Foreign ownership, human capital, and the structure of wages in Japan

Abstract

We examine the differences in the structure of wages between domestic and foreign-owned establishments in Japan. We use high-quality datasets from the Japanese government and construct a large employer–employee matched database consisting of 1 million workers in 1998. Our results confirm that foreign-owned establishments in Japan pay higher wages. We estimate that one percentage increase in foreign-ownership share of equity raises wages by 0.3%. We surmise that this foreign-ownership wage premium can be explained, at least in part, by compensating wage differentials. Workers in foreign-owned establishments are not protected by lifetime employment. They receive higher compensation for being exposed to higher risk and forfeiting their employment security. We also find that in foreign-owned establishments, wages are determined more by general skills, and less by firm-specific skills. These effects become more pronounced among establishments with a higher share of foreign ownership. The gender wage gap is considerably smaller among foreign establishments. Given the lack of long-term prospects for women in the Japanese labor market, foreign-owned establishments may be one source of ‘brain drain’ for highly skilled women there.

Keywords:

• foreign ownership
• gender wage gap
• structure of wages

Notes

 1. In 2001, mid-career hires as a proportion of all hires was less than 10% among one-third of all firms, and less than 30% among half of all firms in Japan (Ministry of Health, Labour and Welfare (MHLW) 2003).

 2. See also Clark and Ogawa (1992) and Genda (1998) who show empirically that effects on wages declined throughout much of the 1980s.

 3. Fifty-two percent of foreign firms had adopted the annual salary system (METI 2002), in contrast to 12.3% among Japanese firms (Ministry of Labour 1999). The MHLW's definition of an annual salary system (nenposei) is ‘a system under which wages are decided on an annual basis and is determined primarily by ability and performance.’ However, in practice, the annual salary system involves the renegotiation of the employment contract and not just wages. Hence, the system is taken to be synonymous with short-term contract and characterizes an employment relationship which contrasts greatly from the implicit long-term contract representative of Japanese firms.

 4. See Recruit Works Institute, Saiyo Brand Chosa (Recruitment Brands Ranking), in Japanese, various years.

 5. The empirical evidence regarding the decline of lifetime employment is mixed. See for example Ono (2010) who surmises that the share of workers in the core (who are presumably covered by lifetime employment) is declining, but the probability of job separations has remained stable for those who are already in the core.

 6. We exclude part-time workers for the following reasons. First, the inclusion of part-time workers introduces too much heterogeneity. Part-time workers in the Wage Census are defined as those who work less than 35 h a week. However, the Wage Census does not provide information on the hours worked. So this category of workers can include anyone that worked between 0 and 35 h. This inclusion thus inflates the variance in wages without being able to effectively control for this variance. And second, we are keenly interested in estimating the seniority (or the tenure) effect on wages. Since part-time workers enter and exit the labor force for reasons that are presumably different from those of full-time workers, the tenure effect of part-time workers does not capture the same effect. For these reasons, it is unconventional to include part-time workers in the dataset at least in the case of Japan. Empirical studies of the Japanese labor market follow this convention [see for example Hashimoto and Raisian (1985), Mincer and Higuchi (1987), Genda (1998) and Abe (2000)]. As much as possible, we would like to make our work comparable to earlier established studies.

 7. Abe (2000), for example, presents econometric issues associated with estimating wage equations from cell mean data as opposed to microdata, and illustrates the bias arising from cell averaging.

 8. In earlier work, Ono (2007) uses a smaller data set where he controls for selection into the foreign firms in looking at patterns of career mobility and earnings in the Japanese labor market. He finds some evidence that selection effects matter, but this did not significantly alter the interpretation of his findings.

 9. Odaki (2000), using the Establishment Census of 1996, estimates this percentage to be 5.9%. This difference is most likely attributable to differences in survey weights. He also estimates the distribution of firms, establishments, and employees as a function of foreign-ownership share of capital.

10. Wage premiums using various categories of foreign-ownership percentage were also estimated but produced similar results and are not reported here.

11. This result can be interpreted to mean that the effect of work experience and tenure are roughly the same in domestic establishments. In fact, the coefficients for work experience and tenure for domestic establishments reported in Table 2 are not statistically different from each other.

12. For review of recent literature on performance-based pay in Japan, see Ohtake (2005, 2007).