search

insongkim / PanelMatch

113 stars 34 forks source link

Lead in PanelMatch #54

Closed pzy0337 closed 4 years ago

pzy0337 commented 4 years ago

The current version returned an error message when I specified lead as a vector (e.g., 0:2). The message also told me to run the function for each increment of lead. Will this issue be fixed in the future?

adamrauh commented 4 years ago

I could be wrong, but this sounds like an old error. Could you try updating?

sergimartinezeui commented 4 years ago

Hi,

When including negative values to the specification of 'lead' part of the PanelMatch command (e.g., lead = -2:1), I get results including pre-treatment placebos (t-2, in that case) compared to the baseline year (t-1). I guess that could be useful for assessing how well matching works for balancing samples and evaluating the pre-treatment parallel trend. I'm not sure, however, to what extent can this be appropriate given that the t-2 estimate is compared to t-1, if I understood it correctly. Is there a way to vary the baseline year?

Thanks!

plotmatchlagsestimates

adamrauh commented 4 years ago

@sergimartinezeui After looking a little bit into this, the package currently does not support this kind of functionality -- so I pushed a patch that will throw an error if the user provides a negative value in the lead argument as a temporary measure.

That said, we think this would be a useful feature and will work on adding this in relatively soon, most likely. I created a fresh ticket to track progress here