Assessing the effects of reducing standard hours: Regression discontinuity evidence from Japan

This paper examines the effect of Japan's reduction of standard hours in the 1990s on labor market outcomes. Given that standard hours become discontinuous at certain establishment sizes, we employ a regression discontinuity design to estimate the impacts of the policy change. The theory predicts that policy change should unambiguously reduce hours worked in establishments with an initial workweek longer than the new standard hours but shorter than the old standard hours. The policy impact is expected to be zero for establishments with short hours, while it is ambiguous for establishments with very long hours. Indeed, the empirical analysis indicates that the reduction of standard hours significantly decreased hours worked in establishments whose hours are predicted to be between the old and new standard hours. Even in establishments with reduced working hours, monthly wages and annual bonuses did not decrease, leading to the increase in labor costs per hour. As a result, the policy change did not contribute to job creation. This paper demonstrates the importance of considering heterogeneous treatment effects, rather than the average treatment effects. When the policy impacts totally differ among groups, as in our case, estimating only the average treatment effect could mask the real impact. In addition, some groups actually affected by the policy may not be the ones targeted by the policy makers; in such cases, if there are heterogeneous treatment effects, unexpected outcomes can occur. (C) 2016 Elsevier Inc. All rights reserved.