Hello everyone,

I am using the -did_multiplegt- command developed by Clément de Chaisemartin Xavier D'Haultfoeuille (C&D) to estimate two-way fixed effects in a setting where treatment has multiple periods. They show that linear regressions with group and period fixed effects identify weighted sums of the average treatment effects (ATE) in each group and period, with weights that may be negative. Due to the negative weights, the linear regression estimand may for instance be negative while all the ATEs are positive. I have estimated this conventional linear regression (two-way fixed effects (TWFE)) using the -reghdfe- command, and I get different estimates, as well as different significance levels.

I would just like to confirm, that intuitively this divergence between estimates makes sense as the conventional TWFE estimate is biased downwards and this is the whole reason why we want an adjusted estimator as C&D have developed? Does it make any sense to have a discussion where both sets of estimates are compared to try and get a sense of the magnitude of the bias? If I find that after using the C&D estimator, effects are no longer significant, while they are in the usual TWFE setting, how can I interpret this result?

Thanks in advance for your help!