|Author Name||KATO Takao (Colgate University) / OGAWA Hiromasa (GRIPS) / OWAN Hideo (Faculty Fellow, RIETI)|
|Creation Date/NO.||March 2016 16-E-060|
|Research Project||Economic Analysis of Human Resource Allocation Mechanisms within the Firm: Insider econometrics using HR data|
|Download / Links|
This paper presents a model of promotion which features two different sources of asymmetric information—disutility of working long hours and on the job training (OJT) ability, or the ability to accumulate human capital on the job via learning by doing. The former is the worker's private information while the latter is the employer's. The firm decides whether or not to reveal its private information on the worker's OJT ability to him/her and how much training it provides to him/her. The worker chooses working hours to signal its commitment to the firm. We show that there always is a separating equilibrium in which the worker's working hours fully reveal his/her commitment level. The firm's optimal feedback policy depends on the nature of the training and learning, and the level of overtime pay. Not revealing private information on the worker's ability could be optimal for the firm under certain circumstances. We argue that two recent changes in the Japanese human resource management system—more selective training and an increasing share of occupations exempt from overtime work payment—may be making information revelation optimal for many firms. We further show that revealing information on the worker's ability tends to be optimal for the firm when many in the workforce have high disutility of working long hours. As such, if the firm can use different feedback policies for men and women, it may reveal its private information on the worker's ability only to women but not to men. If this is the case, there is a testable implication: the incidence of promotion should be more highly correlated with the number of hours worked for women than for men. Using personnel records of a large Japanese manufacturing firm, we find evidence in support of this prediction.