![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
This article examines empirically the proposition that aid to poor countries is detrimental for external competitiveness, giving rise to Dutch disease type effects. At the aggregate level, aid is found to have a positive effect on growth. A sectoral decomposition shows that the effect is (i) significant and positive in the tradable and the nontradable sectors, and (ii) equally strong in both sectors. The article thus provides no empirical support for the hypothesis that aid reduces external competitiveness in developing countries. A possible reason for this finding is the existence of large idle labour capacity that prevents the real exchange rate from appreciating.
Acknowledgements
An earlier version greatly benefited from comments and suggestions from Carl-Johann Dalgaard, Priscilla Mothoora, Thomas Rønde, and Finn Tarp. We are indebted to David Roodman, who generously provided access to his database. All remaining errors are our own.
Notes
1. See Svensson (Citation2000) and Djankov, Montalvo and Reynal-Querol (Citation2006), for example.
2. Our empirical analysis employs aggregate aid data averaged over 4-year periods. In doing so, we miss two important recent developments in the aid effectiveness literature. First, aid heterogeneity has been shown to matter for aid effectiveness (e.g., Clemens et al., Citation2004; Ouattara and Strobl, Citation2008). In the present context, aid for infrastructure may, for example, exert only limited pressure on the real exchange rate to the extent that it helps expand productive capacity. Second, the volatility of foreign aid may have a negative impact on economic growth (e.g., Lensink and Morrissey, Citation2000; Hudson and Mosley, 2009). In particular, volatile aid is likely to impair macroeconomic management, including management of the real exchange rate. Accounting for aid heterogeneity and aid volatility would thus definitely constitute a promising extension of the present analysis.
3. We thank an anonymous referee for pointing us to this issue.
4. The interaction terms reflect the second-order effects considered in the aid-growth literature. In terms of the model in Dalgaard, Hansen and Tarp (Citation2004), these effects correspond to Burnside and Dollar's (Citation2000) claim that aid works with reasonable policies (
); Hansen and Tarp's (Citation2000) suggestion that aid exhibits diminishing returns (
); and Dalgaard, Hansen and Tarp's (Citation2004) finding of higher aid effectiveness with better geographic/climatic conditions (
)
.
5. Effective aid is defined as the grant equivalent of official disbursements constructed by Chang, Fernandez-Arias and Serven (1998), calculated as the sum of official grants and the grant element in concessional loans.
6. Other factors such as changes in relative prices may of course also affect sectoral growth rates but we omitted them because reliable proxies can hardly be constructed in a panel data context.
7. These are Aid/GDP, lagged; (Aid/GDP) squared, lagged; (Policy x Aid/GDP), lagged; Policy x (log Initial GDP per capita); Policy x (log Initial GDP per capita) squared; Policy x (log Population); and a dummy for countries in the Central Francophone Africa zone. This instrumentation strategy is described and motivated in detail in Dalgaard, Hansen and Tarp (Citation2004), but in general it is aimed to reflect donors' overall preference to send aid to the smallest and poorest countries, those with better macro policies and to account for some strategic interests of donors in specific groups of countries (former colonies, important trade partners, or political allies, for example).
8. The same is true for high levels of conflict and ethnic fractionalization. We dropped these variables from the regression shown in column 1. This helped to reduce problems of multi-collinearity, did not change the significance of the rest of the variables in the model and allowed a more precise estimation of the marginal effects.
9. Results for the Agricultural sector have to be interpreted very cautiously because of weaknesses in the specification. For example, with a p-value of 0.67, the Sargan test for over-identification strongly suggests that the set of instruments are not valid.