Coarsened exact matching (CEM) on panel data

[Shruthi Venkatesh]

I have panel data at the individual-quarter level for the years 2007-2012. I'm not able to use dataex because I do not have the rights to share these data (even if anonymized).

Here's a basic format:

input ID y qofy treat treattime X1 X2
1 .333 178 1 0 13423 134
1 .23 179 1 1 23423 234
1 .323 180 1 1 23423 334
2 021 166 0 0 98332 445
2 .211 167 0 0 98038 234
end

I'd like to use CEM to match on the variables X1, X2. I'd like to generate weights, and then run a panel differences in differences regression with ID fixed effects to find an effect of treatment (treat) on my independent variable (y).

I'm guessing I'd need match exactly on qofy and then coarsely on X1 and X2. However, I'm not sure how to run the regression post-estimation to identify the effect of treat on y. How would I go about doing this?

[Shruthi Venkatesh]

I was able to figure this out using the examples from: https://gking.harvard.edu/files/gkin.../cem-stata.pdf

[Stephanie Stahlberg]

Hi Shruthi. Can you share more about what you found? I have read the above article but I still have not figured it out. Thanks