Nearest neighbor matching without replacement - Binary dependent variable

[Marie-Pier St-Vincent]

Greetings,

I am new to STATA and would like to conduct one of my methodologies presented in my thesis. I have a panel data, including financial and operational metrics on different time periods of many companies according to their method of going public. The independent binary variable = 0 for traditional IPOs and =1 for companies that went public by merging with a SPAC. Thus, I believe my dependent variable is also the treatment. I would like to do a nearest neighbor matching without replacement, from a logit regression, and I would like to match on size, age and industry. The goal is to obtain a newly constructed control group sample from which I'll be able to draw accurate comparison of the 2 samples financial and operational metrics. I thought about using psmatch2, but I struggle with the syntax. Can anyone help me with this? Thank you very much!

[ericmelse]

Dear Marie-Pier,

You should consider the user community contributed package kmatch by Ben Jann:

* Set up
ssc install kmatch, replace
h kmatch

In the help file you will find several examples.
Here is a presentation of Ben that is helpful, in my opinion, presented at the UK User Group meeting in London, September 7–8, 2017: Kernel matching with automatic bandwidth selection