Propensity score matching problem

[Johnny YANG]

I'm doing a propensity score matching estimation. The code is as follows,
psmatch2 highMacroSubFirm OperatingLeverage Markup PriceResearch OperatingLeverage, outcome(Value) noreplacement logit ate
psgraph
pstest OperatingLeverage Markup PriceResearch OperatingLeverage
My problem is whatever matching factors I try, there are at least more than half of the observations are off support (19000/31000). Please help me with this, thank you!

[Johnny YANG]

The problem has been solved. Yesterday I divided the sample with mean value, thus the highMacroSubFrim number is way lower that the number of lowMacroSubFirm. Today I tried to divided the sample by the median value. It worked! Hope others can learn from my mistake.

[Ain Qurat Ul]

Hi ..
I am a beginner. I am also using PSM But I have no o
idea how to take the start.
I Need to run a logit model on the basis of a binary variable including All governance ，ownership and control variables.
After this to apply PSM. Would you help me which command I can use in stata？
for logit model，caliper matching and PSM ？

[Johnny YANG]

Hi ..
I am a beginner. I am also using PSM But I have no o
idea how to take the start.
I Need to run a logit model on the basis of a binary variable including All governance ，ownership and control variables.
After this to apply PSM. Would you help me which command I can use in stata？
for logit model，caliper matching and PSM ？

psmatch2 governance-binary-variable-here all-your-control-variables-here, outcome(variable-you-want-to-compare-here) noreplacement logit ate

sorry for the late reply. I haven't logged in for the recent days. Hope it helps!

[Johnny YANG]

Hi ..
I am a beginner. I am also using PSM But I have no o
idea how to take the start.
I Need to run a logit model on the basis of a binary variable including All governance ，ownership and control variables.
After this to apply PSM. Would you help me which command I can use in stata？
for logit model，caliper matching and PSM ？

After the code I wrote in the last reply, you will get some results like "ATT" "ATU" "ATE" which are the effect size of the binary variable on the variable-you-want-to-compare. ATT are the effect size within the treated group, the other two are the effect sizes of the untreated and average group.
Then if you want to test the matching quality, you need to run tests such as,
psgraph, which will tell you how many entries are matched and unmatched
pstest all-the-controls-here, which will tell you how biased is the matching for each control variable.