The macroeconomic effects of adjustment lending: A review and evaluation

Abstract

Does external aid have an impact on economic growth, if so, through what channels? Does the domestic policy environment matter? In the past ten years, these questions have been the subject of a large number of empirical studies. This article surveys these studies and identifies the areas where there is consensus and those where there is not. It is argued that to move forward, empirical work must be based more explicitly on structural models of the link between aid and growth, notably by incorporating relationships between aid, government fiscal responses, private sector investment, and economic growth.

Keywords:

• Adjustment lending
• Aid and growth
• Private sector investment
• JEL Codes: E6

Notes

Burnside and Dollar (2000). A widely circulated version of the paper appeared as early as 1997.

World Bank (1998).

See references in Easterly et al. (2003, footnote 1) to a speech by President Bush and articles in the Economist and the Washington Post.

The Aid measure they use is a new measure of effective development aid (EDA) developed in Chang et al. (1998). This measure attempts to measure the grant component of capital flows more accurately than the ODA measure conventionally used. The section on ‘More Theory‐based Specifications’ contains a critical discussion of the aid measures used in this literature.

Note that Burnside–Dollar assert that the significance of the (Aid)² term is highly dependent on the presence of outliers in the data in the Aid dimension.

Some of the relevant literature is discussed in the section ‘Who Receives Aid.’

G^c (Gⁱ) can be thought of as the sum of expenditures on different categories of consumption‐type (investment‐type) goods.

McGillivray and Morrissey (2001) contains a thorough survey of the literature dealing with points 1 and 3 below.

See McGillivray and Morrissey (2001) for a comprehensive survey.

It is also important to investigate possible changes over time in fiscal policy responses to aid.

See McGillivray and Morrissey (2000).

Table 2 in McGillivray and Morrissey (2001) contain an overview of results of seven recent fungibility studies.

It is not possible of course to include all items of the budget constraint in the regression since this would lead to perfect colinearity among the regressors. Hence, the effect of any one of the included variables has to be interpreted as the impact relative to the excluded variable.

Since we did not have access to the full regression results, only point estimates could be calculated in certain cases, identified by NA in the column containing t‐values. The reason is that the point estimates represents the difference (or sum) of two regression coefficients. The standard error of this difference (or sum) involves not only the standard error of each individual coefficient estimate but also their covariance. We did not have access to estimates of these covariances.

The database goes under the acronym MONA, which stands for Monitoring of IMF Arrangements.

Poverty reduction is probably a better indicator of the ultimate objective of these institutions, but it can be argued that high growth and low inflation contribute significantly to reducing poverty. See Dollar and Kraay (2002).

Ul Haque and Khan (1998) and Bird (2001) contain surveys.

The + and − signs indicate whether the variable has a positive or negative influence on the probability of success of the program.

Branson and Hanna (2000) extend the empirical specification of Dollar and Svensson (2000) by treating countries that have gone through multiple programs as one observation, and including previous evaluation results as an additional explanatory variable. They find that success (failure) in a previous program is a good predictor of success (failure) in subsequent ones. This suggests that countries can build up reputation as good or bad program performers that dissipate only gradually over time.

See, for example, Dhonte (1997).

See Bird & Rowlands (2002) for a systematic survey.

Barro & Lee (2001) is an exception in that they use an instrumental‐variables estimation technique to control for the possible endogeneity of the decision to enter an IMF program. Their main concern is however not about the effect of IMF programs on access to the international capital market but on economic growth. They find that there is no such effect.

Recall the sensitivity of the Burnside–Dollar results to the chosen specification and sample period.

Svensson (2003) provides some related evidence. He shows that there is no relationship between the percentage of committed funds in adjustment programs that is disbursed and the reform effort of the recipient.