I wonder if there is a way to identify the MDE (Minimum Detectable Effect) in a propensity score matching (PSM) model for the average treatment effect on the treated (ATT). I am using STATA 14 with the command psmatch2. But I can learn another software. Suppose that
T= treatment status (1 treated , 0 control) X= confounders Y = output
First I estimate the propensity score
logit T x1 x2 .... xn predict scoring
Then I use the command psmatch2 to find the ATT.
psmatch2 T , kernel out( Y ) pscore(scoring)
I know that there are some formulas and software to estimate the MDE and the power for randomized experiment and regression discontinuity, but I don´t know a similar formula for matching.
Suppose that I find a ATT effect of 3% that is not statistically significant, in this case I want to know if $ ATT<MDE$.