6. New look at the aid-growth literature
Contrary to the two previous analyses, we prefer to divide the
aid-growth literature in two major periods separated by two almost simultaneous
turning points in the second half of the nineties. If we want to clarify a
little bit the debate about aid effectiveness, we need to focus on the recent
literature published after the mid-nineties. This does not mean that earlier
articles are useless but they were quite contradictory and have been abundantly
criticised. Nevertheless, we think necessary to discuss this early literature
despite all its shortcomings to understand the transition to the last
generation of studies.
14 Mosley (1980), Mosley et al. (1987) and Boone (1994) were then
opposed to Gupta & Islam (1983) and Levy (1988)
6.1. The first part of the literature
In the early nineties, after a few decades of inconclusive
assessments of aid, emerged some kind of «micro-macro paradox» as
named by Mosley (1987). There was indeed a striking contrast between the
relative optimism of a majority of studies conducted at the micro level and the
relative pessimism regarding the ability of aid to increase growth performance
at the country scale. Projects appeared to be successful whereas there was no
visible impact of aid at the macro level. At that time, a few pessimistic
studies such as Bauer (1993) have been particularly influential in promoting
the idea that aid is wasted or even detrimental for recipient's
development15.
If we except Hansen and Tarp (2000, p16) which is, to our
knowledge the only article to consider that «A very few highly influential
studies in each generation have argued the negative», the majority of
authors agree about the pessimistic evolution of the first decades of
aid-growth research:
McGillivray (2006, p16): «For many decades the research
literature on the country-level impacts of aid often sent ambiguous messages as
to whether aid was effective in promoting growth and reducing poverty... The
overall consensus about these impacts was rather pessimistic ... «
Clemens et al. (2004, p7): «(after a decade of
controversy)... Many researchers have taken Boone (1996)'s findings as
confirmation of the «macro-micro paradox»... «
Kanbur (2003, p16): «Throughout the assessments of aid
that have been carried out over the last few decades, there seems to have
emerged a micro-macro paradox.»
Lensink and White (1999, p64): «Overall, the strict
macroeconomic studies... seem to be inconclusive with respect to the
effectiveness of aid... Most authors argue that the macroeconomic impact of aid
was modest or that aid did not have any affect on economic growth.»
Hudson (2004, p1 85): «Hence, until recently, the
dominant view within economics had become pessimistic with respect to the pas
effectiveness of aid...»
In the early nineties, the «micro-macro paradox» due
to the general pessimistic conclusions of macro studies led to an unprecedented
crisis of legitimacy of aid to development. In consequence the overall aid
flows to developing countries fell drastically. Nowadays this paradox is well
understood. Indeed, a particular project may reach its own objectives and
create simultaneously important distortions. If the evaluation at the local
level is not
15 Schwalbenberg (1998) argues that their evidence is largely
anecdotal and denies their finding that aid has led to the adoption of
damageable economic policies.
conducted at «shadow price» its conclusion may be
positive even if the global balance sheet is in deficit.
Moreover, the high number of pessimistic macro studies can be
explained as well by analysing the framework of this early literature. Growth
models have been important in influencing the specification of empirical
relations. Until recently, the most widespread models used to examine the
macroeconomic impact of aid were the different types of «gap
models»16. In short, these models consider the availability of
capital goods as the constraining factor to enhance economic growth in less
developed countries. In this sense, aid may spur growth by raising investment.
In other words, foreign aid fills the gap created by insufficient savings or
exports.
This early literature has been extensively criticized. Apart
from the poor quality of the data available for these different articles, the
criticisms focused essentially on two unrealistic assumptions of the «gap
models». The first problematic hypothesis concerns the assumed linear
relationship between investment and growth through a constant capital-output
ratio. As Easterly (2003, p31) emphasises: «Most economists since Solow
(1957) have felt uncomfortable with a Leontief-style production function that
does not allow the substitution of labour for capital.»
The second critical assumption of the «gap models»
has been exposed by Boone (1996). It is the strong hypothesis that aid
necessarily finances investment rather than consumption. Morrissey (2005)
points out that there is probably not more than one third of total foreign aid
devoted to productive investment. The largest part of aid is actually dedicated
to health expenditures, education, emergency aid or debt cancellation which
will have very little impact on growth in the short run.
As a consequence of these two core assumptions, until the
mid-nineties, economists only tested linear relationships between aid and
growth despite the improvements in growth literature. As Durbarry et al. (1998,
p1) document: «Many aid-growth investigations ignore many of the recent
advances in growth theory». Easterly (2001) tests the credibility of
such
16 There exist different variant of the «gap
models». The most employed one is the so called «two gap model»
of Chenery and Stout (1966), combination of the trade gap model (Balassa, 1964)
and of the saving gap model (Rosenstein-Rodan, 1969; Fei and Paauw, 1965).
gap models for a sample of 88 recipient countries. For only one
case, Tunisia, the «gap» approach seems to hold empirically.
There are definitely many reasons to expect a nonlinear
aid-growth relationship. Diminishing return of aid is actually very plausible.
First, Hadjimichael et al. (1995), Durbarry et al. (1998) or Lensink and White
(1999) refer to the limited absorptive capacity of recipient
countries17. Hence, there can be «too much» aid intended
for a particular country if aid inflows can not be productively utilized. This
can for example happen if country's management capacities are not sufficient.
Another explanation of these diminishing returns could be the fact that aid
does not flow neutrally into a country. The recipient economy can be made worse
of because of the immiserization problem18. This may happen
when domestic inputs required for a particular project are greater than its net
output. In other words, this is the case of a very costly project leading to a
net loss for the «beneficiary» country19.
Secondly, Lensink and White (1999) present several early
studies pointing at a negative relationship between foreign aid and domestic
savings. They also argued that development aid could enable countries to hold
an overvalued currency and thereby stimulating capital flight and depressing
the competitiveness of a country. Finally, aid may encourage unproductive
public expenditures so that the positive impact of foreign assistance could be
off set by government reaction. This has been called the fungibility
problem20. All these arguments demonstrate the weak theoretical
background and the abundant empirical failings of the first part of the
literature.
Therefore, in the early nineties, the only realistic statement
was given by White (1992, p121) «We know surprisingly little about aid's
macroeconomic impact. The combination of weak theory with poor econometric
methodology makes it difficult to conclude anything about the relationship
between aid and growth.»
17 There can be first diminishing and even negative returns to
aid. This can be illustrated by the aid of a Laffer curve. Estimates of the
turning point at which the impact of aid on growth turns negative vary
extremely: Some authors (Dalgaard and Hansen, 2001; Burnside and Dollar, 1997)
estimated this turning point below 20% of GDP whereas others (Lensink and
White, 2001; Gomanee et al., 2002) found it higher than 40%.
18 The famous «Dutch disease» is a particular case of
this problem. (See Kanbur, 2003, p1 1)
19 This could be the case of development agencies' projects
requiring important co-financings from the «beneficiary» country but
leading to relatively small output.
20 However Lensink and White (2000) underline that fungibility is
not necessary a problem if recipient countries pursue appropriate growth
objectives.
| 
