Why Women’s Share among Managers Is So Low in Japan: A Statistical Fallacy or A Shadow of the Employment System?

Women’s share among managers is extremely low in Japan, compared with other OECD countries. We investigate two possible factors accounting for it: (a) the inconsistent application across countries of the classification of occupations, and (b) the international differences in employment systems. By analyzing the OECD Survey of Adult Skills (PIAAC) data and others, we show the following three empirical results. (1) The proportion of managers based on the International Standard Classification of Occupation is much smaller in Japan than in other OECD countries. The proportion of workers classified as managers among total employment is over 10% in the United States and the United Kingdom, while it is only about 2% in Japan. If the work of managers is distributed in each country equally, there would not be such large differences across countries in the fractions of the employees who are managers. Such large gaps could stem from differences in the organizational structures of companies across countries or from differences in employment practices across countries. But the low women’s share among managers in Japan might simply be because of differences in the scope of the category of workers classified as managers. In fact, the differences in women’s share among managers across countries becomes much smaller, if we take another definition of managers, based on the number of subordinates. (2) The effect of educational attainment on promotion is much larger in some OECD countries than in Japan. In France, for example, grandes-écoles are closely related to the managerial category of the socio-economic classification called cadres. Many graduates of grandes-écoles are promoted to the category of cadres at the start of their professional careers. By contrast, in Japan it is quite rare to see a worker being promoted to managerial category at the start of her career, even if she is a graduate from a prestigious university. (3) We estimate a promotion function (probit model), in which the dependent variable is a dummy variable taking the value one if the worker is promoted to manager, zero if otherwise, with labor market experience, schooling years and female dummy being included among the covariates. The estimation results are consistent with the following two hypotheses, though the consistency is weak for (ii). (i) Women’s share among managers is higher in those countries where the speed of promotion to manager is faster. (ii) Women’s share among managers is higher in th