Do large capital inflows hinder competitiveness? The Dutch disease in Ethiopia

Abstract

This article investigates whether large inflows of foreign aid and remittances have had a damaging impact on the Ethiopian Real Exchange Rate (RER). We improve the current empirical literature by: (i) compiling a unique quarterly dataset to provide a larger sample size and enable the modelling of important intra-year dynamics – which should lead to better model specifications; (ii) providing a new empirical approach (Unobserved Components (UC)) to test the ‘Dutch disease’ hypothesis; and (iii) using several cointegration approaches to further test the robustness of our conclusions. Our results suggest that there are two main long-run determinants of the RER in Ethiopia: trade openness is found to be correlated with RER depreciations, while a positive shock to the terms of trade tends to appreciate the RER. Foreign aid is not found to have a statistically significant impact, while there is only weak evidence that remittances are associated with RER appreciations. The lack of empirical support for the ‘Dutch disease’ hypothesis suggests that Ethiopia has been able to effectively manage large capital inflows, thus avoiding major episodes of macroeconomic instability. We believe that most African countries will therefore be able to absorb large inflows of foreign capital without damaging their external competitiveness.

Keywords:

• real exchange rate
• foreign aid
• time series models
• Africa

JEL Classification::

• C22
• F35
• O24
• O55

Acknowledgements

The author would like to thank the financial support provided by the Fundação para a Ciência e a Tecnologia.

Notes

¹ The authors do not find similar effects from remittance flows.

² Since remittance inflows predominantly finance the (private) purchase of (nontradable) local goods, we may expect remittances to have a stronger impact on the RER than foreign aid inflows.

³ This has been suggested by several Computable General Equilibrium (CGE) studies.

⁴ According to the OECD-DAC 2011 Development Co-operation Report, Ethiopia has received $3.8 billion of Official Development Assistance in 2009, second only to Afghanistan.

⁵ Moreover, we avoid problems of size and power usually found in unit root and cointegration tests.

⁶ While (domestic) nontradable prices were proxied by Ethiopia's CPI (foreign), tradable prices were proxied by the (trade partner's) PPI or WPI.

⁷ They accounted for 73% of total trade flows during the period 1981–2008.

⁸ Our dataset was compiled from the IMF's Balance of Payments Statistics (BoPS), which reports data from the Central Bank of Ethiopia, and the International Financial Statistics (IFS).

⁹ Alternative measures were also calculated (e.g. combining prices for coffee and agricultural raw materials), but they performed poorly.

¹⁰ A similar approach is followed by Bourdet and Falck (2006) to proxy for workers’ remittances.

¹¹ The ECM-test is a t-test on the lagged dependent variable, which in practice assesses the statistical significance of the adjustment coefficient of the error-correction term. The bounds test approach uses the conditional unrestricted ECM and performs an F-test (or Wald-test) on the long-run coefficients. The distribution of both tests is nonstandard, and therefore we use the tabulated values from Pesaran et al. (2001).