I was using teffects psmatch for matching unit-by-unit, for example firm-with-firm. From the code, I saw that we should put all covariates to control and the accurate the match.
The match is nice during matching progress when I gradually put the firm-level variable into the code; however, when I put the country-level control variables (e.g., ln(gdp)), the code turns error:
I am wondering why it happened and should we include country-level to propensity matching score for firm-level units?
Many thanks and best regards.
The match is nice during matching progress when I gradually put the firm-level variable into the code; however, when I put the country-level control variables (e.g., ln(gdp)), the code turns error:
Code:
logit failed to estimate propensity scores; computations cannot proceed
Many thanks and best regards.
