The long-run decline in the share of agricultural and food products in international trade: a gravity equation approach to its causes

Abstract

The objective of this study is to determine the causes of the loss of share of agricultural products and food in international trade. The article compares, using a gravity model, the impact of various factors upon bilateral trade in agricultural products, in manufactures and in total trade, between 1963 and 2000 for a representative sample of 40 countries. The results clearly demonstrate how the low demand elasticity for agricultural products and food, the high degree of protectionism to which they were subjected and their meagre share in intra-industrial trade are the principal causes of their relatively slow growth.

Keywords:

• agri-food trade
• international trade
• gravity model
• second globalization

JEL Classification:

• N70
• N50
• F14
• Q17

Acknowledgements

This study has received financial support from the Ministry of Science and Innovation of the Spanish Government, project ECO2009-07796 and the Department of Science, Technology and Universities of the Government of Aragon, Research Excellence Group for ‘Agri-food Economic History’. We are also grateful for the suggestions of the editor of the journal and the comments of the reviewers, and further wish to thank Ana Angulo, Albert Carreras, Domingo Gallego, Joan Ramon Rosés, Marcela Sabaté and Daniel Tirado for their comments.

Notes

¹ See, for example, the analysis performed by Head and Ries (2001), which examines only 6 years.

² Estimation of the SD of the first difference of the annual natural logarithm of the nominal bilateral exchange rate, for both countries in the 10 years preceding the period t (data for exchange rates taken from WDI CD-ROM, 2004)

³ Africa (Algeria, Egypt, Morocco, Nigeria, Sudan, the Ivory Coast), Asia (China, India, Indonesia, Israel, Japan, Malaysia, Saudi Arabia); North America (Canada, México, United States); Latin America (Argentina, Brazil, Chile, Colombia, Ecuador, Nicaragua, Peru); Europe (Belgium-Luxembourg, Denmark, Finland, France, Germany, Greece, Italy, Ireland, Norway, Portugal, Spain, the Netherlands, the United Kingdom); Oceania (Australia, New Zealand).

⁴ In order to obtain a balanced panel trade flows with a value of 0 are replaced by a figure for minimum trade ($100), following previous research e.g. Raballand (2003) or Schumacher and Siliverstovs (2006) having a similar specification to our approach to the gravity equation. The most common alternative to this method consists of eliminating those trade flows with a value of 0. Frankel (1997) performs a comparative analysis of the two methods and finds negligible differences between the two alternatives. We have preferred the first approach, since it permits us to use more sophisticated econometric methods to correct the recurrent problems of estimation in previous research. Nevertheless, due to the drastic lack of data, exports from China, the Ivory Coast, Nigeria, Sudan, Saudi Arabia and Uruguay to the remaining countries were eliminated. Note that exports from the remaining countries to these countries remain in the sample. Thus, the sample comprises the trade flows among 40 countries of origin × 39 countries of destination × 38 years − (6 × 39 × 38 trade flows were eliminated due to a lack of data) = 50 388 observations.

⁵ Beck and Katz (1995) demonstrate that the SEs of PCSE are more precise than those of Feasible Generalized Least Squares (FGLS), the other alternative to solve the above-mentioned problems.

⁶ Following Feenstra (2004), the use of alternative methods produces similar results.