Two-way fixed effects regression model

On the Use of Two-Way Fixed Effects Regression Models for Causal Inference with Panel Data argues that the model relies on the assumption of linear additive effects.