A Piece of Evidence of Equal Pay for Equal Work

Active debates on equal pay for equal work are continuing. Although various case examples and statistical data are being used in the process, an important piece of evidence—i.e., one that shows the relationship between productivity and wages—seems to be still lacking. While there are significant differences in pay among workers, whether such disparities exist for good reasons or constitute irrational discrimination cannot be determined simply by comparing pay rates. The key is whether the amount of pay for each worker is reasonable for the level of productivity. This article attempts to present one piece of evidence regarding this point.

Working style reform has been a big policy issue continuing through 2016. The Ministry of Health, Labour and Welfare (MHLW) established a study group, and discussions are underway to work out measures for the realization of equal pay for equal work (Note 1). In the fall of 2016, the Council for the Realization of Work Style Reform was established at the Prime Minister's Office. The improvement of treatment for non-regular workers—i.e., those other than permanent fulltime employees—including the realization of equal pay for equal work is one of the key themes of the council. Those discussions are expected to continue in 2017.

The difficulty of this issue is that we cannot tell whether the observed pay disparities are based on sound economic reasoning or attributable to irrational reasons such as discrimination simply by comparing the rates of pay offered for similar jobs. In the light of economic theory, this is essentially the question of whether the amount of pay for a worker matches the level of his or her productivity (Note 2). For instance, if wages for non-regular workers are low relative to their productivity, raising the wages is desirable from the perspectives of both efficiency and equity. On the other hand, if their wages match their productivity, raising the wages is detrimental to efficiency and may cause unemployment. In this case, in order to raise the wages for non-regular workers, it is necessary to implement measures to enhance their productivity.

As wages are observable, precise measurement of individual workers' productivity would bring us close to solving the problem. However, it is generally quite difficult to measure the productivity of individual workers, except for those in certain types of professions in which the output is measurable at the individual level, such as professional athletes and commissioned workers. For instance, the productivity of school teachers can be measured to some extent by the degree of improvement in students' academic performance and that of university researchers by the number of research papers published or cited. However, it is virtually impossible to measure the productivity of white-collar workers of large companies or construction workers where cooperation within the team is essential. Such challenges of measuring the productivity of individual workers is a big factor in making it difficult to solve this problem.

However, even though it is impossible to measure the productivity of individual workers, it is possible to measure the productivity of individual companies. Some empirical studies led by Hellerstein and Neumark (1995) and Hellerstein et al. (1999) focused on this point. For instance, if we have data on the ratios of female employees at companies, we can infer whether wages for women are, on average, reasonable for their level of productivity by comparing the relationship between the ratio of female employees and their level of productivity or the relationship between the ratio of female employees and their level of wages. By taking such an approach, some attempts have been made to estimate the productivity-wage gap (defined as a difference in the degree of contribution to productivity and that to the average wages) by age, gender, race, educational attainment, and so forth. However, very few studies have focused on non-regular workers. One of the few exceptions is Garnero et al. (2014). Using Belgian employer-employee matched panel data, they showed that part-time workers are paid less than what they should deserve for their level of productivity. Unfortunately, it seems that such empirical findings are extremely limited in Japan, and we do not have sufficient evidence that provides the basis for policy discussion (Note 3).

Relating to this point, I would like to present some of the preliminary findings from my recent analysis. Specifically, I measured the productivity-wage gap for selected types of workers such as part-timers and women using a dataset developed by linking data from the Basic Survey of Japanese Business Structure and Activities and those from an original survey of companies (Note 4). I do not go into details, but my estimates based on data collected from some 2,400 companies show that wages for both part-time workers and women are generally reasonable for the level of their respective contribution to productivity (see Figure 1) (Note 5).

In other words, while part-time workers' contribution to total factor productivity (TFP) is lower than that of fulltime workers, the productivity-wage gap is negligible as wages for the former are also lower than those for the latter. A comparison of female and male workers showed relatively small differences whether in the level of contribution to TFP or in wage levels, and no productivity-wage gap was observed.

Note: Based on data from some 2,400 Japanese companies, the relationship between TFP and the average wages was estimated, using the ratio of part-timers to total employees and the ratio of female to total employees as explanatory variables. The labor share has been adjusted for the calculation of TFP differentials. The black dot in the center of each bar is a point estimate, while the orange and blue bars represent 95% confidence intervals. When the point estimate of the average wage is lower than that of TFP in the graph, the productivity of the category is high relative to the wage level. In the opposite case, the productivity of the category is low relative to the wage level.

This finding—i.e., productivity and wage levels are generally balanced—indicates that Japanese companies, when seen on the whole, set reasonable wages, and we cannot say that they discriminate against certain types of workers in wages. Of course, this is just an average picture. In reality, there must be workers who are underpaid and overpaid for their respective levels of contribution to the productivity of their companies. Therefore, if equal pay for equal work is realized in the strict sense that everyone is paid reasonably commensurate with the level of contribution to productivity, some part-timers would see their wages raised and just as many would find their wages reduced.

From the macroeconomic point of view, so far as productivity and wage levels are fairly balanced in average terms, the realization of equal pay for equal work does not have much to do with the compensation of employees or the labor share in an economy as a whole. On the other hand, this clearly points to the crucial importance of human capital investment in a way to raise the productivity of low-productivity workers, if Japan is to reduce overall wage disparities. Such an initiative will also serve as a growth strategy.

In this regard, it is quite reasonable that the ongoing policy debate on equal pay for equal work focuses not only on wage levels but also on equal opportunities for education and training. Raising both the productivity and wage levels by means of appropriate human capital investment, such as the development of career paths for employees including part-timers and female workers, should be the pillar of the government's economic policy. On the other hand, it is hoped that a guideline for equal pay for equal work to be finalized in due time and its application will not end up imposing yet another layer of excessive compliance rules to impede the efforts to address the problem of long-hour working and improve productivity—the other pillars of the working style reform (Note 6).

What has been discussed is no more than the result of cross-sectional analysis based on a limited sample, and thus has its limitations. Furthermore, data on the characteristics of employees are not necessarily sufficient. In order to derive a definitive conclusion and policy implications, it remains a challenge to conduct a detailed analysis of a large data set containing more abundant information, and I would like to count on researchers in this field to contribute to this endeavor. Also, it should be noted that while there are various types of non-regular workers, my analysis focused only on part-time workers.