Green productivity and bilateral trade flows in an augmented gravity model – A panel data analysis

Abstract

Motivated by the debate in the trade liberalization and the environment literature, this article examines the effect of enhancing green productivity (GP) on bilateral trade flows. The uptake of per capita ISO14001 certification counts is used to measure GP. The existing literature provides other key determinants of bilateral trade flows. This article employs an augmented gravity model and presents panel data analysis on 26 countries from 1995–2004. Since GP is closely related to quality management, this article also examines the joint effect of the measure of quality management systems (QMS) and the measure of GP. Several fixed effects regression equations are estimated. The results support the hypothesis that enhancing green productivity is a positive and statistically significant determinant of real bilateral exports. The joint significance of the measures of GP and QMS is also supported. This article lends empirical support for the new trade theory and Linder's hypothesis and is consistent with those obtained in the existing literature.

Keywords:

• green productivity
• bilateral trade flows
• gravity model
• interaction effects
• panel econometrics

JEL Classifications:

• F18
• Q56
• C33
• F14
• F15

Acknowledgments

Comments from Prof. H. Benchekroun, Prof. R. Cairns, Prof. J. Galbraith, Prof. J. Kurien, Prof. N.V. Long of McGill University, and Prof. E. James of Dawson College are gratefully acknowledged.

Notes

 1. As in Baltagi, Egger, and Pfaffermayr (2003), this article examines flows of real bilateral exports.

 2. The results are robust to an alternate specification of the model using levels. The use of panel data in a gravity equation model of bilateral trade flows with distances between countries is intended to address potential simultaneity bias. Many studies used the instrumental variable method to tackle endogeneity. However, as highlighted by Baier and Bergstrand (2007), in many cases the instruments used are less than fully convincing while the use of panel data is effective in addressing endogeneity issues.

 3. According to the pollution havens hypothesis stricter environmental regulations in one country either encourages domestic production facilities to move to countries with less stringent environmental regulations or encourages increasing imports from those countries.

 4. See Jha, Markandya and Vossenaar (1999) for case studies on 11 developing countries.

 5. There is a considerable amount of research in this area including a specialized survey by Jaffe et al. (1995).

 6. Of these, productivity gains appear to be the most important with proponents claiming that firms which implement ISO14001, can save significant costs by way of enhanced operational efficiency (Rondinelli and Berry 2000). Further details can be found in Haufler (2001).

 7. Equation (2) presents one of the four alternative specifications used. The first alternative is associated with the GDP of the two countries i & j as the economic proxy for the mass of the two bodies, while in the other two alternatives mass is associated with GDP per capita and with both GDP and GDP per capita.

 8. LGDT_ijt  = log(GDP_it  + GDP_jt ).

 9. LSIM_ijt  = log[1−(GDP_it /GDP_it  + GDP_jt )² – (GDP_jt /GDP_it  + GDP_jt )²].

10. LRFAC_ijt  = |log(GDP_it /population_it ) – log(GDP_jt /population_jt )|.

11. Baltagi, Egger, and Pfaffermayr (2003) have used bilateral transportation costs as the difference between imports measured at c.i.f and exports reported at f.o.b. This data was unavailable for many of the developing countries included in this article. As in (Bikker 1987; Harris, Kónya, and Mátyás 2002; Jug and Mirza 2005; Porojan 2001) and others, bilateral geographical distances between two countries is used as a proxy for bilateral transportation costs.

12. LSIM14pc_ijt  = log[1−(ISO14pc_it /ISO14pc_it  + ISO14pc_jt )²−(ISO14pc_jt /ISO14pc_it  + ISO14pc_jt )²].

13. LSIM9pc_ijt  = log[1−(ISO9pc_it /ISO9pc_it  + ISO9pc_jt )²−(ISO9pc_jt /ISO9pc_it  + ISO9pc_jt )²].

14. The bilateral flows of exports are obtained from the Direction of Trade Statistics (2006) while the data for GDP, population size, export deflators and GDP deflators are obtained from the Yearbook of International Financial Statistics (2006). All values are in US dollars and this article uses the STATA 10 S/E for the statistical analysis.

15. Distances between major cities of countries is expressed in KM, downloadable from www.geobytes.com

16. A joint hypothesis test of LDIS with (αβ)_ij in Model II gives a F value of 409.04 indicating that these variables are jointly significant in explaining changes in LRX.

17. It may be noted that when LSIM9pc is used without LSIM14pc it appears with a positive but statistically insignificant coefficient. Thus quality management alone does not appear to have a significant impact on bilateral real exports.