The Impact of Aid on Total Government Expenditures: New Evidence on Fungibility

DOIhttp://doi.org/10.1111/rode.12286
Published date01 August 2017
Date01 August 2017
The Impact of Aid on Total Government
Expenditures: New Evidence on Fungibility
Łukasz Mar
c*
Abstract
Aid is said to be fungible at the aggregate level if it raises government expenditures by less than the total
amount. This happens when the recipient government decreases domestic revenue, decreases net
borrowing, or when aid bypasses the budget. This study makes three contributions to both fungibility and
fiscal response literature. First, fungibility at the aggregate level is re-examined on a larger recent panel
1980–2012, distinguishing between short- and long-term impact of aid. The results indicate that aid is
partly fungible in the long run and highly fungible in the short run. Second, to account for aid bypassing
the budget, technical cooperation is used as a proxy for off-budget aid. Off-budget aid is found to be
non-fungible and on-budget aid is partly fungible. Third, fungibility of bilateral and multilateral aid is
analyzed: the results indicate lower fungibility of multilateral aid.
1. Introduction
Over the last 40 years aid
1
has accounted for a substantial part of developing
countries’ gross domestic product (GDP) and has been one of the main sources of
government revenues in these countries. The empirical literature has tried
(especially from the mid-1990s) to quantify aid’s impact on growth and well-being
in developing countries. The results were mixed in the studies up to 2009 (for
reviews and discussion see, e.g. Tarp, 2006; Roodman, 2007; Arndt et al., 2010).
However, the most recent literature (see Arndt et al., 2010, 2011, 2015; Alvi and
Senbeta, 2012; Juselius et al., 2014; Mekasha and Tarp, 2013; Galiani et al., 2014)
mostly reports a positive impact of aid, although smaller than expected in the 1970s
and 1980s.
2
A large part of aid goes to the public sector. Therefore, to better
understand the impact of aid on growth and poverty reduction, it is crucial to
determine the impact of aid on the behavior of one of the major actors in the
development process, i.e. the recipient government.
Two strands of literature focus on the impact of development assistance on the
behavior of recipient governments: fungibility studies and fiscal response studies.
*Mar
c (Corresponding author): Overseas Development Institute, Research and Policy in Development
(RAPID), 203 Blackfriars Road, London, SE1 8NJ, UK. E-mail: l.marc@odi.org.The author is also
affiliated to Vistula University, Faculty of Business and International Relations, 3 Stokłosy Str., 02-787,
Warsaw, Poland. The author is especially grateful to Chris Elbers and Jan Willem Gunning for many
useful comments and insights during numerous talks when he worked at the Vrije Universiteit (VU)
Amsterdam and Tinbergen Institute. He is also grateful to Maarten Lindenboom, Nicolas van de Sijpe,
Sweder van Wijnbergen, anonymous referees, and seminar participants at Tinbergen Institute, VU
Amsterdam, Warsaw Economic Seminar, Ruhr Graduate School in Economics (RGS) conference, the
Centre for the Study of African Economies (CSAE) conference, and African Economic Conference for
useful comments at the early stage of this research, and to Lisette Swart, Melinda Vigh, Jannis Visser,
Lisa Winkler, Marcin Zamojski, and especially to Zlata Tanovi
c, for other help. The author is grateful to
the United Nations University World Institute for Development Economics Research (UNU-WIDER)
for allowing him to use WIDER’s premises and network to finalize this article, and to Jukka Pirttil
a,
Channing Arndt, Miguel Ni~
no-Zaraz
ua and internal seminar participants for useful suggestions. This
article is a thoroughly revised version of an earlier paper (Mar
c, 2012). The usual disclaimers apply.
Review of Development Economics, 21(3), 627–663, 2017
DOI:10.1111/rode.12286
©2016 John Wiley & Sons Ltd
This article contributes to both. Fungibility is a broad term that describes situations
when recipients respond to aid by changing the way they use their own resources.
Aid is called fungible if the provision of goods and services intended by the donor
is not fully achieved, because aid was used for other purposes.
3
McGillivray and
Morrissey (2004) and Morrissey (2014) distinguish three types of fungibility: general
fungibility, sector (categorical) fungibility and the additionality of aid.
4
General
fungibility describes a situation when aid intended to finance public investment is
diverted to government consumption. Aid is fungible on a sector (categorical) level
if it is intended for one sector, but at the margin finances expenditures in other
sectors.
This study focuses on the third aspect of fungibility: the additionality of aid.
5
It
investigates by how much an additional euro of aid increases total government
expenditures. At the aggregate level,
6
aid is fungible when one additional euro of
aid increases total government expenditures by less than one euro and fully
fungible when government spending does not increase at all. This can happen when
aid channeled through the budget substitutes for other sources of government
revenue. This way the government is able to decrease taxes, decrease borrowing, or
increase its reserves.
Morrissey (2014) points out that there is surprisingly little evidence on the
additionality of aid. Although there is some evidence that aid is partly fungible at
the aggregate level, the extent of fungibility of aid and its components and the
importance of fungibility for the development process are still under debate. Yet,
the question whether aid increases government expenditures or is used to decrease
taxes or borrowing needs matters for at least three reasons. First, it may be a
crucial channel to look at when investigating the impact of aid on growth. Second,
donors are usually interested in financing increased expenditures rather than tax
decreases or reductions in recipient’s borrowing. Third, there is an ongoing
discussion that aid may have a negative impact on the quality of institutions when it
substitutes for own government’s sources of revenues and reduces incentives to
collect taxes (Carter, 2013; Moss et al., 2008).
This study provides evidence on the extent of fungibility at the aggregate level,
using a rich dataset of 118 countries for the period 19802012. I take into account
the dynamic properties of the process, discuss endogeneity of aid in regressions and
explicitly distinguish between the short- and long-term impact of aid. The focus of
the fungibility literature is on the short-term effects of aid: it tests whether aid
increases government expenditures in the same year. However, aid may also have a
long-term effect on the level of government expenditures. Investment in, for
example, water supply and sanitation may reduce water-related diseases and, in the
long term, decrease health spending and total government expenditures. Moreover,
since aid is usually committed for a longer period, donors may be more interested
in a long-term impact of aid on government behavior rather than short-term
adjustments.
The inclusion of lagged government expenditures and data averaged over several
years allows one to estimate the long-term impact of aid on government
expenditures. Furthermore, a change in government expenditures may be different
depending on whether aid is increasing or decreasing, especially if the change in aid
disbursement is unexpected. This study also tests the hypothesis of symmetric
response to increases and decreases of aid.
If government expenditures do not increase by the amount of aid disbursed, it
follows from the government budget constraint that domestic revenues adjust, or
628 Łukasz Mar
c
©2016 John Wiley & Sons Ltd
the net borrowing adjusts, or that aid bypasses the budget. This study investigates
two out of these three channels: off-budget aid and government revenues.
7
A large part of aid is never recorded in the budget, either because it is spent in
the donor country, or because it directly reaches ultimate beneficiaries. The
differences between these two components of aid should be taken into account, as
on-budget aid has a direct impact on government expenditures, whereas the impact
of off-budget aid is indirect because it is not recorded in the budget. Van de Sijpe
(2013b) made the first attempt at distinguishing between recorded and unrecorded
aid flows at the sectoral level. I follow this approach at the aggregate level and
estimate the impact of on- and off-budget aid on government expenditures.
The impact of aid on government revenues is especially interesting for donors, as
they are worried that aid may discourage tax effort and lead to lower domestic
revenues of the government. This study provides additional evidence on the impact
of aid on total government revenues and contributes to the fiscal response
literature.
Fungibility of aid has been tested at the aggregate, sectoral and regional level. To
my knowledge, the fungibility of bilateral and multilateral aid has not been
investigated using data that became available since 1990. Yet, bilateral and
multilateral aid differ for at least three reasons (Ram, 2003): (i) donor motives, (ii)
aid conditionalities, and (iii) closeness of the relationship between the donor and
the recipient. This study fills this gap.
The paper proceeds as follows. The next section focuses on a literature review.
Section 3 presents metho dology, data and the choice of control variables. Afterwards,
the results for aggregate aid are presented in section 4. Section 5 discusses and
presents results for off- and on-budget aid, and section 6 for bilateral and multilateral
aid. The robustness of the results is tested in section 7. Section 8 concludes.
2. Literature Review
There is a large literature that investigates categorical and general fungibility.
8
Generally, aid is found to be fungible at both sector and general levels, however
Morrissey (2014) points out that the extent of general and categorical fungibility is
overstated because part of the aid never reaches the recipient country’s budget.
Fungibility detected at the sectoral or regional level does not mean that aid is
fungible at the aggregate level. Aid intended for the health sector that is diverted
to roads may still increase total government expenditures by the full amount. In
this case aid is not fungible at the aggregate level. For example, Devarajan et al.
(2007) report fungibility of aid at the sectoral level, while finding that at the
aggregate level each dollar of aid leads to a 90 dollar cent increase in government
expenditures in a sample of 18 sub-Saharan African countries in the period 1971
1995. In a similar study, Feyzioglu et al. (1998) show that for their sample of 14
countries aid is not fungible at the aggregate level and that there is no tax relief
that could be associated with fungibility. However, the number of observations is
very small and, additionally, this result is not robust. Namely, when the number of
countries is increased, aid appears to be fungible.
9
Remmer (2004) uses an error
correction model and finds a positive relationship between aid and government
expenditures both in the short and in the long term. Chatterjee et al. (2012) analyze
fungibility at the aggregate level and the impact of disaggregated aid (into
investment aid, non-investment aid and social infrastructure aid) on various
measures of government expenditures for years 19722000 and a group of 67
AID IMPACT ON TOTAL GOVERNMENT EXPENDITURES 629
©2016 John Wiley & Sons Ltd

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT