Does unmeasured ability explain the wage premium associated with technological change?: Quantile regression analysis

Abstract

By using the quantile regressions of earnings equation, we find that the educational wage premium is higher in industries with rapid technological change than in industries with slower technological change at every decile in the distribution of wage residuals. The wage premium associated with the technological change is mostly explained by the returns to workers' unobserved heterogeneities, which are correlated with education, rather than the rents of high-tech industries.

Acknowledgements

This research was supported by the Fund for Promoting Basic Research at the College of Business and Economics in Yonsei University.

Notes

¹ Johnson (1997), Katz and Autor (1999) and Acemoglu (2002) present excellent reviews of the literature on the relation between earnings inequality and technological change.

² Taylor (2002) also considers how technological change has influenced the return to occupation as well as education.

³ Boyle and McCormack (2002) find that broad agreement existed across countries in the factors which influence labour demand despite considerable differences in the cross-country nature of labour market institutions.

⁴ The distinction is also important for policy makers who intend to train workforces. If returns to college education increase, then policies aiming at the expansion of college graduates are needed. In contrast, if returns to ability increases, then there should be policies to improve the ability of an individual. These may include policies improving the quality of college education or improving other types of skills not learned in college.

⁵ They do not, however, show the empirical evidences. They just develop a growth model and explain the observed pattern in the US and other advanced countries over the past several decades.

⁶ The earnings premium associated with computer usage could also reflect unobserved ability, if more able workers are assigned to jobs using computers (DiNardo and Pischke, 1997).

⁷ In Korea, the entrance quota of college has been strictly controlled by the government. Before 1980, the government allowed only a gradual increase in college enrollment with the increase in population. The educational reform in 1980 restructured the quota system and increased the number of college graduates dramatically.

⁸ Choi and Jeong (2005) also show that the skill premium associated with technological change is mostly explained by the returns to workers’ unobserved heterogeneities. They use the two-stage double fixed effects, developed by Bartel and Sicherman (1999). However, their results suffer from the insufficient number of industry switchers as well as from the problems of the two-stage, double fixed effects model mentioned earlier in this article.

⁹ The KLIPS is a household panel survey conducted by the Korea Labor Institute in order to measure changes in the well-being of individuals and families over time and the determinants of well-being. The target population comprises all persons living in urban areas (except Jeju Island and the residents of institutions). The first reference year of the survey was 1998 and individuals selected for the survey are interviewed once a year. For the first panel, around 5000 households were included as samples, composed of about 13 300 adults aged 15 years and over.

¹⁰ Korea Productivity Center (2001).

¹¹ The change of TFP for the years 1990 to 1997 was added, along with the change of TFP for the years 1980 to 2000, to eliminate any possible distortion caused by financial crisis broken out at the end of 1997.

¹² Korean Ministry of Science and Technology (various years).

¹³ Hur et al . (2002).

¹⁴ To preserve space, only the results using R&D as a proxy for technological change are reported in the appendix. Other results are available upon request.

¹⁵ To check the stability of bootstrap estimates of SEs, some parameters have been bootstrapped 500 times. The results were not qualitatively different from the ones with 200 replications.