Manual matching for difference-in-differences

[Johannes Henkel]

Hello, I’m trying to run DiD regression on matched pairs. I have matched these pairs manually based on a number of characteristics. These pairs share the same Match_ID value. (see screenshot). I have a total of 280 rows in excel (140 matches). Would appreciate help on how to run the DiD on these pairs and how to best visualise the results. Thank you

[Fei Wang]

If all pairs are well organized, then you may directly implement DiD estimation as below.

reg y treated##time control_variables, vce(cluster clubid)

The coefficient of treated#time measures the treatment effect. BTW, "Time" should be 1 in the last line of your screenshot.

It's difficult to visualize something for a two-period DiD. It's usually useful to visualize DiD with multiple pre-treatment periods to verify the parallel-trend assumption.

[Johannes Henkel]

Fei Wang, thanks for your answer! Preliminarily results seem insignificant but at least I’m on the right track!

The two values (pre/post) are averages of 3 years pre and 3 years post. So in total 6 values per club.