Conditioned Export-Led Gowth Hypothesis : A Panel Threshold Regressions Approach

This paper proposes a reassessment of the export-led growth hypothesis focusing on conditioning effects from countries initial level of GDP per worker, human capital stock, and exports share in GDP. For this purpose a panel threshold regression technique was applied over selected cross-country panel data, covering a broad sample of 72 countries and two sub-samples over the period from 1974 to 2003. Special attention was given to the 5-years data averaging procedure, using panel unit root tests, and to the variables measures choice, where a sensitivity analysis is proposed. Overall, the evidence reported favors the export-led growth hypothesis, where the relationship between exports and growth was showed to be not as trivial as linear specifications would indicate