Two-group, three-period staggered difference-in-differences design and evaluation of labour market effects of minimum wages in China

We examine inference in a two-group, three-period staggered difference-in-differences setting where conventional pre-trends testing is infeasible. By imposing structured assumptions on treatment effect heterogeneity and dynamics, we develop a novel framework that enables reliable inference on the direction of treatment effects while relaxing the parallel-trends assumption. We apply this framework to assess the effects of minimum wages in China, a context in which the validity of existing DiD-based findings may be compromised by the inability to validate pre-trends. The results indicate that minimum wages substantially increase the wages of low-skilled rural migrant workers, with no detectable effects for high-skilled workers. We find no evidence of employment reductions; conversely, the findings suggest a positive employment effect for low-skilled workers.