Technological convergence, competitiveness, and welfare: A study of international manufacturing industries

Abstract

In this study, a monopolistic competition model is used to investigate the effects of international technological convergence on factor rewards, output composition, and welfare. Four testable hypotheses on the impact of technological convergence on follower's and leader's competitiveness and welfare are presented. We then use 1993–2001 data from 128 manufacturing industries in 35 countries to test these hypotheses. Results show that followers' relative wages and global value-added shares increase with technological convergence. Followers benefit from convergence's positive income effect. Leader's own technological progress is the key to its welfare improvement, while terms-of-trade effects appear less important.

Keywords:

• competitiveness
• manufacturing industries
• productivity growth
• technological convergence
• welfare

Acknowledgments

The authors thank Steven Buccola (Oregon State University) for helpful comments on an earlier version of the article and Mark Gehlhar (ERS/USDA) for providing access to international trade data. Gopinath acknowledges support from the Agricultural Experiment Station of Oregon State University.

Notes

1. If there are increasing returns to scale at the level of national industries, then countries with larger industry outputs will have higher measured TFP even if technology is identical (Harrigan 1999, 268).

2. Relaxing the assumption of constant fixed cost complicates the analysis but does not affect our basic results.

3. Note that complete factor price equalization requires convergence in both fixed cost and marginal cost. Here, marginal cost convergence reduces the technological and wage gap.

4. If both fixed and marginal costs converge, the follower's global production share and relative wage will rise, while its imported share of consumption falls. Terms of trade remain unchanged in both countries. However, both countries benefit from the increased number of goods.

5. The terms ‘technological’ and ‘productivity’ convergence are used synonymously. Recall in our theoretical model that a country's technological level is measured by its labor productivity (x/l or x*/l*). However, technological level is measured empirically here by total factor productivity (based on inputs of both capital and labor) rather than by labor productivity. The latter does not allow one to identify the separate influences of technology and capital growth (Bernard and Jones 1996b).

6. We also have tried the translog function to estimate TFP, and the F-test rejects the hypothesis of constant returns to scale. Therefore, we choose the approach in the text that suppresses substitution between capital and labor, but permits variable returns to scale.

7. Technological convergence in this study is β-convergence, referring to followers' ‘catch-up’ toward the leader.

8. To avoid perfect multicollinearity between Δln(LTFP_i ) and η_0i , we employ industry dummy variables classified at a more aggregate industry level. See the Results and discussion section for more details.

 9. Estimates of equation (4) can be used to decompose both absolute and relative TFP growth of followers.

10. Difference between Δln(LTFP_ci ) and Δln(RTFP_ci ) is the leader's productivity growth rate, which is perfectly collinear with the industry dummy variable, φ_0i. We therefore, as in equation (9), employ industry dummy variables at a more aggregate level.

11. US industrial correspondences between ISIC Revision 2 and Revision 3 are taken from the US Bureau of Census. We assume this correspondence is applicable to every nation. Some countries' data are available for certain years in both revisions, which enable us to test the average difference between the data reported in Revision 3 and those converted, through the US industry correspondences, from Revision 2 to Revision 3. Results of t-tests indicate that none of the data differences in value-added, employment, or gross fixed capital formation is significantly different from zero at the 5% significance level. Hence, we apply to other countries the US correspondences between the two revisions.

12. Manufacturing value-added price index and output price index are computed as the ratio of current to constant manufacturing value added; gross-fixed-capital-formation price index is computed as the ratio of current to constant gross fixed capital formation in the aggregate economy; and the consumer price index (CPI) of the aggregate economy is used to deflate wages.

13. We follow Hall et al.'s (1988) procedure to obtain base-year capital stock data. Given that I_{t 0} is base-year investment, initial capital stock K_{t 0} equals I_{t 0}/(d + g), where g is pre-sample annual growth rate of new capital. Country-specific pre-sample capital growth rates are derived as the average annual growth rates of gross fixed capital formation in the aggregate economy during the 10-year pre-sample period (World Bank 2005). We set the depreciation rate (d) at 8% per year.

14. The import (export) price index is calculated as the ratio of current to constant imports (exports) of goods and services in the aggregate economy.

15. The 26 industries where the US is the second technological leader are ISIC1511, 1514, 1520, 1551, 1553, 1712, 1729, 1730, 2102, 2109, 2423, 2429, 2610, 2692, 2695, 2710, 2813, 2893, 2914, 2923, 3140, 3150, 3190, 3230, 3311, 3313. US production data are unavailable in 10 of 128 industries (ISIC 1820, 2212, 2213, 2219, 2310, 2320, 2330, 3710, 3720, 3999).

16. Carree, Klomp and Thurik (2000) also found inter-industry differences in labor productivity convergence across manufacturing industries in 18 OECD countries during 1972–1992.

17. As in equation (10), we employ three-digit industry dummies in equation (9).