Support for ‘anti-globalist’ policies across the developed world suggests that many people are concerned about the impact of globalisation on employment and wages. This column examines the wage premium for exporting among firms in Japan, using linked employer-employee data to control for other factors that may affect wages. Exports and wages clearly are correlated in Japan's manufacturing sector, especially for smaller-scale plants and firms.
When we look back in the future, the latter half of the 2010s may be seen as the era of anti-globalism. The UK anounced its withdrawal from the EU in 2016 and, in the US, Donald Trump's administration has proposed policies that go against trade liberalisation such as withdrawal from the Trans-Pacific Partnership (TPP) agreement and raising the tariffs of steel and aluminium products and Chinese products.
Policies of anti-globalism in democratic countries such as the US and the UK imply that there are a certain number of people concerned that globalisation of the economy will result in the loss of employment and expansion of wage disparity and who therefore support these policies. However, other factors, such as technological innovation and labour institutions, can also affect employment and wages. Does globalisation still have an impact even after controlling for these other factors? In this regard, I analysed wage inequality in Japan as an example.
How can we distinguish the effect of globalism and other factors on wage inequality?
In Japan, as in other countries, the globalisation of the economy began progressing rapidly in the 1990s, and trade and foreign direct investment have also quickly expanded. Overseas production activities are also being actively carried out through the introduction of offshoring and other activities. Generally, it is thought that firms engaged in globalisation benefit in various ways. Many empirical studies of the relationship between globalisation of firms and productivity/profit have been conducted since the 2000s, and several have indicated that benefits such as improvements in productivity come from international activities such as exporting.
When firms benefit from globalisation, do their workers profit from such benefits? For example, do workers at exporting firms receive higher compensation than those of firms that do not export? Previous studies focused on salary, a representative indicator of benefits for workers, and indicated that average wages at exporting firms are higher than those at non-exporting firms. The difference between them is called the ‘wage premium’ of exporters, and it has been observed in various countries around the world (Schank et al. 2007). However, we should not conclude that workers in exporting firms obtain high wages by exporting due to this wage premium because wages are affected by other firm characteristics (e.g. size, and individual workers' characteristics, academic background, employment pattern, job category, and gender). Only after confirming that the wage premium of exporters exists when these various characteristics are controlled for can we say that workers at exporters receive higher wages.
Therefore, to control for the characteristics of firms and workers, an increasing number of studies have been produced in recent years examining the wage premium of exporters by using linked employer-employee data, combining data from firms and their employees.
Working from recent studies, in Ito (2017) I construct Japan's cross-sectional employer-employee data by merging plants' data from the Census of Manufacture implemented by the Ministry of Economy, Trade and Industry (METI); the Economic Census for Business Activity implemented by the Ministry of Internal Affairs and Communications (MIC) and METI; and worker data from the Basic Survey on Wage Structure implemented by the Ministry of Health, Labour and Welfare (MHLW). The study estimates a Mincer-type wage function in Japan's manufacturing sector to determine the existence of part of the wage premium that cannot be explained by other characteristics of workers and plants, i.e. a part purely correlated with exports.
Estimation result of the Mincer-type wage function
The estimation of the Mincer-type wage function indicates that there is clearly a wage disparity due to the presence or absence of exports concerning wages. This result differs from previous domestic and overseas research including Schank et al. (2007), Munch and Skaksen (2008), Tanaka (2015), and Endo (2016).
However, unlike previous research, in my paper I do not use panel data of workers due to restrictions on data and cannot exclude the possibility that differences in the unobservable workers' abilities and plant characteristics are affected. For this reason, I also estimated the wage function by plant size because the same scale plants are expected to have small differences in characteristics that cannot be observed. The result shows that, at plants with 300 or fewer employees and firms with 299 or less employees, the impact of export on wage disparity is remarkable.
Additionally, in order to compare the relative degree of influence of the part correlating purely with export among the export premieres of wages, we performed Blinder-Oaxaca decomposition on 2012 data. The impact of exports was 9.5% for the entire manufacturing industry, less than 10%, but for plants with 100 or fewer employees, it is over 1/3. Even for plants with 101 to 200 employees, it is 20%, accounting for a relatively large weight (Table 1).
Mechanism behind the big influence of export on wages of small firms
Why did exports bring wage disparity only to smaller plants and firms? Two mechanisms are conceivable.
One is rent sharing, as Helpman et al. (2010) suggested. In general, smaller scale businesses tend to have lower performance. Among such firms, starting exports will greatly improve the performance compared with other firms, and this will be reflected in wages. In contrast, in large-scale firms, non-exporting firms can also afford to raise wages due to other achievements rather than exporting and, as a result, the wage disparity between exporting and non-exporting firms is not clear.
The other is a mechanism that raises wages as a result of training and/or recruiting employees for starting to export. Generally, small and medium-sized firms are not sufficiently structured to develop human resources such as in-house training, compared with large firms. If such firms decide to start exporting, they need to recruit personnel who can conduct export business urgently and pay higher wages for new employees or employees with such ability. Consequently, the wage disparity between exporting and non-exporting firms grows. In contrast, large firms in general are actively engaged in human resource development regardless of whether or not they export, and when non-exporting firms start to export, there is no need to additionally hire and train human resources by raising wages.
Conclusion
In Ito (2017), I reveal that exports and wages clearly are correlated in Japan's manufacturing sector, especially for smaller-scale plants and firms. Thus, should Japan adopt an anti-liberalisation policy such as tariff increases to protect widening wage inequality?
In reality, raising tariffs is not a substantial solution. Milanovic and Squire (2007) indicate that tariff reduction has a limited relationship of tariff reduction and wage inequality between skills and industries. Tariff cuts also have the effect of raising industry-level labour productivity of the industry where tariffs are lowered by reducing employees mainly in less productive firms (Trefler, 2004, LaRochelle-Côté, 2007). The best way for Japan is to seek further trade liberalisation. For wage inequality, the effective policy is to take care of less productive firms by helping them improve productivity or exiting from the market smoothly, although modest efforts are needed.
Editors' note: The main research on which this column is based first appeared as a Discussion Paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.