Should we test anticipation effect for Diff-in-Diff in natural event?

[Phuc Nguyen]

Normally, the anticipation effect is tested as an assumption for Diff-in-Diff. However, it seems that it is necessary for examining the impact of laws on firms' behavior or else. For example, we want to test if the anti-collusion affects firms' assets growth. Therefore, we expect that firms will know about the establishment of the laws(due to media, government discussion, insider information), leading to firms change their assets growth before the actual event dates.

On the other hand, I am wondering if we need to test anticipation effect if the event date is a natural event ( not a laws, etc), saying an example that the level of vaccinated people/population is 30% or a tsunami, I do not think anticipation test is needed, can I ask your opinion then?

[Clyde Schechter]

These are matters of judgment and I think they have to be decided case by case, not be general rules. I would, for example, treat the two examples of natural events you mention differently.

Level of vaccinated people/population reaching 30% is, at least approximately, foreseeable. Before it happens, those agents who have some stake in that event will be watching the statistics rise towards 30% and make projections of when it might be reached. They won't always be right, but they are likely to make plans based on their projections. An anticipatory effect is possible, maybe even probable. By contrast, the timing of tsunamis is not predictable, and there is no way there can be an anticipatory response, other than, in general, those in tsunami-prone areas are likely to, in general, take certain precautions at all times--but that wouldn't show up as an anticipatory effect in an analysis anyway. I would see no need to look for anticipatory effects in this situation.