Journal of International Development

J. Int. Dev. (2010)

Published online in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/jid.1701

FOLLOW THE LEADER: HOW DUTCHDEVELOPMENT NGOs ALLOCATE THEIR

RESOURCES – THE CONTRADICTORYINFLUENCE OF DONOR DEPENDENCY

BART LOMAN, IOANA POP and RUERD RUBEN*

Centre for International Development Issues (CIDIN), Radboud University Nijmegen, Nijmegen,

The Netherlands

Abstract: Public funding to non-governmental development organisations (NGDOs) influ-

ences their allocation choices regarding preferred recipient countries for development aid.

Instead of strict poverty-orientation criteria for aid allocation, NGDO country choice becomes

increasingly aligned with bilateral priorities. Based on original data on Dutch aid delivery to

developing countries, combined with detailed information on NGDOs financial structure and

country characteristics, we analyse whether and how donor dependency and public subsidies

influence NGDO country choice. Independent NGDOs appear strongly inclined to follow

public aid priorities. This is attributed to risk-avoidance and complementarity-seeking of

independent NGDOs and deliberate non-alignment for identity-seeking of dependent NGDOs.

Copyright # 2010 John Wiley & Sons, Ltd.

Keywords: non-governmental development organisations; country choice; donor depen-

dency; financial structure; risk avoidance; identity-seeking

JEL Classification: F35, DO2, C31, G11

1 INTRODUCTION

Foreign aid to development countries has increased significantly in recent years. The

world’s major donor countries of the OECD Development Assistance Committee (DAC)

disbursed globally US$ 104 billion in Official Development Assistance (ODA) in 2008, an

increase of more than 60% compared a decade earlier. Given this large increase, much

*Correspondence to: Ruerd Ruben, Centre for International Development Issues (CIDIN), Radboud UniversityNijmegen, P.O. Box 9104, 6500 HE Nijmegen, The Netherlands. E-mail: r.ruben@maw.ru.nl

Copyright # 2010 John Wiley & Sons, Ltd.

B. Loman et al.

attention has been devoted to the underlying determinants for the geographical aid

distribution of aid expenditures between countries and regions. While earlier research has

been primarily focused on bilateral and multilateral donor behaviour (e.g. Rowlands and

Ketcheson, 2002; Mascharenhas and Sandlers, 2006), there has been a recent surge of

attention on explaining the expenditure pattern of non-governmental development

organisations (NGDOs) (see Masud and Yontcheva, 2005; Dreher et al., 2007; Koch,

2009).

Most of this NGDO research selected determinants based on juxtaposing a needs-based

model with a donor-interest based model. According to the needs-based model, donors

would focus primarily on the poorest (Riddell and Robinson, 1995) and poorly governed

countries (Fowler and Biekart 1996). Following a donor interest model, NGDOs would

have less altruistic motives and especially allocate their funds to countries according to

political and economic interests of their back donor. This model is inspired by Edward and

Hulme’s (1997) proposition that official funding from governmental back donor weakens

NGDO’s independency.

Available verification of these two models for NGDOs points foremost to the latter one.

With the exception of NGDOs funded by the European Union (Nancy and Yontcheva,

2006), country-specific studies of Swedish, Swiss and Dutch NGDOs conclude that

especially bilateral aid is a primary determinant for NGDO country allocations (Dreher

et al., 2007; Nunnenkamp et al., 2008; Koch and Loman, 2009). This was also confirmed in

a cross-country study on major NGDOs from 15 OECD countries (Koch et al., 2008).

Arguably, the most in-depth study on determinants for NGDO aid expenditures concluded

that geographical choices were ‘rather governmental, that is influenced by the preference of

the back donor’ (Koch, 2009: 163).

Interestingly, little or no research has been conducted on the effect of NGDO funding

structures on their aid allocation decisions, while this could be of significant importance for

country choice. For instance, Koch (2009: 1155) mentions that many Northern NGDOs are

struggling for greater ‘room to manoeuvre’ from their back donors in their aid allocation

decisions. This implies that NGDOs would make other geographical choices in the absence

of such back donor’s pressure. Moreover, it raises the question whether financially more

autonomous NGDOs would in fact provide more poverty-targeted aid than their

counterparts (Nunnenkamp et al., 2008: 21). NGDOs might actually prefer to allocate their

funds according to the needs-based model, but are restricted by political and economic

interest of their back donor.

This article contributes to the discussion on NGDO aid expenditures in two ways. First,

we will test determinants for geographical aid allocation of Dutch NGDOs and their

alignment with bilateral funding priorities using a new and expanded dataset. Previous

research on Dutch NGDOs only focused on the expenditures of four major Dutch

organisations (Koch and Loman, 2009). However, the Dutch NGDO sector is characterised

by a multitude of organisations, which together channel about 40% of bilateral aid. We

compiled a detailed dataset of 73 major Dutch NGDOs and their country-wise and sector-

wise expenditures to ODA recipients in 2007, covering an estimated 70% of expenditures

of the Dutch NGDO sector. This detailed database allows us to test the determinants for

Dutch NGDO aid allocations. Second, we analyse whether and how the internal financial

structure of Dutch NGDOs influences their aid allocation decisions, compiling an

independency and subsidy index for the NGDOs. We use these two indexes as interaction

variables with commonly used determinants for aid allocation to verify whether back-

donors interests modify NGDO country choice.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Contradictory Influence of Donor Dependency

2 NGDO DEPENDENCE AND DONOR ALIGNMENT

Development aid is assumed to be strongly oriented towards poverty alleviation, and

therefore geographical distribution of aid is likely to focus on the poorest countries, regions

and population segments. On the other hand, aid relationships are subject to certain path

dependency and thus replicate historical allocation patterns. Moreover, political motives

regarding good governance are increasingly influencing the choice of preferred recipient

countries for delivering development aid.

Academic research on aid channels often distinguishes between two models for

explaining aid allocations: the ‘recipient needs’ and the ‘donor-interest’ model (Dudley

and Montmarquette, 1976; McKinley and Little, 1977). The former model assumes that

development aid is likely to focus on the poorest countries and regions, whereas the latter

model explains NGDO aid allocation from the priorities of lead funding agencies.

Empirical evidence has been somewhat contradictory. Nancy and Yontcheva (2006) find

that most EU-funded NGDOs favour poverty motives in their country choice, but other

studies focussing on (inter)national NGDOs (e.g. Dreher et al., 2007; Nunnenkamp et al.,

2008; Koch, 2009; Koch and Loman, 2009) do not confirm this and instead reveal that aid

relationships are subject to donor dominance.

Development aid delivered through NGDOs might either follow the priorities of lead

bilateral funding agencies or could search for a more independent allocation profile. The

latter decisions are commonly assumed to depend on their internal financial structure, e.g.

the degree of funding received from back donors (Koch, 2009). Recently, some new

empirical studies on the determinants for NGDO aid allocation (Nancy and Yontcheva,

2006; Dreher et al., 2007; Nunnenkamp et al., 2008) confirm the importance of bilateral aid

for NGDO country choice. These studies seem to suggest that Northern NGDOs closely

follow the policy priorities of their public counterparts.

Another reason for closely looking at the underlying determinants of NGDO country

choice refers to the variability within the non-governmental development sector regarding

their internal funding structure. Earlier studies assumed that all NGDOs are donor-

dependent, while in fact there are many organisations that are capable of raising substantial

private donations through campaigning, charitable giving and membership fees. The

philanthropy literature indicates that fundraising income of Dutch NGDOs has more than

doubled during the last decade (Beckers, 2003), while maintaining reasonable cost ratios.

There are two competing hypotheses regarding the potential impact of donor funding on

NGDO country choice. The standard idea is that NGDOs that largely depend on public

subsidies will feel obliged to spend their resources in the same countries where their back-

donors are operating. This is further reinforced by the fact that bilateral donors increasingly

demand complementarities between public and private aid allocations, as part of the Paris

agenda for aid harmonisation (Rogerson, 2005). The alternative hypothesis pays more

attention to the importance of identity-seeking and the result-orientation for NGDOs, that

requires them to deliver tangible output to guarantee public support and societal legitimacy

(Clegg et al., 2007). This implies that even ‘dependent’ NGDOs that are less capable of

raising substantial own resources feel obliged to develop a recognisable profile to maintain

themselves eligible for public funding. Instead of free-riding on ongoing projects funded

by other agencies in the same country, they might prefer to engage in more innovative,

albeit risky, projects in settings where less donor presence is recorded.

Further testing of both hypotheses requires detailed insight in the aid allocation pattern

and the financial structure of NGDOs. Available data from Dutch NGDOs permit us to

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

B. Loman et al.

scrutinise the relationships between donor dependency and country allocation of aid. This

provides opportunities for deepening the debates on the possibilities and likelihood of back

donor-influenced aid patterns and on the frequently contested relationship between donor

funding and (in)dependent decision-making within the NGDO community. Whereas some

European bilateral donor agencies tend to reinforce their influence over non-governmental

development aid operations, the funding instrument might be insufficient. Instead,

understanding the internal logic of NGDO aid allocations could offer new insights into the

trade-offs that reign the priority-setting agenda of mixed public–private agencies, where

risk-avoidance and identity-seeking arguments simultaneously ask for attention

(Yescombe, 2007).

The resulting aid allocation pattern is thus likely to be influenced by two opposing

forces: (a) risk-avoidance for NGDOs that are drawing mostly on private fundraising, and

(b) identity-seeking for NGDOs that compete for public funding. This may result in a rather

unexpected pattern of country choice for aid allocation, where the former type of (more

independent) NGDOs seek to align their activities with major bilateral priorities, whereas

the activities of the latter category of (subsidy-dependent) NGDOs are more difficult to get

dovetailed with backdonor interests. The contradictory role of donor dependency becomes

visible at the interface between public objectives of aid harmonisation and private

objectives for identity-seeking. Strangely enough, reducing the donor dependency in

NGDO funding might thus result in greater prospects for public–private alliances in

international development cooperation.

3 DATA AND APPROACH

We composed an unique data set that permits to analyse the determinants for the

geographical allocation of development aid by Dutch NDGO’s to developing countries.1

Our dataset includes 73 NGDO’s with a total expenditure of 840 million euro in the year

2007. This is considered representative for about 70% of all development NGDO’s

operating from the Netherlands. The data set thus contains a total of 7211 combinations of

Dutch NGDO expenditures in 99 so-called OECD-DAC countries.2 We combined the aid

allocation data with information on the financial structure from the Dutch Central

Fundraising Bureau (CBF). On average, Dutch NGDOs depend for 36% on government

subsidies and are capable of raising private funding up to 53% of their budget (see

Table 1).3 The Dutch civic aid landscape includes a small number of large traditional

NGDOs that are strongly dependent on state support and fundraising campaigns, alongside

a large number of medium-size and smaller agencies that rely on mixed funding from

public and private sources. Public funding for NGDOs used to be based on their supposed

strong poverty orientation through direct linkages with local civil society structures.

However, the rules applied in the competitive tendering system for NGDO aid also include

criteria related to streamlining with official aid priorities in terms of country and sector

choice (Ruben and Schulpen, 2009).

1All expenditures in the initial dataset are split up by sectors based on the OECD Creditor Reporting System (CRS)codification. This allowed to exclude humanitarian aid which is strongly correlated with emergency situations.2We selected all official DAC recipients in 2007 excluding the category ‘small development islands’.3Note that this is an unweighted average for all 73 NGDOs. Especially for the larger Dutch NGDOs, the subsidydependence from public sources is substantially higher.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Table 1. Descriptive statistics

Variable N Mean SD Min Max

Dependent variables

NGDO aid to OECD-DAC countries 7211 0.136 0.343 0 1

Total NGDO aid (000 US$) 983 502,239 999,577 51 11,630,148

Country characteristics

GDP per capita (US$) 7211 2.59620 3.25724 114.00 19.16600

Billateral aid (�100.000) 7211 9.27 29.30 0.00 259.00

Polity index 7211 1.84 6.25 �10.00 10.00

Dutch export value (�100,000) 7211 333.00 705.00 0.3934 4350.00

Religion – Christian 7211 0.47 0.50 0.00 1.00

Religion – Muslim 7211 0.38 0.49 0.00 1.00

Religion – Other 7211 0.14 0.35 0.00 1.00

Dutch NGDO aid 2006 (�1000) 7211 2112.19 4023.10 0.00 28000.86

Distance (km) 7211 6397.69 2720.52 1087.00 11987.00

Budget NGDO 2006 (�100,000) 6127 203.99 487.47 0.00 2185.00

Subsidy dependency

Subsidy index (2006) 5039 0.37 0.33 0.00 1.00

Subsidy index with imputationsa 7211 0.36 0.32 0.00 1.00

Independence index (2006) 5039 0.53 0.35 0.00 1.00

Independence index with imputationsa 7211 0.55 0.34 0.00 1.00

Sources: World Bank, OECD and Dutch CBF statistics.aBoth subsidy index and independence index for the year 2006 had missing values (particularly for the subset ofcountries where an NGDO does send aid). In order to deal with this situation we imputed missing values usingregression methods. The initial and final variables maintain the same standard deviation.

Contradictory Influence of Donor Dependency

We used several variables to assess the validity of the need-based or donor-interest aid

allocation model. For the need-based model, we relied on GDP per capita – measured in US

dollars – as an indicator of poverty corresponding to year 2006.4 The governance level is

measures by the Polity IV index, which uses a composite index accounting for the

combinations of monarchies, anocracies and democracies. The continuous variable ranges

from �10 to þ10, where �10 signifies a country that is a full autocracy while þ10 signifies

a country that is a full democracy. In the probit regression we used the continuous variable

(since there are no assumptions on the normality of the variables) while in the OLS

regression we categorised the variable using country information. The donor-interest

model was tested by assessing the bilateral aid sent to a specific country, as sum of public

and private aid, measured in US dollars (OECD-DAC, 2008), corresponding to year 2006.

This measures whether NGDOs follow their back donor in aid allocation. Second, we

included the export level between The Netherlands and a specific country from the dataset,

corresponding to year 2006 to test for economic interests from the back donor. Both

variables were transformed in natural logarithms to guarantee normality.

To test the effect of the funding structure of a NGDO, we first calculated an

independency index. We used data for 2006 from the CBF to assess the internal financial

4We introduced this variable into the models after natural logarithm transformation. Other possible variables forpoverty incidence, including Gini coefficients (accounting for income inequality) and Human Development Index(accounting for multi-dimensional poverty) were also used in earlier model specifications but resulted to be non-significant.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

B. Loman et al.

structure of Dutch NGDOs, taking into account the percentage of the own funding derived

from (1) private fundraising, (2) joint campaigns and (3) profits from stocks. The higher the

share of own funding, the higher the assumed independency from the back-donors. In the

second place, we also tested the effect of the funding structure in a separate model

using the subsidy index. It takes the share of public funding raised from ‘subsidies

from governments and others public agencies’ in the overall NGDO budget. In this

respect, the subsidy index is a measure of dependency on Dutch government support: a

higher subsidy index indicates more reliance on public funding. Both the independency

index and the subsidy index are measures for the influence of back donor’s interests.

We cannot use, however, both measures in the same statistical model since they are

highly correlated. Therefore, we included both indexes in two separate models to test their

effect.

Other control variables that were included refer to structural parameters of donor and

recipient countries. First, Dutch NGDOs aid lagged for year 2006 measured in constant US

dollars (as reported by OECD-DAC) is used to test for path dependency. We thus assume

that historical aid patterns may influence current aid allocations. Second, the distance

between the donor country and the receiving country is used as control variable, indicating

that geographical distance might influence country choice. Third, we tested for the

similarity between the main religion in receiving countries with the dominant religion in

The Netherlands, looking at the proportion of the population belonging to particular

denominations (Christian, Muslim and other religions), using the Christian religion

majority as reference. The total population (World Bank, 2006) is introduced to control for

size differences between the countries. Finally, we control for differences in size between

NGDOs, measured by the total budged reported for year 2007 (CBF, 2008). Table 1

provides the descriptive statistics for the 7211 combinations of NGDO donor aid with

recipient countries.

3.1 Analytical Methods

We based our analyses on a two-stage Heckman approach which assumes that there is a

difference between the selection of a country as aid recipient and the actual amount sent to

a specific country. We therefore computed two dependent variables: (a) a selection model

for the determinant of country choice, and (b) an allocation model that explains the amount

of NGDO aid allocated in a specific country. The selection variable has values of 0 or 1 to

show whether or not a Dutch NGDO delivers development aid to a country. The allocation

variable contains the actual amount of development aid that was sent to a particular

country. The distribution of the aid allocation variable is rather skewed, since many

NGDO – country combinations do not actually deliver development aid.

The structure of the dataset exhibits several major characteristic that warrant due

attention. First, the cases of NGDO-country combinations are not fully independent from

each other, since some countries receive development aid consistently from a consortium

of several NGDOs. Moreover, some countries are especially targeted for bilateral aid and

thus are also likely to receive more NGDO aid. This implies that the internal structure of the

database is such that any ordinary one-level analysis might deliver biased results. In order

to correct for the non-independence of observations, we clustered the standard errors at

country level in all our analyses.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Contradictory Influence of Donor Dependency

A second concern refers to the most appropriate way for analysing the data. In line with

the available literature we disaggregated that the decision for allocation is a two stage

process, involving (1) the selection of the recipient countries, followed by (2) the decision

regarding the amount of aid sent. Following this approach, selection and allocation can be:

determined by different sets of variables, and the same predictors can have different effects

in the each of the two stages of the process. Even while we acknowledge that in the

literature direct estimates for selection are used in a one stage Tobit procedure

(Nunnenkamp et al., 2008), we consider a two stage approach more appropriate for

separately analysing both decisions.

The statistical procedures used to estimate the selection and the allocation of aid to

certain country rely on probit regression (for the selection stage) and an OLS regression

employed for the subsample of countries that receive aid from Dutch NGDOs. However,

these procedures still have to address the problem of selection bias that results from the fact

that country selection is a non-random process. If we do not take this issue into account our

analysis might lead to biased estimates in the OLS models. In our case, the necessity to

address this selection bias is even more stringent because of the high proportion of

countries that do not receive any aid from Dutch NGDOs or that only receive aid from a

small number of NGDOs. An appropriate response to this problem requires using the

Heckman procedure, that relies on computation of the Inverse Mill ratio as an estimator

that corrects for inherent selection bias.

The models were estimated in STATA 10 software that has incorporated both required

data correction routines: the Heckman selection model and the clustering of standard

errors. Because of the requirement of not using the same variables in the selection and in

the allocation model, we used common religion only in the selection equation. We thus

argue that common religion might be considered more important in the primary selection

process then for the allocation of funds. Shared value attributes between the donor and

receiver (such as colonial history or religion) were also found to be important for

establishing the relation between donors and receivers (Koch et al., 2008), however for

deciding how much funds should be allocated towards selected NGDOs, other factors are

likely to be more relevant.

4 RESULTS

We first present the general results concerning determinants influencing NGDO selection

of target countries to get insight into driving forces for the spatial allocation of

development aid resources by Dutch NGDOs. This analysis reveals to what extent need-

based or donor-driven interests are most critical for NGDO engagement with particular

developing countries. Hereafter, we include specific parameters regarding their

(in)dependency on fundraising and the public subsidies to detect the relative importance

of the NGDO financial structure on country choices. In each case, we distinguish two

(related) decisions: the factor influencing country selection (coefficients of probit model)

and the factors determining the allocation of the amount of resources devoted to particular

countries (using an OLS regression models with correction for selection bias).5

5The coefficients for the Inverse Mill ratio were fully significant in all models, thus warranting the use of theHeckman approach to correct for selection bias.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Tab

le2

.D

eter

min

ants

of

Du

tch

NG

DO

cou

ntr

ych

oic

e(p

rob

itre

gre

ssio

n)

Mo

del

1A

1B

Wit

hin

dep

end

ence

ind

exW

ith

sub

sid

yin

dex

Wit

hin

dep

end

ence

ind

exan

din

tera

ctio

nef

fect

s

Wit

hsu

bsi

dy

ind

exan

din

tera

ctio

nef

fect

s

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Sel

ecti

on

stag

e

GD

Pp

erca

pit

a(l

n)

�0

.095�

0.0

49

�0

.095

0.0

49

�0

.096

0.0

50

�0

.09

50

.04

9

Bil

ater

alai

d(l

n)

0.0

14��

0.0

06

0.0

14��

0.0

64

0.0

16��

0.0

08

0.0

12�

0.0

08

Ex

port

sfr

om

NL

(ln

)�

0.0

05

0.0

39

�0

.055

0.0

38

�0

.004

0.0

38

�0

.00

50

.03

8

Po

liti

cal

reg

ime

(po

lity

ind

ex)

0.0

15��

0.0

07

0.0

15��

0.0

66

0.0

16��

0.0

07

0.0

15��

0.0

07

vD

utc

hN

GD

Oai

d2

00

6(l

n)

0.0

75���

0.0

14

0.0

75���

0.0

14

0.0

75���

0.0

14

0.0

76���

0.0

14

Rel

igio

n(C

hri

stia

nre

f)

Mu

slim

�0

.119

0.0

74

�0

.120

0.0

74

�0

.126

0.0

76

�0

.12

50

.07

5

Oth

erre

ligio

n�

0.0

28

0.0

78

�0

.024

0.0

79

�0

.011

0.0

82

�0

.01

80

.07

9

Bu

dg

etN

GD

O2

00

6(0

00

US

S)

0.0

01���

0.0

00

0.0

0���

0.0

00

0.0

01���

0.0

00

0.0

01���

0.0

00

Po

pu

lati

on

(ln

)0

.143���

0.0

42

0.1

43���

0.0

42

0.1

43���

0.0

42

0.1

42���

0.0

42

Dis

tan

ce(k

m)

0.0

00��

0.0

00

0.0

00�

0.0

00

0.0

00

0.0

00

0.0

00

0.0

00

Ind

epen

den

cein

dex

0.0

60

0.0

61

0.1

03

0.1

44

Su

bsi

dy

ind

ex�

0.1

34

0.0

73

�0

.18

10

.17

5

Inte

ract

ion

ind

epen

den

cyIn

dex

�b

ilat

eral

aid

�0

.003

0.0

10

Inte

ract

ion

sub

sid

yin

dex

�b

ilat

eral

aid

0.0

04

0.0

11

Co

nst

ant

�2

.599���

0.3

63

�2

.556���

0.3

65

�2

.669���

0.3

70

�2

.54

4���

0.3

68

Lo

gp

seu

do

lik

elih

oo

d�

40

02.7

08

�3

99

7.2

1�

39

84

.83

8�

39

94.3

27

Sig

nifi

can

ceat

10

%.

Sig

nifi

can

ceat

5%

.S

ign

ifica

nce

at1

%.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

B. Loman et al.

Contradictory Influence of Donor Dependency

4.1 Country Needs versus Donor Interests?

Looking at the effects of the predictors in the selection model (Table 2, 1A) compared with

the corresponding allocation model (Table 3, 2A) reveals that different independent

variables explain both types of decisions. It appears that the Dutch NGDOs rely on a

wide range of criteria for deciding where the money will go, but when deciding how much

money will be devoted to a specific country things are much simpler: the amount sent

in the previous year is the main criterion that guides the financial allocations in the current

year.

The poverty incidence reflecting the needs of a country is consistently not

significant, neither for the country selection nor for allocation of development aid.

In the selection models, the GDP per capita is only significant at 10%, but it

loses significance in the allocation model. Otherwise, the political regime is far

more important for country selection; it seems that most Dutch NGDOs tend to avoid

too difficult political environments and often select countries with a more democratic

regime.

The size of NGDOs – as measured by its yearly budget – is important in all selection

models. As expected, larger NGDO are active in more different countries. However,

resource allocation towards selected countries does not depend any more on NGDO size.

Whereas larger NGDOs increase the number of countries targeted as recipients, they split

up their funds among the recipients, resulting in comparable smaller allocations. The

country size as reflected by the population also influences the selection as recipient: larger

countries have more chance to be selected. Moreover, distance between the donor country

and the recipient country appears significant in the selection model. It appears that Dutch

NGDOs have preference for more distant countries when selecting recipients for their

funding. Finally, sharing a common (Christian) religion does not influence the probability

of a country to be selected.

The back donor’s interest (e.g. bilateral aid) turns out to be highly significant only in the

selection models, whereas other indicators of economic interest measured through the

amount of exports do not show to have any influence on NGDO aid allocations.

Interestingly enough, the more bilateral aid a country receives the higher its chances to be

selected as recipient for development aid by Dutch NGDOs. This point towards close

coordination between NGDO and public aid operations.

Path dependency plays a dual role in each of the two models. In the selection process it

seems that the more funds a country received in past years, the higher the chance that it will

continue receiving resources the next year. This might indicate a policy of capitalising on

previous collaboration efforts and on-going multi-annual project commitments by the

NGDOs. However, in the allocation model, the more funds a country received from Dutch

NGDOs this will lead to fewer funds allocated in subsequent years. This points towards a

policy of gradual reduction of aid relationships in order to avoid overly dependent

relationships.

Introducing the subsidy index and the (in)dependency index into the selection and

allocation models, we observe a significant effect of both indexes, but only for the decision

to allocate funds to a particular country. The two indexes are strongly negative correlated,

reflecting the fact that the two sources of funding – government and own sources – might be

considered complementary for Dutch NGDOs. The effects of the two indexes in the models

reflect this negative correlation between them: the independency index has a negative

effect on the amount of funds a country receives, whereas the subsidy index has a positive

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Tab

le3

.D

eter

min

ants

of

fun

dal

loca

tio

nb

yD

utc

hN

GD

Os

(OL

Sre

gre

ssio

n)

Mo

del

2A

2B

Wit

hin

dep

end

ence

ind

exW

ith

sub

sid

yin

dex

Wit

hin

dep

end

ence

ind

exan

din

tera

ctio

nef

fect

s

Wit

hsu

bsi

dy

ind

exan

din

tera

ctio

nef

fect

s

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Co

effi

cien

tR

obu

stst

and

ard

erro

r

Aid

allo

cati

on

stag

e

GD

Pp

erca

pit

a(l

n)

0.0

85

0.1

31

0.0

75

0.1

35

0.0

82

0.1

29

0.0

64

0.1

34

Bil

ater

alai

d(l

n)

�0

.026

0.0

19

�0

.027

0.0

21

�0

.066���

0.0

23

0.0

02

0.0

27

Ex

port

sfr

om

NL

(ln

)0

.004

0.0

95

0.0

18

0.1

01

0.0

07

0.0

94

0.0

25

0.0

99

Po

liti

cal

regim

e(a

uto

crac

ies

ref)

An

ocr

acie

s0

.385

0.2

13

0.3

43

0.2

19

0.4

18��

0.2

07

0.3

98

0.2

10

Dem

ocr

acie

s0

.031

0.2

32

�0

.009

0.2

39

0.0

55

0.2

31

0.0

46

0.2

37

Du

tch

NG

DO

aid

20

06

(ln

)�

0.0

91���

0.0

34

�0

.104���

0.0

38

�0

.085��

0.0

34

�0

.103���

0.0

38

Bud

get

NG

DO

20

06

(00

0U

S$

)0

.000

0.0

00

0.0

00

0.0

01

0.0

00

0.0

00

0.0

00

0.0

00

Po

pu

lati

on

(ln

)�

0.1

59

0.1

07

�0

.187

0.1

14

�0

.166

0.1

06

�0

.196

0.1

12

Dis

tan

ce(k

m)

0.0

00

0.0

00

0.0

00

0.0

00

0.0

00

0.0

00

0.0

00

0.0

00

Ind

epen

den

cein

dex

�1

.103���

0.2

37

�2

.006���

0.4

75

Su

bsi

dy

index

0.8

95���

0.2

67

1.7

97���

0.5

83

Inte

ract

ion

indep

end

ence

index

�b

ilat

eral

aid

0.0

76��

0.0

34

Inte

ract

ion

sub

sid

yin

dex

�b

ilat

eral

aid

�0

.077

0.0

43

Con

stan

t1

8.3

77���

1.0

08

17

.87

3���

1.0

89

18

.80

8���

1.0

29

17

.47

8���

1.0

67

Sourc

es:

bas

edo

nO

EC

D-D

AC

and

Worl

dB

ank

dat

a.R

elig

ion

isth

euniq

ue

var

iable

that

dif

fere

nti

ates

bet

wee

nse

lect

ion/a

lloca

tion

model

.S

ign

ifica

nce

at1

0%

.S

ign

ifica

nce

at5

%.

Sig

nifi

cance

at1

%.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

B. Loman et al.

Contradictory Influence of Donor Dependency

effect. We can interpret this as follows: the more an NGDO relies on its own

sources of finance, lower amounts of aid will be allocated to particular countries that

already receive bilateral aid. Otherwise, the more NGDOs rely on governmental funding,

the higher their capacity to allocate larger amounts of development aid to a particular

country.

4.2 Interdependencies Between Bilateral and NGDO Aid

We also wanted to test whether the NGDO status of being more or less (in)dependent in

terms of the funding sources is important in conjunction with the variables that measure

the interest of back donors. The results provide insights in the complex relation between

the back donor country policies and the relative independence that NGDOs enjoy

regarding their funding and allocation decisions. Tables 2 and 3 (1B and 2B) summarise the

findings of the extended model that accounts for interactions between bilateral and NGDO

aid.

The hypotheses that is subject of analysis refers to (1) whether NGDOs that are more

independent from their back donor in terms of their sources of funding take their selection

or allocation decisions based on different criteria, and (2) whether the condition of being

less dependent on bilateral funds leads to different weight of back donor’s interests. We

used an interaction term between the subsidy or independency index with bilateral aid

expenditures, a variable that was found to be significant in the process of selection recipient

countries. If the interaction term is significant it provides support for both our hypotheses.

As before, we permit that this interaction might result in different effects between the

selection and the allocation stages.

We first notice that the interaction effect between the subsidy-index and bilateral aid is

not significant, and that the performance of the model resembles the outcomes without

interaction term. It seems that the influence of receiving funding from the government does

not differentiate between NGDOs with respect to their decisions regarding country choice

and resource allocation.

However, when we analyse the effect of the independency index (i.e. the share of own

funding in the total NGDO budget) we observe some different outcomes. While the

selection stage is still consistent with the model without interaction terms, we do find

changes in the parameters of the allocation stage. It appears that the more an NGDO

depends on external funding, the country characteristics for fund allocation of the back

donor become highly relevant. The interaction effect is, however, positive, indicating that

when more bilateral aid is allocated in a specific recipient country, less development aid

will be allocated by NGDOs from their own resources. In other terms, NGDOs that do not

have own funding (thus relying on subsidies and donor funding) will select a country based

on how much bilateral aid it receives. However, they will allocate more of their own funds

towards countries that receive lower levels of bilateral aid. This can be interpreted as a

policy of strategic NGDO aid allocation that tries to avoid too close relationships with

bilateral donor activities.

Finally, addressing the situation of NGDOs that raise own funding (meaning that the

independency index >0), the role of bilateral aid relationships tends to becomes gradually

more important. NGDOs that depend more on their own resources are increasingly orienting

their allocations in line with bilateral aid when deciding how much aid to allocate to a

particular country. We can further analyse this relationship by writing the NGDO aid

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

-1,0000,0001,0002,0003,0004,0005,0006,0007,0008,000

0 10 3020 40 50 10090807060

Independence Index

Effect of Bilateral Aid on NGDO Aid

Eff

ect

of

Bila

tera

l Aid

Figure 1. Effect of bilateral aid on NGDO aid allocation

B. Loman et al.

allocation Equation in the following way in order to explain the effect of the interaction term:

AidAllocationðlnÞ ¼ b0 þ b1 � indep:index þ b2 � bilat:aidðlnÞ þ b3

� indep:index � bilat:aidðlnÞ þ bnXnþ 2(1)

where aid allocation and bilateral aid are measured in natural logarithms, Xn represents the

other structural variables in the models with their corresponding b effects and s is the error

term. When the independency index is 0, NGDOs have no own funding and thus the terms

for b1� indep.index and b3� indep.index� bilat.aid(ln) become 0. In that case, we have a

significant (negative) effect from bilateral aid on NGDO allocations, indicating that the

more bilateral aid is delivered to a chosen recipient country, the less NGDO development

aid will be allocated. However, when we address the situation of an NGDO with own funds,

meaning that the independency index is higher than 0, the effect of the bilateral aid can be

expressed mathematically in the following form:

AidAllocationðlnÞ ¼ b0 þ b1 � indep:index þ bilat:aidðlnÞ� ½b2 �þb3 � indep:index� þ bnXnþ 2

(2)

Using this equation, we can easily assess the effect of bilateral aid (the back donor’s

interest) for various degrees of own funding of the NGDOs. Calculating the effect of the

bilateral aid for various degrees of own funding results in the conclusion that the more an

NGDO is depending on its own financial resources, the more the initial negative effect of

the bilateral aid upon NGDO aid allocation decisions to particular country becomes

reduced. This is illustrated in Figure 1, where the alignment with bilateral aid is reinforced

when independence in funding increases. Note that for a value of 50% for the independency

index, the effect of bilateral aid is 3.7 instead of �0.0658 for the situation where the

independency index equals to 0 (see Table 2, Model 2B), while for a value of the

independency index of 100 the effect of the bilateral aid increases to 7.353. This indicates

that the more an NGDO is own-funded the more it will be inclined to follow the bilateral

back donor’s priorities in deciding how much development aid will be allocated to

particular countries.6

6Another interesting outcome refers to the fact in the model with interaction term the parameters for the politicalregime of a country tend to be modified. Countries being ‘anocracy’ increase the allocation compared to countriesthat are ‘autocracies’. Taking into consideration the effect of political regime in the selection and allocationequations, we can conclude that the NGDOs prefer to be present in more democratic countries (thus avoidingautocratic regimes) but they allocate their funds mainly in the intermediary regimes (thus not directing mostfunding to the democratic countries).

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Contradictory Influence of Donor Dependency

Public agencies thus have more influence on the NGDO decisions for allocating aid to

certain countries apart from their usual incidence through conditional co-funding. Even

with growing NGDO independency from back-donors, their aid allocation priorities are

likely to differ less.

5 CONCLUSIONS AND OUTLOOK

Funding structures of Dutch NGDOs show substantial diversity in terms of their

dependency on public donor support and private fundraising. Most agencies rely on a

mixture of public and private resources and thus have to respond in their operations towards

different audiences. Consequently, aid allocation patterns are likely to exhibit similar

variability with respect to country choice and the degree of alignment with back-donor

priorities. Whereas poverty incidence is generally advocated as an important driver for

country choice, the prevailing political regime and the bilateral aid flows appear as most

important predictors for NGDO country choice. Moreover, path dependency is clearly

confirmed in the sense that past NGDO aid allocations lead to higher current commitments.

Finally, whereas Dutch NGDOs do engage more often with countries where substantial

Dutch bilateral aid is allocated, there is a clear trend of diminishing the volume of

allocations in these countries, thus avoiding full harmonisation of development aid

activities and maintaining an independent profile.

Empirical analyses of the driving forces for aid allocation usually suffer from several

methodological constraints related to selection bias of the sample and non-independence of

observations. We therefore relied on a two stage approach for analysing the sequential

decision process of the country selection and the aid allocation of NGDOs. This procedure

provides new and detailed insights in the underlying determinants for country choice.

While the initial selection process is relatively straightforward involving more or less

objective political and socio-economic criteria that a country should meet in order to be

selected as recipient for Dutch NGDO aid, for the determination of the aid resource

allocation the role of the internal NGDO funding structure and its dependency on public

donor support or private fundraising becomes far more important.

Using the independency index and the subsidy index as indicators for the reliance on,

respectively, own fundraising and public donor support, we find strong and significant

effects on NGDO fund allocation decisions. Against our expectations, NGDOs with higher

independence from back-donors are likely to allocate less funding in bilaterally selected

partner countries, whereas NGDOs that rely more on private fundraising select an aid

allocation pattern that is better aligned with public aid agencies. This implies that

complementarities between public and NGDO aid are easier reached with agencies that are

less dependent on back-donor funding.

We explain this consistently registered inverse relationship between NGDO

independence and bilateral aid priorities against the background of new theories regarding

the complex motivation structure for public–private alliances. NGDOs relying on private

fundraising and campaigning only have to show to the general audience some global

indicators of development progress that are likely to be generated in countries where

bilateral donors are widely present. On the other hand, publicly funded NGDOs are subject

to far more stringent outcome monitoring and impact assessment and therefore tend to

focus their activities in countries where they can more easily demonstrate their unique

performance attributes.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

B. Loman et al.

In policy terms, our results seem to indicate that donors should not expect too much from

their financial leverage in co-funding NGDO programmes for reinforcing the desired

complementarities between public and private development efforts. For reaching the

objectives of aid harmonisation, due attention should be given to the inherent objective

structure of mixed-funded NGDOs where priorities of risk-avoidance and identity-seeking

are carefully balanced against the background of their required performance indicators.

ACKNOWLEDGEMENTS

Data collection for this research has been conducted within the framework of the IS

Academy programme ‘Civil Society organisations’ co-funded by the Dutch Ministry of

Foreign Affairs. The NGDO database is publicly accessible at www.ngo-database.nl.

Research assistance by Mrs. Ioana Pop is gratefully acknowledged.

REFERENCES

Beckers R. 2003. Trust, accreditation, and philanthropy in the Netherlands. Nonprofit and Voluntary

Sector Quarterly 32(4): 596–615.

CBF. 2008. Database Central Bureau of Fundraising. Available at http://www.cbf.nl/Database_

goede_doelen/1_zoek_vergelijk.php [Accessed on 12 December 2008].

Clegg SR, Rhodes C, Kornberger M. 2007. Desperately seeking legitimacy: organizational identity

and emerging industries. Organization Studies 28(4): 495–513.

Dreher A, Molders F, Nunnenkamp P. 2007. Are NGDOs the Better Donors? A Case Study of Aid

Allocation for Sweden. Working Paper No. 1383. Kiel Institute for the World Economy.

Dudley L, Montmarquette C. 1976. A model of the supply of bilateral foreign aid. American

Economic Review 66: 132–142.

Edwards M, Hulme D. 1997. Too close for comfort? The impact of official aid on NGDOs. In:

M, Edwards D Hulme (eds).NGDOs, State and Donors, Too Close for Comfort? Macmillan Press:

London.

Fowler A, Biekart K. 1996. Do private agencies really make a difference? In: D Sogge (ed).

Compassion and Calculation: The Business of Private Foreign Aid. Pluto Press: London.

Koch DJ. 2009. Aid from International NGDOs, Blind Spots on the Aid Allocation Map. Routledge:

Abingdon.

Koch DJ, Loman B. 2009. ‘The geographic choices of Dutch. NGDOs’. In: P Hoebink (ed). Dutch

Yearbook of International Cooperation. van Gorcum: Assen.

Koch DJ, Dreher A, Nunnenkamp P, Thiele R. 2008. Keeping a Low Profile: What Determines the

Allocation of Aid by Non-Governmental Organizations? Working Paper No. 1406. Kiel Institute

for the World Economy.

Mascharenhas R, Sandlers T. 2006. Do donors cooperatively fund foreign aid? International

Organizations Review 53(1): 589–627.

Masud N, Yontcheva B. 2005. Does Foreign Aid Reduce Poverty? Empirical Evidence from

Nongovernmental and Bilateral Aid. IMF Working Paper 05/100. Washington, D.C., IMF.

McKinley RD, Little R. 1977. A foreign policy model of US bilateral aid allocation. World Politics

30(1): 58–86.

Nancy G, Yontcheva B. 2006. Does NGO Aid Go to the Poor? Empirical Evidence from Europe. IMF

Working Paper No. 06/39. Washington D.C.: International Monetary Fund.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid

Contradictory Influence of Donor Dependency

Nunnenkamp P, Weingarth J, Weisser J. 2008. Is NGDO aid not so different after all? Comparing the

allocation of Swiss aid by private and official donors. Working Paper 1405. Kiel Institute for the

World Economy.

OECD-DAC. 2008. International Development Statistics. Paris: Organisation for Economic Co-

Operation and Development (OECD), Development Assistance Committee (DAC).

Rogerson A. 2005. Aid harmonisation and alignment: bridging the gaps between reality and the Paris

Reform Agenda. Development Policy Review 23(5): 531–552.

Riddell RC, Robinson M. 1995. Non-governmental Organizations and Rural Poverty Alleviation.

Clarendon Press: Oxford.

Rowlands D, Ketcheson I. 2002. Multilateral aid coordination by the international financial

institutions: an examination of Canadian development assistance to sub-Saharan Africa. In:

BA Mark (ed). New Perspectives on Foreign Aid and Economic Development. Praeger: Westport,

Co.

Ruben R, Schulpen S. 2009. How to allocate public funding to nongovernmental development

organizations: a critical assessment of the Dutch co-financing system. Nonprofit and Voluntary

Sector Quarterly 38(2): 287–306.

World Bank. 2006. World Tables. World Bank: Washington D.C.

Yescombe ER. 2007. Public Private Partnerships: Principles of Policy and Finance. Elsevier Science

& Technology: Oxford.

Copyright # 2010 John Wiley & Sons, Ltd. J. Int. Dev. (2010)

DOI: 10.1002/jid