What a difference a source makes! An analysis of export data

Abstract

This study examines real export data from two common sources: IMF and UN Commodity Trade Statistics to determine the comparability of these data in level and percentage change form, and if the export-led growth models are robust to the data source. Additionally, the comparability of the data deflated at a disaggregated level data or by a single export unit price index is examined. The results reveal that the data are neither comparable nor in a number of cases, correlated, suggesting that much of the debate regarding export led growth is fuelled by data choice and methodology of deflation.

Notes

¹ Percentage change is included since most longitudinal studies use the export data in the percentage change format to correct for unit roots often found in level data.

² The sample contains 16 Latin American countries: Argentina, Bolivia, Brasil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, México, Panamá, Perú, Paraguay, Uruguay and Venezuela.

³ Oxford Latin American Economic History Database – Methodology, http://oxlad.qeh.ox.ac.uk/sources.php

⁴ The results of the correlations, for aggregate series in both levels and percentage change, are available upon request from the author.

⁵ For in-depth analysis of export led growth literature see Ahmad (2001). For studies that use IFS data see Bahmani-Oskooee et al. (1991); Jung and Marshall (1985). For a study that uses COMTRADE data see Amin Gutiérrez de Piñeres and Ferrantino (2000). More recent papers on the topic of export-led growth are Abu-Quarn and Abu-Bader (2004), Balanguer and Cantavella-Jorda (2004) and Abual-Foul (2004).

⁶ Simple Granger tests are limited to capturing short run dynamics between exports and growth when employing a simple bi-variate model.

⁷ Results for all countries in the sample are available from the author upon request.