Portal de Periódicos da CAPES

Industry-specific human capital, knowledge labour, and industry wage structure in Taiwan

2004; Taylor & Francis; Volume: 36; Issue: 2 Linguagem: Inglês

10.1080/0003684042000174047

1466-4283

Kam Ki Tang, Yi‐Ping Tseng,

Labor Movements and Unions

Abstract This paper investigates the role of industry-specific human capital (ISHC) in determining industry wage structure. The model presented in this paper distinguishes between knowledge labour and physical labour. Knowledge labour is physical labour embodied with ISHC. It is postulated that more ISHC-intensive industries, such as high-tech industries, pay higher wages and the wage premiums increase with workers’ experience. The hypothesis is tested using a merged sample of 1997–1999 manpower utilization survey data from a newly industrialized economy—Taiwan. The findings show support for the effect of ISHC. Acknowledgements We would like to thank Jeff Borland, Noel Gaston, participants of a seminar at the Bond University, participants of the 2001 Conference for Economists in Perth, and the referee for their constructive comments and suggestions. Notes 1 For the earlier vintages, see Slichter (Citation1950), Dickens and Katz (Citation1987), Krueger and Summers (Citation1987, Citation1988), Katz and Summers (Citation1989), Gibbons and Katz (Citation1992), Helwege (Citation1992), Haisken-DeNew and Schmidt (Citation1997), and the references cited therein. In particular, Helwege (Citation1992) warns that there are large overlaps between industry wage structure and occupation wage structure, such that reported inter-industry wage differentials could be due to the insufficient disaggregation of occupations 2 Human capital can be formally classified into general, industry-specific, and firm-specific, respectively. General human capital does not depreciate when workers move across industries or firms; the industry-specific one depreciates completely when workers move across industries but not across firms within an industry; while the firm-specific one depreciates completely when workers move across firms. 3 Another very different approach is the searching-theoretic approach (e.g. Montgomery, Citation1991), which has received increasing attention in recent years. 4 The loss of ISHC is not in full; due to the Cobb–Douglas function in Equation Equation3, a 1% change in the deployment of physical labour will lead to less than 1% change in that of knowledge labour. 5 Since human capital is only industry-specific but not firm-specific, intra-industry movement incurs no opportunity cost to labour. 6 A dot over a variable, say X, denotes a derivative with respect to time, i.e. dX/dt. 7 With a Cobb–Douglas function, allowing β to differ across industries will not change the equilibrium ISHC intensity of an industry. It will add another exogenous element to the determination of inter-industry wage differentials without enhancing the model's intuition. 8 The process can be modified into learning-by-training by specifying a share of physical labour hour spent on human capital investment. As long as the share is constant, it will not change the properties of the model. 9 W R can be interpreted as reservation wage because it is equal to the payoff for staying out of the job market, so a job that pays less than W R will not attract any worker. 10 Equation Equation8 only sets the relative wages between industries. The absolute wage levels will be determined by incorporating the labour demand-side equations derived from the firm's optimization problem. 11 For more details, see Tang and Tseng (Citation2001). 12 For instance, Katz and Summers (Citation1989) suggest that labor has more leverage to extract rent in a capital intensive industry. 13 In this sense, our theoretical result is consistent with the unmeasured labor ability argument in the literature. 14 Since the data are a very short panel with large sample size, random effect model, in general, is preferred to fixed effect model. Moreover, by nature, fixed effect model ignores all the non-matched observations, which account for one-third of our total sample, making the sample size in some industries too small. 15 According to Taiwan's Ministry of Economic Affairs, SMEs are those enterprises with less than 200 employees in the mining and quarrying, manufacturing and construction sectors, and those in other sectors with less than 50 employees. 16 Source: The Small and Medium Enterprise Administration, Ministry of Economic Affairs . 17 MUS is very similar to the US Current Population Survey (CPS) widely used in the literature, e.g. Dickens and Katz (Citation1987), Katz and Summers (Citation1989) and Juhn et al. (Citation1993). CPS contains information on monthly earnings, personal characteristics, industry, occupation, and firm size. However, it was used as cross-sectional data in the previous literature, while we match each two consecutive MUS to form repeated short panels. 18 It actually includes the returns to firm-specific skill as well. However, since the majority of Taiwanese firms are very small, most firm-specific skills are industry-specific rather than firm-specific. 19 The experience-specific industry wage differential is defined as the difference between predicted log wage of an industry and the log wage of base industry for a given experience group. It is calculated from the coefficients of D jit E it and . 20 As mentioned earlier, despite ISHC being accumulated through learning-by-doing in our model, it can be modified to learning-by-training without changing the theoretical results. 21 Choosing other low-tech industries as the base industry does not change this general pattern. 22 In fact, the latest government industry classification has a separated category for computer and communication products. Detailed industry classifications can be found at .