A bug in the DID?

[kzmh78]

Hi, I did the Callaway and Sant'Anna (2021) example on page 1294 of Users Guide2.
Then, on page 1296 of the Users Guide, the coefficient of TREATED is -0.036549, but my estimate is 5.776771 (see eq01).

However, as is, if I do the CS Group-Time Effects on page 1296, I get the results on page 1297 (see spool_eq01).

Is it because my procedure is wrong that the TREATED coefficient is not -0.036549? Or is it an error in EViews13?

[EViews Gareth]

Bug in EViews for that initial estimate - thanks for reporting, we'll issue a fix asap.

[EViews Gareth]

Patch released

[kzmh78]

Thank you for releasing the Patch !
I estimate using Nov 7 2023 build ver, the coefficient of TREATED is now -0.036549, as written on page 1296 of the Users Guide2.

[EViews Gareth]

Great. You can verify that is the correct number by estimating a two-way fixed effects Least Squares model with a constant and Treated as the regressor.

Command:

Code: Select all

equation eq_test.ls(cx=f, per=f) lemp c treated
show eq_test