Question about estimation failure when no "nevertreated" control groups and simultaneous continuous treatment are present

[EddieLUO97]

Dear Dr. Callaway,

Thank you for developing the contdid package.

I'm currently working on a project involving a national policy that was implemented simultaneously across all observational units (cities) in a specific year. Each city received a different treatment intensity, measured by a continuous, time-invariant dose variable. As a result:

1. All units are treated in the same post-treatment period;

2. No unit remains untreated throughout the panel (i.e., no "never-treated" group exists);

3. Treatment intensity varies across units, but is fixed over time.

Given that treatment occurs at the same time for all units, I suspect that Cases 1 and 2 in the contdid framework may not be directly applicable, as they seem to rely on staggered treatment timing and the presence of a never-treated control group. However, I would appreciate your confirmation on whether this interpretation is correct.

More importantly, I also encounter an error when applying Case 3: Data-Driven Nonparametric Model for Treatment Effects, even after collapsing the panel into a two-period setup (pre/post). My understanding is that Case 3 is designed to accommodate non-staggered treatment and continuous doses, so I wonder:

Is this estimation failure due to the lack of a never-treated group, even in the Case 3 setting?

Any clarification you can provide would be greatly appreciated. Thank you again for your work on this valuable package.

[gabrielcruzfer]

Dear Dr. Callaway,
Thank you for the contdid package and paper, they have been incredibly helpful. I am working on the same setup: a one-time policy with continuous, time-invariant doses and no never-treated group. I would appreciate any guidance on how to apply in this context.
Thanks!!