Vertical Specialisation and its Determinants

Abstract

Vertical specialisation (VS) and outsourcing have gained prominence in international trade recently and could represent an entry point for developing countries into world markets as well as a channel for technology transfer. However, entry into international supply chains often requires just-in-time delivery and close to zero fault rates, making market entry via VS difficult. This paper analyses the importance of infrastructure and quality of institutions for VS, for total trade and with a focus on the clothing and electronics sectors. It is found that good governance and an open trade policy is strongly related to VS in electronics. Infrastructure, particularly ports, is strongly related to VS in the clothing sector.

Acknowledgements

Useful comments and suggestions on this and earlier versions of the paper from Linda Orvedal, Henri de Groot, Zdenek Drabek, Paul Swaim, participants at the ‘75 Years of Development Research’ conference, Cornell University, 7–9 May 2004 and participants at the Nordic International Trade Seminars, Copenhagen, 14–15 May 2004 and an anonymous referee. The opinions expressed in this paper is personal and do not engage the OECD or its member countries. The usual disclaimer applies.

Notes

1. The shares are calculated by the author from the WTO database on trade in goods and services and the World Banks's World Development Indicators for world GDP.

2. See for instance Krugman and Venables (1995) for a seminal contribution. This paper constructs a two-country, two-sector, one-factor model. One of the sectors produces differentiated products which are traded subject to iceberg transport costs, while the other produces a homogenous product that is freely traded and subject to constant returns to scale in production.

3. The relation between distance and trade costs has been established in the large and growing literature using the gravity model for estimating the determinants of bilateral trade. See Anderson and Wincoop (2004) for a recent survey. Source of data on distance between the capitals of the countries in question is found on http://cepii.fr/anglaisgraph/bdd/distances.htm

4. Weighted least squares regressions use the glogit command in Stata. The main results are qualitatively the same using both methods.

5. When including the tariff level rather than the level relative to the sample average, the coefficients were larger: −1.70 (significant at a 10% level) for own tariffs in the VS share regression and −2.58 (significant at a 5% level) on own tariffs in the export share regression.

6. From 2004–2005, China's exports of clothing (SITC code 84) to the US increased by 47 per cent, while the corresponding figure for India was 33 per cent. China's exports to the EU increased by 42 per cent during the same period, while India increased its exports by 18 per cent (see United States International Trade Commission, accessed at: http://dataweb.usitc.gov/ and The European Commission, accessed at: http://ec.europa.eu/comm/trade/issues/sectoral/industry/textile/stats.htm).

7. Island and landlocked dummies were explored in the regressions for the other sectors as well, but turned out to be statistically insignificant and did not add any explanatory power to the regressions.