, which
then disburses to the final beneficiaries. The long route of
passing through government bureaucracies is avoided.
There are, however, some differences in approach. Kiva
focuses on individuals, or entrepreneurs, and provides a per-
son-to-person link. Its feedback mechanisms and other infor-
mation on the site are all geared towards establishing a con-
nection between people. Global Giving, in contrast, highlights
the worthiness of the projects they are proposing for funding.
Donors contribute directly to those activities, rather than to
individuals.
Both Kiva and Global Giving report their administrative
costs for developing and maintaining the websites and provid-
ing the matching and information services that permits the
short route of funding from person to person to occur. Those
costs have averaged around 15 percent. Both companies, how-
ever, are relatively new and still expanding, and costs may be
expected to fall significantly if their business model is success-
ful and expansion continues at its recent rate.
Higher transaction costs are incurred at the financial in-
termediary that conveys the money from Kiva to the micro-en-
trepreneur and that collects repayments. Studies suggest that
interest rates on microcredits are between 30 to 50 percent in
order to cover default risk and transaction costs.27 At the up-
per end of that rate, the total costs for private giving through
ter on The Millennium Challenge Corporation: Analysis of Compact Development

and Future Obligations and Current Disbursement of Compact Assistance indicated
that 27 percent of in-country disbursements between July 2005 and end-De-
cember 2007 were for administrative, audit, fiscal, and procurement ex-
penses. Additional disbursements were made to cover MCC headquarters ad-
ministrative, due diligence and audit expenses, http://www.gao.gov/new.
items/d08577r.pdf.
27. See Consultative Group to Assist the Poor, Making Sense of Microcredit
Interest Rates (CGAP Donor Brief No. 6, 2002), available at http://www.cgap.
org/gm/document-1.9.2434/DonorBrief_06.pdf.
microcredit, to take just one example, are probably about the

same as for official aid.
C. Agency Costs
From the perspective of individual donors, channeling
funds through official aid agencies has the drawback that it is
the agency that decides what projects to fund rather than the
donor. For some, this may not be a cost but a benefit. If an
official agency learns about what works in development, has an
active evaluation mechanism, strong project review and imple-
mentation structures, and appropriate financial controls, it
may provide valuable services to the individual donor.
On the other hand, if the official agency chooses projects
on a different basis from what an individual would choose, or
imposes conditions on its giving, there may be agency costs. In
the past, one large source of agency costs came from tied aid, a
practice where procurement was linked to the country of ori-
gin of the funds. It is unlikely that altruistic individual donors
would care about whether a particular good or service was pro-
cured from a specific country, whereas it is more obvious why a
national government may care about such linkages. Estimates
of the cost of tied aid vary, but averaged between 15 to 30 per-
cent.28 For some types of aid, like technical cooperation, the
costs of tying may be even higher.
A further source of agency costs comes from differences
in approach about what makes for an effective aid interven-
tion. Individual donors may have views about project size, gen-
der of recipient, sector, and other characteristics that differ
from official agency views. The greater the difference, the
greater the agency costs of transferring aid through official
channels rather than through direct person-to-person giving.
IV. THE DISTRIBUTION OF AID FLOWS: EVIDENCE FROM

INTERNET PHILANTHROPY
Assuming that the allocation of aid through GlobalGiving
and Kiva reflects philanthropic citizens’ preferences regarding
development assistance, this Part analyzes data on internet-
28. CATRINUS J. JEPMA, OECD DEV. CTR., THE TYING OF AID 15 (1991)
(aggregating several empirical studies, but acknowledging variation among
results).
based transactions to determine the extent to which private

preferences differ from official aid agency preferences, as well
as to examine the factors that affect the supply of private devel-
opment aid. We proceed in two steps. First, using informa-
tion on disbursements by recipient country, we compare the
allocation of official development aid for projects and pro-
grams to GlobalGiving’s disbursements for projects, and of of-
ficial microcredit lending to Kiva’s microloans. Official aid
categories are defined to overlap as closely as possible with
GlobalGiving and Kiva activities; humanitarian aid, debt relief,
technical cooperation, administrative expenses, and other
forms of official aid are excluded from this analysis. Second,
moving beyond country-specific factors affecting aid and
microloan allocations, we examine the funding rates of spe-
cific project grant or microloan requests—the “velocity” at
which individual grant and loan requests are funded once they
are posted on the GlobalGiving or Kiva websites.
A. Sensitivity of Official and Online Aid to Recipient-Country

Characteristics
Several observers of foreign aid have argued that assis-
tance would have greater impact on growth and poverty were
it better targeted to poorer countries, and to countries with
better “institutions.”29 Empirical work on this subject has
found increasing selectivity of aid money along these lines, es-
pecially after the end of the Cold War.30 In the 1990s, for ex-
ample, bilateral and multilateral aid was allocated more selec-
tively in terms of objective criteria, whereas in the 1980s the
same donors tended to be more arbitrary. Whether private aid
is equally selective remains a puzzle, mainly due to the poor
quality of data on private aid allocation. Some skeptics of pri-
vate development aid believe that aid allocation is influenced
by considerations that are unrelated to the needs of the poor
in recipient countries. In particular, increased competition
29. See David Dollar & Victoria Levin, The Increasing Selectivity of Foreign
Aid, 1984-2003, 34 WORLD DEV. 2034 (2006) (citing a World Bank Study as-
serting that foreign aid would be more effective if given more selectively).
30. See Anne Boschini & Anders Olofsgård, Foreign Aid: An Instrument for
Fighting Communism?, 43 J. OF DEV. STUD. 622, 641 (2007) (discussing the
Bush administration’s approach to foreign aid, which stressed selectivity of
recipients).
among NGOs for funding has prompted these groups to capi-

talize on the misery of the world’s poor in order to perpetuate
and fund themselves. As a result, NGOs may hop from crisis to
crisis, forever seeking the next development cause or humani-
tarian disaster that will mobilize money rather than taking the
long-term perspective that is needed to achieve lasting devel-
opment progress.
GlobalGiving and Kiva disbursements can serve as useful
proxies for private grants and private microloans. We com-
pare four aid categories: (i) net project and program aid from
aid agencies (ODA); (ii) microlending by official aid agencies;
(iii) GlobalGiving disbursements; and (iv) Kiva disbursements,
to examine differences in aid selectivity across recipient coun-
tries for these different types of assistance. All disbursements
are per-capita for recipient countries.
FIGURE 3: SENSITIVITY OF PRIVATE AID DISBURSEMENTS TO

INSTITUTIONAL QUALITY IN RECIPIENT COUNTRIES
Note: Disbursements are log scale. Estimates are derived from stochastic simulations
of a regression equation in which total population (log) and GDP/capita (log) are
fixed at their means, and where Democracy is adjusted to different values along the
sample percentile. Bands show +/– 95% confidence intervals.
Figures 3 and 4 shows the sensitivity of these four differ-

ent aid flows to poverty (measured as per-capita GDP) and
quality of institutions, proxied by the “Polity” index of democ-
racy, which scores countries from -10 (most non-democratic)
to +10 (most democratic).31 We see that official aid is quite

selective. More money goes to poorer countries and to more
democratic countries. Comparing official project aid with
GlobalGiving, we see similar tendencies, but GlobalGiving is
less selective. Democracy is less of a factor for GlobalGiving
donors. Official microcredits, too, are allocated more toward
democratic countries, as with ODA. Kiva lenders, however, are
more likely to help those living in less democratic countries,
perhaps because they are in greater need.32
Private online aid thus appears to be less influenced by
recipient country-specific factors such as GDP per capita or the
quality of country institutions than official aid and microlend-
ing. Overall, philanthropic citizens do not behave like official
donors. Given that both Kiva and GlobalGiving are internet-
based platforms drawing from the same group of potential do-
nors, it may seem surprising that allocations of funds differ be-
tween the two of them. One explanation is that the funding
mechanism drives the cross-country allocation. GlobalGiving
is based on a project-funding concept. If poor countries have
developed an expertise in project preparation to attract
money from official donors, then those same countries may
have better projects to offer. In the case of Kiva, all loans flow
through participating financial intermediaries. These may be
more developed or more widespread in slightly richer develop-
ing countries (or as countries grow richer). Similarly, Kiva do-
nors may feel a special empathy for entrepreneurs struggling
in non-democratic systems, and hence have a higher propen-
sity to lend to them.33
31. Polity data is taken from MONTY G. MARSHALL & KEITH JAGGERS, POL-
ITY IV PROJECT: POLITICAL REGIME CHARACTERISTICS AND TRANSITIONS 1800-
2007 40-45 (2008), http://www.systemicpeace.org/inscr/inscr.htm. Poverty
data is taken from World Bank, World Development Indicators, http://ddp-
ext.worldbank.org/ext/DDPQQ/member.do?method=getMembers&userid
=1&queryId=135.
32. Note that the number of recipient countries varies considerably
across development flows. Countries that receive annual ODA do not re-
ceive these private forms of aid annually. These comparisons focus on a
common set of twenty countries that received all four types of flows between
2006 and 2008.
33. Note that these comparisons—between official aid, official
microcredit, GlobalGiving, and Kiva disbursements—are in a common
group of twenty-nine countries to avoid various selection biases due to
GlobalGiving’s or Kiva’s screening procedures (which may limit their ability
FIGURE 4: SENSITIVITY OF PRIVATE AID DISBURSEMENTS TO

POVERTY IN RECIPIENT COUNTRIES
Note: Disbursements are log scale. Estimates are derived from stochastic simulations
of a regression equation in which total population (log) and Democracy are fixed at
their means, and where GDP/Capita (log) is adjusted to different values along the
sample percentile. Bands show +/– 95% confidence intervals.
B. Factors Affecting Variation in Funding Rates for GlobalGiving

and Kiva
If traditional country-, sector-, and risk-related factors do
not seem to have a strong effect on the willingness of individ-
ual philanthropists to give aid, what does? Individual philan-
thropists may be more likely to focus on project-specific fac-
tors, such as the individuals who are receiving the funds, the
specific purposes for which the funds will be used, and the
amounts requested. This section examines the effects of these
factors on the rate at which projects are funded by website visi-
tors. The analysis relies primarily on information regarding
funding rates for GlobalGiving and Kiva projects (measured in
dollars per hour) to determine the relative significance of
these various factors. A project that is funded very quickly can
be assumed to appeal more deeply to more people. The fund-
ing rate would therefore reveal information about the prefer-
to transact with NGOs or MFIs in some countries) or, in Kiva’s case, due to
some countries’ capital controls that may restrict microcredit flows.
ences of donors with respect to the project or recipient indi-

vidual.
Figure 5 shows monthly average dollars-per-hour funding
for GlobalGiving projects (between May 2003 and December
2008) and Kiva loans (between April 2006 and December
2006). Kiva’s rate is, on average, tenfold faster than GlobalGiv-
ing’s (recall that average project size is much smaller for Kiva).
For Global Giving, with some exceptions in 2003 and in 2008,
the funding rates have remained fairly constant. Kiva’s rose
between 2006 and early 2008, fell through 2008, and then
jumped at the end of the year.
FIGURE 5: GLOBAL AND KIVA FUNDING RATES
1. Allocations by Country
Turning to country-specific lending rates, Figure 6 com-
pares average hours to fund (for GlobalGiving projects, log
scale) and average seconds to fund (Kiva loans, also log scale)
on a set of common countries, i.e., countries to which Kiva
loans have been disbursed and countries for which GlobalGiv-
ing projects have been completely funded. This graphic may
be considered a comparison of “relative” funding rates. The
graph does show that some East Asian countries (Indonesia,
Nepal, and Philippines) receive funding from Kiva faster than

other countries, while Cameroon and Sierra Leone have been
favored by GlobalGiving funders.
FIGURE 6: COUNTRY-SPECIFIC FUNDING RATES
Overall, however, there does not seem to be a strong

country-specific bias in terms of funding rates for either
GlobalGiving or Kiva. Similarly, when one compares funding
rates across countries to the same country-specific factors ex-
amined in the previous section—quality of institutions, level of
poverty, etc.—there is no clear relationship between these
country-level indicators and funding rates. This pattern can
also be seen across regions (in Table 2), with two exceptions.
First, North African states tend to receive GlobalGiving money
faster than countries in other regions. Second, for the online
community of both organizations, there is a clear bias against
Eastern Europe and the Commonwealth of Independent
States (CIS) of the former Soviet Union. For both GlobalGiv-
ing and Kiva, the EE/CIS region receive funds at the slowest
rate—approximately two-thirds as fast as for other Kiva re-
gions, barely one-tenth for GlobalGiving.34
34. There is little evidence that GlobalGiving donors or Kiva lenders are
more inclined to fund projects in countries that receive lower amounts of
TABLE 2: FUNDING RATES BY REGION
Dollars Dollars
per Hour per Hour
Region (GlobalGiving) (Kiva)
East Asia/Pacific 1.5 160.8
Eastern Europe/CIS 0.2 114.1
Latin America/Caribbean 0.4 162.8
Middle East 0.7 171.2
North Africa 5.3 —
South Asia 1.2 —
Sub-Saharan Africa 1.0 172.9
2. Allocations by Project Purpose
Both organizations also collect information regarding the

purposes for which funds are to be used by prospective grant-
ees or borrowers. Table 3 lists funding rates by GlobalGiving’s
and Kiva’s different sectoral classifications. For GlobalGiving
the fastest recipients tend to be project sponsors involved with
disaster relief followed by human rights, reflecting the orienta-
tion of GlobalGiving’s donors towards these areas. Meanwhile,
the traditional areas of official development—education,
health, human development, childhood development, eco-
nomic development, even climate change—fare poorly by
comparison. In the case of Kiva, one does see some traditional
areas of development—education and health—being funded
quickly. On the other hand the traditional areas of
microfinance, namely, entrepreneurship in retail and eco-
nomic sectors (construction, transport, retail, services) do not
obtain funds as quickly (only manufacturing does relatively
well). Finally, arts and entertainment—unusual sectors for
microcredit—are funded fast.
official aid, suggesting little evidence of a substitution effect between private

aid and ODA. Additional analyses of funding rates controlling for country-
specific factors do not find any significant effect of levels of ODA per capita.
See Raj Desai & Homi Kharas, Do Philanthropic Citizens Behave Like Govern-
ments? 17 (Wolfensohn Cent. for Dev., Working Paper 12, 2009) (reporting
statistical data on differences between private aid and ODA).
TABLE 3: SECTORAL FUNDING RATES

Global Giving Kiva
Dollars per Dollars per
Theme Hour Sector Hour
Animals 1.5 Agriculture 151.3
Children 0.7 Arts 201.5
Climate 0.5 Clothing 141.6
Democracy 2.1 Construction 125.3
Disaster 6.2 Education 251.8
Economic Development 0.2 Entertainment 202.2
Education 1.3 Food 179.5
Environment 0.4 Health 207.7
Finance 2.9 Housing 113.1
Gender 1.0 Manufacturing 194.5
Health 1.5 Personal Use 124.2
Human Development 0.0 Retail 146.2
Human Rights 5.0 Services 161.8
Sport 0.7 Transportation 130.3
Technology 0.2 Wholesale 146.7
3. Kiva’s Allocation by MFI Rating

Kiva also rates the MFIs through which it does business,
based on repayment records and other factors, as “one-star”
(riskiest) through “five-star” (least risky). If Kiva’s lending
community behaved as traditional microlenders, presumably,
these star ratings should have some impact on the rate at
which funds are provided. Table 4 lists average funding rates
by MFI star rating. It is clear that, at $95/hour, single-star-
rated MFIs receive funds at a slower rate than better-rated or-
ganizations. But once an MFI moves beyond the one-star
threshold, there is not much extra funding speed to be gained
from acquiring three-, four-, or five-star status.
TABLE 4: MICROFINANCE INSTITUTION RATING AND FUNDING

RATES (KIVA)
MFI Rating Dollars
(stars) per Hour
1 92.8
2 165.3
3 161.4
4 148.2
5 168.0
4. Allocation by Loan-Specific Factors

Using information from Kiva’s microcredit projects, for
example, we can examine the relationship between three loan-
specific factors and their funding rates: loan amounts, dura-
tion, and the gender of the borrower. Figures 7 through 9
show these relationships graphically. In figure 7, for example,
we see that the relationship between loan size and funding
speed is U-shaped, small loans and large loans are funded
faster than mid-sized loans. It is likely that small loans are eas-
ily funded by a handful of visitors to the website. But it is also
the case that website visitors may gravitate towards the larger-
sized, group-based loans. Figure 8 shows that the funding rate
declines as loans increase in duration.
FIGURE 7: AMOUNT AND FUNDING RATES (KIVA)
Figure 9, finally, shows that Kiva’s lenders also follow one

of the tenets of microfinance—that women are preferable as
borrowers.35 Except for a few months in 2006 and one month
in 2007, the female lending rate is faster than the male rate (in
35. The graph shows monthly funding rates for projects with male and
female borrowers (if for group loans, than if greater than 50 percent of the
group is male or female) between April 2006 and December 2008.
some cases twice as fast). Figure 9 also shows the overall in-
creasing funding-rate trend for Kiva loans until early 2008.
FIGURE 8: LOAN DURATION AND FUNDING RATES (KIVA)
FIGURE 9: GENDER AND FUNDING RATES (KIVA)

C. Effects of Donor Fragmentation
Internet platforms offer an innovative mechanism for ty-

ing together a fragmented community of individual philan-
thropists and bundling modest contributions into large
amounts. Thus far, we have examined the effects of project-
specific characteristics on the rate at which projects are
funded by website visitors, on the assumption that these micro-
level characteristics trump country-specific factors that deter-
mine the flows of official aid.
But there is also the question of how donors interact. In
other words, the newest website visitor to GlobalGiving or Kiva
will respond not only to the various characteristics of the pro-
ject or loan request, but to what other donors have done up
until that point. In particular, a large number of small dona-
tions funding a project may send different signals than a simi-
lar project that has been funded by a small group of donors.
On the one hand, donors (and lenders) may respond posi-
tively to highly dispersed projects (being funded by a large
number of small-scale donors) as a result of some sort of phil-
anthropic “herd.” Alternatively, donors may be more inclined
to fund more concentrated projects, seeing an opportunity to
be part of a small group of patrons.
We examine these relationships in figure 10 using
GlobalGiving project data. The graph measures the standard
funding rate (dollars per hour) against donor concentration
which is measured as the sum of the squared shares of contri-
butions from all donors. A concentration measure of 1, then,
implies that the project has been funded by a single donor, a
measure of 0.5 implies two donors (0.52 + 0.52), and so on. We
remove all single-funded projects from the data (since there is
no interaction between donors for these projects).
The graph shows that, on average, the more concentrated
the project, the slower that additional donors will fund it,
while more dispersed projects with lots of donors are funded
quickly—evidence of a philanthropic herding effect. But the
variation in funding rates is much higher for concentrated
projects, suggesting that as more and more donors contribute
to projects, the rates at which these projects increases and con-
verges at approximately $5/hour.
FIGURE 10: CONCENTRATION AND FUNDING RATES

(GLOBALGIVING)
V. CONCLUSIONS
We have shown that the selectivity of development assis-
tance varies between official and private donors and lenders.
Official aid tends to be more responsive to country-specific fac-
tors in aid recipient countries, especially the income level and
institutional environment. Elected officials in the largest bilat-
eral donors must justify the use of taxpayer funds. Hence their
concern—both in rhetoric and in practice—with using these
funds to reduce poverty and in allocating these funds to coun-
tries with lower levels of corruption, more democratic institu-
tions, etc. These same factors seem to influence private indi-
viduals—at least those who give through internet contribu-
tions—to a far lesser degree. Individual citizen-donors and
citizen-lenders, however, do care about project-specific factors,
such as project size, the gender of the recipients, and the spe-
cific purposes for which the funds will be used.
Our findings suggest that both the aid instrument and
platform matter a great deal. In particular, the fact that Kiva’s
participants are lending money (in small amounts) rather
than making grants to entrepreneurs may be driving the high
level of activity on Kiva’s website as well as its rapid expansion,

despite the fact that Kiva’s lenders have no recourse in the
event of default. GlobalGiving, being a site that bundles dona-
tions into larger amounts, has expanded more slowly.
Moreover, the person-to-person connections that are the
heart of Kiva’s platform may account for the significant growth
in number of lenders and amounts disbursed at Kiva over
three years. Although Kiva’s lenders actually lend to MFIs, not
to individuals (or groups of individuals) directly, the percep-
tion that specific micro-entrepreneurs are the recipients drives
much of activity on Kiva’s website. Indeed, our findings show
that individual lenders do not care nearly as much about MFI
risk factors as they do about the specific entrepreneurs they
believe they are funding.36
Our findings have implications for policy. Individuals al-
locate aid differently than official donors. This indicates that
private aid and official aid are complementary: official aid sup-
ports countries, private aid supports people. With different
preferences, formal coordination between these different do-
nors may not be needed. Instead, each needs to understand
when and how it can partner with the other to meet differing
objectives.
We also suggest that there are few inherent advantages to
official aid over private aid in terms of collective action. New
internet-based technology appears to have reduced the advan-
tage that official agencies once held in terms of transaction
costs. By its very nature, official aid has disadvantages in terms
of agency costs. Multiple private aid platforms can provide a
choice for donors to give money to recipients in developing
countries in a much more direct way than ever before. The
36. Recent discussions have pointed out discrepancies between how Kiva
has been perceived by its users (as a platform to connect directly with micro-
entrepreneurs in developing countries) and how it works in practice (as a
mechanism that bundles interest-free loans to MFIs that then on-lend the
funds to entrepreneurs at interest), thus severing the connection between
specific lenders and borrowers. See David Roodman, Kiva is not what it Seems,
CENT. FOR GLOBAL DEV., Oct. 2, 2009, http://blogs.cgdev.org/open_book/
2009/10/kiva-is-not-quite-what-it-seems.php (alleging that Kiva modestly
misleads users in order to raise additional funds for entrepreneurs); Stepha-
nie Strom, Confusion on Where Money Lent via Kiva Goes, N.Y. TIMES, Nov. 9,
2009, at B6 (questioning the role of Kiva in extending microcredit to entre-
preneurs in the poorest countries).
rapid growth of private aid may be attributed to the attractive-

ness of this “short route” to giving, but not all recipient coun-
tries are organized to take advantage of this. Our findings sug-
gest that the design of projects can be fine-tuned to make
them more attractive to donors. To give an example: it is
probably more effective to invest in providing assistance to en-
trepreneurs to allow them to develop a sensible project idea
than to invest in building the capacity of microfinance in-
termediaries.
Finally, an implication of the rapid expansion of internet
lending is that aid recipients would do well to organize them-
selves to take advantage of new forms of private aid. In some
countries like India, MFIs must first obtain approval from the
Reserve Bank before they can borrow abroad. That is an obvi-
ous barrier to accessing private loans from Kiva, for example.
India could benefit by relaxing its rules for highly concessional
credits.
The phenomenal growth of internet-based giving is testi-
mony to the potential for private aid to reach a scale which can
be significant in global terms. What has not been shown is
that organizing aid in this fashion is more effective for devel-
opment. A comparison of development effectiveness between
public and private aid platforms is an important direction for
future research.

Democratizing Foreign Aid

Uploaded by

Copyright:

Available Formats

Democratizing Foreign Aid

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Democratizing Foreign Aid

Uploaded by

Copyright:

Available Formats

\\server05\productn\N\NYI\42-4\NYI403.

txt unknown Seq: 1 13-AUG-10 13:06

DEMOCRATIZING FOREIGN AID: ONLINE

RAJ M. DESAI & HOMI KHARAS*

* Raj M. Desai is a Nonresident Senior Fellow at the Wolfensohn

1112 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

Arthur, Rockefeller, and Hewlett foundations, to hundreds of

2010] DEMOCRATIZING FOREIGN AID 1113

We argue that private aid provided by newly enfranchised

1114 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

GlobalGiving and Kiva online platforms results in different aid

II. THE CHANGING LANDSCAPE OF DEVELOPMENT AID

2010] DEMOCRATIZING FOREIGN AID 1115

penses of bilateral aid agencies; relief on debts that were un-

1116 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

up are breaking down. The key issues facing development aid

A. Internet Philanthropy—Global Giving and Kiva

Despite the growth of private aid in general, and internet-

2010] DEMOCRATIZING FOREIGN AID 1117

microfinance institutions (MFIs). The Kiva website, similarly,

1118 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

the borrower (both GlobalGiving’s and Kiva’s websites also

14. A minimum $10 contribution is required for GlobalGiving. See gener-

2010] DEMOCRATIZING FOREIGN AID 1119

TABLE 1: BASIC INDICATORS

Averages GlobalGiving Kiva

Figure 1 shows monthly gross disbursements for GlobalGiving

1120 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

FIGURE 1: GROSS DISBURSEMENTS VIA GLOBALGIVING AND KIVA

Figure 2 shows all grant and loan requests for GlobalGiv-

2010] DEMOCRATIZING FOREIGN AID 1121

FIGURE 2: PROJECTS (GRANTS AND LOANS) FUNDED

1122 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

III. ISSUES IN PRIVATE AID PROVISION

15. See generally DEBORAH STONE, THE SAMARITAN’S DILEMMA: SHOULD

2010] DEMOCRATIZING FOREIGN AID 1123

The new forms of internet-based giving offer different an-

A. Collective Action Problems

1124 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

cause each individual, behaving rationally, tries to free-ride on

2010] DEMOCRATIZING FOREIGN AID 1125

platforms, the premise is that the act of individual giving is

20. There is considerable evidence of “psychic” rewards to charitable giv-

1126 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

such organizations.22 The growth in private aid is all the more

22. See Robert O. Keohane & Joseph S. Nye, Introduction, in GOVERNANCE

2010] DEMOCRATIZING FOREIGN AID 1127

In all, transaction costs on official aid could amount to 60 per-

ter on The Millennium Challenge Corporation: Analysis of Compact Development

1128 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

microcredit, to take just one example, are probably about the

IV. THE DISTRIBUTION OF AID FLOWS: EVIDENCE FROM

2010] DEMOCRATIZING FOREIGN AID 1129

based transactions to determine the extent to which private

A. Sensitivity of Official and Online Aid to Recipient-Country

1130 INTERNATIONAL LAW AND POLITICS [Vol. 42:1111

among NGOs for funding has prompted these groups to capi-

FIGURE 3: SENSITIVITY OF PRIVATE AID DISBURSEMENTS TO

Figures 3 and 4 shows the sensitivity of these four differ-