Productivity growth in Korea: efficiency improvement or technical progress?

Abstract

This paper shows that the productivity gains in Korean manufacturing are mostly from efficiency improvement rather than from technical progress. These findings are contrary to those of previous sectoral studies of Korean and Taiwanese manufacturing, but are consistent with those of cross-country studies. Regression results show that both domestic and foreign R&D played an important role in increasing efficiency and technical progress in Korean manufacturing. However, domestic R&D has more effect on technical progress, while foreign R&D has played a relatively stronger role in efficiency improvement.

Acknowledgement

We are grateful to professors Frank S. T. Hsiao and Keith Maskus for helpful comments and suggestions. We are also indebted to Dr Sunmi Chang for help in collecting data.

Notes

¹ See Färe et al. (1994) for a graphical explanation.

² Ray and Desli (1997) emphasized the importance of variable-returns-to-scale (VRS) in using a reference technology. In some cases, however, the VRS method has an infeasible solution (Ray and Desli, 1997, p. 1037). In response to Ray and Desli (1997), Färe et al. (1997) commented that constant-returns-to-scale captures long-run results, while the VRS is appropriate for the short run. Since our study analyses the long-run productivity trend for 1970–1996, we use the method of Färe et al. (1994).

³ See Appendix A for details of the manufacturing industry classification.

⁴ The 14 countries are Australia, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, the UK, and the USA.

⁵ Refer to www.oecd.org for how to match the SITC and ISIC. Also, see Appendix A for the industry classification. Korean R&D data are available by industry from 1976.

⁶ We considered the relative importance of each industry within each specific group when we calculated the summary index of one specific group for each year. The geometric mean is appropriate for calculating summary indexes over the period, since the Malmquist index is multiplicative.

⁷ Note that values of the Malmquist index or any of its components that are less than 1 denote deterioration in performance, whereas values greater than 1 denote improvements in the relevant performance.

⁸ According to previous studies, the annual growth rate of the TFP of Korean manufacturing was 2.0% over the period 1967–1989 in Park and Kwon (1995, p. 349), 3.0% over 1966–1990 in Young (1995, p. 660), 1.26% over 1975–1990 in Nadiri and Kim (1996, p. 31), and 2.0% over 1966–1988 in Kim (2000). Each study showed a different TFP growth rate. This might result from different estimation methods, different data sources, or different periods. Park and Kwon (1995) and Kim (2000) estimated sectoral productivity growth, while Young (1995) and Nadiri and Kim (1996) base their results on the aggregate production.

⁹ Hong and Kim's (1996) data set is for 36 Korean manufacturing industries for 1967–1993, while Madani's (1996) data set is for 52 industries for 1972–1991.

¹⁰ Lee et al. (1998) used Hong and Kim's (1996) data set.

¹¹ See Keller (2002) for a detailed explanation. Keller (2002) used an input–output matrix and a technology flow matrix to construct the domestic other-industry R&D capital stock, but we use only an input–output matrix because technology flow matrix is not available for Korean manufacturing.

¹² The elasticity of TFP with respect to the same sector R&D in Keller (2002) is 0.074 for 13 industries in eight OECD countries over the period 1970–1991.

¹³ The difference between the coefficients of lnR&D_S and lnR&D_F is larger in the CH method than in the LP method. This shows that the CH method might overestimate the effect of foreign R&D stocks.