The validity of the ELG hypothesis in the MENA region: cointegration and error correction model analysis

Abstract

The export-led growth (ELG) hypothesis is examined for nine Middle East and North Africa (MENA) countries in three-variable vector autoregressive and error correction models. When considering total exports, the results reject the ELG hypothesis in almost all of these countries. When only manufactured exports are examined, no support is found for ELG in countries with relatively low shares of manufactured exports in total merchandise exports but strong support in countries with relatively high shares. These findings suggest that promoting exports may contribute to economic growth only after a certain threshold of manufactured exports has been reached.

Acknowledgements

We would like to thank the participants in the Eastern Economic Association conference 2001, participants of seminars at the Hebrew University, Northeastern University, and Ben-Gurion University for their helpful comments. The usual disclaimer applies.

Notes

The ELG hypothesis is considered one of the main pillars of the free trade school of thought that emerged in the 1980s. The other major school of thought, which is known as the protectionism school and is based on Prebisch (1950), calls for the adoption of policies of import substitution rather than promoting exports to stimulate economic growth.

See Giles and Williams (2000) for a comprehensive survey of the empirical literature.

This region encompasses the 21 members of the Arab League, plus Iran, Israel, and Turkey.

Chow (1987), Bahmani-Oskooee et al. (1991), Biswal and Dhawan (1998), and others use this definition of the ELG hypothesis.

Abu-Bader and Abu-Qarn (2003) is among the very few studies that addressed the MENA region applying proper time series techniques to investigate the relationship between economic growth and two categories of government expenditures, namely civilian and military expenditures.

They concluded that the effect is negative in each direction.

In these cases, the results were hindered by the presence of serial correlation.

For countries with no cointegration detected, Xu performed SGC on first differences.

Potential variables include the exchange rate, terms of trade, investment, and government spending. An example is found in Glasure and Lee (1999).

See Serletis (1992) and Riezman et al. (1996).

No cointegration was detected between the variables LGDP and LM that are I(1).

Radelet (1999).