Empirical strategy and data description
Our main empirical strategy is to derive measures of the product cycle at a high level of disaggregation over a recent period, employing a widely used index of concentration or diffusion (the Herfindahl-Hirschman index) and a second index of the level of relative income associated with revealed comparative advantage in the export of a particular good, to capture the concept of downstreaming. Measurement of diffusion and downstreaming correspond to the two phases of the traditional product cycle. Since the measurement of diffusion is also a measure of concentration, it can also be used as an indicator of Marshallian agglomeration economies that may inhibit downstreaming and lead to persistence in comparative advantage.
Main features of the dataset
Export data for 15 economies for the period 1997-2006 were obtained from the UN COMTRADE system maintained by the United Nations Statistical Division.4 This ten-year period is shorter than those often used to test hypotheses relating to the product cycle, but long enough that disaggregated data can be used without product definitions changing too much.5 Observations were taken on exports to the world, as reported by the exporting economy, of all Harmonized System (HS)-6 level subheadings, hereinafter “products.” The selected products include all those in 21 HS-2 chapters selected from the 96 regular chapters, as listed in Table 8.A1 in the Annex. Broadly speaking, the product landscape consists of chemicals and allied products; machinery, electronics, and instruments; transportation equipment; and armaments. For comparability over time, the products are defined using the HS 1992 nomenclature. Products which for at least one year in the time period had no exports reported by any of the 15 economies are dropped.6 In total, this yields 2 035 products.
The economies included in the dataset are listed in Table 8.2. They include the six largest Organisation for Economic Co-operation and Development (OECD) economies and nine Asian economies. Together, these 15 economies represent approximately 70% of world exports of the products in question, though the percentage varies from product to product. Where available, re-export data is subtracted from gross exports to yield net export data for the included economies and years.7
The HS-2 chapters are selected so as to include all products defined as ATPs by the US Census Bureau (hereafter “Census”), as well as chapters related to these chapters by type of product. Table 8.A3 presents the categories of ATP products, while Table 8.A4 provides a tabulation of the number of ATP products falling in each HS chapter. The ATP products, defined at the HTS-10 level,8 are selected based on the expert judgment of Census staff regarding the technology intensity of products. The list of products used to construct China’s High and New Technology Product Import and Export Statistics Catalogue corresponds closely to the Census ATP list.9 Because the ATP list represents an independent judgment about technology intensity, it is a useful reference point to compare with inferences about technology intensity drawn from the trade data.