Trade specialization, export productivity and growth in Brazil, China, India, South Africa, and a cross section of countries

See also Lucas (1988, 1993) and Grossman and Helpman (1991) for theoretical formulation and modelling.

The data is based on the Harmonised System Classification (HS 1992) at 6 digits level (COMTRADE dataset 2006).

The detailed estimates are presented in Tables 9 and 10 in the appendix. Each taxonomy encompasses the following sectors and commodities: (1) Primary: foodstuffs, cotton/wool, minerals; (2) resource-based: prepared meats/fruits, beverages, wood products, vegetable oils, ore concentrates, petroleum/rubber products, cement, cut gems, glass; (3) low-tech: textile, garments and footwear, leather manufactures, pottery, simple metal parts/structures, furniture, jewellery, toys, plastic products; (4) medium-tech: synthetic fibres, chemicals and paints, fertilisers, plastics, iron, pipes/tubes, automotive (vehicles and parts), engineering; and, (5) high-tech: electronic/electrical (office/data processing/telecommunications equipment, and other (pharmaceuticals, aerospace, optical/measuring instruments, cameras).

A typical example is the concessions granted under the import substitution industrialization (ISI) strategy. Incentives were also nurtured under regional trade agreements. China and India in Asia, and Brazil and Mexico in Latin America are examples of industrial policy to develop competitiveness. However, in some instances, ISI deviated from its goals because of deficiently coordinated industrial policy aimed at upgrading skills, technology, institutions and infrastructure.

The increased efforts of domestic companies to improve their technological capabilities have further pressured foreign companies to enhance their R&D activities in China in order to being able to maintain their competitive (Chen 2007; Walsh 2007). Foreign and domestic companies’ operations in China have evolved from simple technical service and product localization units to fully technology development units, where both research and development are carried out (Santos-Paulino et al. 2008).

The within-groups estimator is thus likely to perform well only when the time dimension of the panel is large, as in our case. See also Arellano (1993) and Arellano and Bond (1991, 2001).

The results are robust to the inclusion of additional regressors such as investment, labour force ratios, and other country groups such as DRAGONS (Hong Kong, Singapore and South Korea), TIGERS (Malaysia, Phillipines, and Thailand), OECD, Latin America, Africa, and Asia. However, due to missing data for a significant number of observations in the sample, these were omitted from the chosen specifications. Also, the results were generally robust to including the SEG dummy variable on its own among the regressors, however, this variable was generally poorly determined and hence is not reported.

Further estimations including the trade dissimilarity index (A _i ) were also performed, but the results were not statistically sound and therefore are not reported.