Country and industry-level determinants of vertical specialization-based trade

Abstract

This paper investigates country and industry-level determinants of vertical specialization-based trade. Industries that engage in this pattern of trade are identified through their use of offshore assembly provisions in the US tariff code. Findings explain why industries engage in vertical specialization-based trade and shed light on factors that enter production location decisions. Identifying factors that encourage vertical specialization-based production and trade will enhance our understanding of industry strategy and how trade patterns will evolve as the process of globalization continues. Results suggest vertical specialization-based trade will continue to grow relative to total trade.

Keywords:

• Vertical specialization
• production sharing
• offshore assembly provisions

Acknowledgments

The author is grateful to Donald Bruce and two anonymous reviewers for providing many helpful comments.

Notes

¹Hummels et al. (1998) describes outsourcing as the relocation of component production to another country and subsequent importation of components from that country. Components are then used to make a good that is not exported. If the final good is exported, both outsourcing and vertical specialization are said to have occurred. Vertical specialization-based trade refers to the value of imported inputs embodied in goods that are exported.

²See Clark et al. (1989), and Clark et al. (2000).

³Imports under both items 9802.00.60 and 9802.00.80 are included in the present study. The latter accounts for 99% of all OAP imports. More than half of the 4-digit Standard Industrial Classification (SIC) industries use the OAPs. For a list of countries and industries engaged in OAP assembly and vertical specialization, see US International Trade Commission (1998, 1999).

⁴A US International Trade Commission (1988) publication identified four types of offshore operations that use OAPs: (a) foreign-based manufacturers that use US-based components; (b) US producers that moved assembly to low wage countries; (c) US-based producers that produce abroad to expand export markets, but send some goods back to the US; and (d) multinational firms that process metals.

⁵Variable definitions and summary statistics are presented in the Appendix.

⁶Feenstra et al. (2000) found goods exported from the US for further processing were more skilled labor intensive than other goods US industries produce.

⁷It is common practice to interpret differences in per capita income across countries as differences in capital-labor endowment ratios. Unskilled labor abundant countries will have relatively low levels of per capita income. See, for example, Helpman & Krugman (1985), Helpman (1987), and Clark & Stanley (1999). A preferred approach would be to use the actual capital–labor endowment ratio. These figures are unavailable for most countries in the present study and are likely to be unreliable when available.

⁸See Dollar (1992), and Clark (1997).

⁹See Dunning (1979).

¹⁰For example, Helleiner (1973) and Grunwald & Flamm (1985) discuss labor costs, transportation costs, tariffs, scale economies, and the ability to segment production processes as determinants of offshore activity.

¹¹See Helleiner (1973).

¹²See Vernon (1974).

¹³See Markusen (1995) and Helleiner & Lavergne (1979).

¹⁴See Grunwald & Flamm (1985: 240).

¹⁵Foreign tariffs and non-tariff measures also influence the location of production. These data are unavailable on a disaggregated basis for the countries included in the present study.

¹⁶See Grunwald & Flamm (1985: 34–37) for a discussion of this issue and other important exclusions from OAP data.

¹⁷A likelihood ratio test is used to compare random effects models with a model that does not include random effects. We reject the null hypothesis that random effects do not contribute to the model at all standard levels of significance. Fixed effect probit and tobit models are not readily available.

¹⁸Factors that influence the availability to search for a partner and establish relationship-specific investments in an environment of incomplete contracts influence the success of outsourcing. See Grossman & Helpman (2005).

¹⁹The black market premium exchange rate measures the deviation of the black market exchange rate from the official rate. A negative coefficient is consistent with the hypothesis that greater exchange rate distortion reduces the likelihood of vertical specialization.

²⁰Perhaps our tax measure reflects the ability of a foreign country to support the infrastructure necessary to attract and sustain offshore assembly.

²¹Swenson's (2004) study of OAP determinants found the OAP import share was positively related to a country's production costs and negatively related to competitor costs. Negative relationships were established between the OAP share and GDP, per capita GDP, and distance.

²²The Barro & Lee (1994) index of political rights varies from 1 to 7, with ‘1’ indicating ‘most free.’ A negative coefficient is expected for this variable because we hypothesize that greater political rights will raise the likelihood of vertical specialization-based trade.