Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: margins returning "not estimable"

From   Suryadipta Roy <>
Subject   st: margins returning "not estimable"
Date   Tue, 14 Aug 2012 08:17:32 -0400

Dear Statalisters,

I am running the following (country-pair and time) fixed effects
regression (sample size > 100,000) :
areg BilateralTrade Treatmentt##c.ContinuousVar1
Treatment##c.ContinuousVar2 ExporterIncome ImporterIncome
ExporterPopulation ImporterPopulation Distance ImporterExchangeRate
ExporterExchangeRate GSP LegalOrigin CommonCurrency Colony
TimeDummies, absorb(pair) robust cluster(pair);

This is followed by - margins Treatment - which is returning "not
estimable" results.
- matrix H = get(H)-  followed by - matrix list H - reveals that the
hidden H-matrix (as referenced in the Base Reference manual) after
-areg- consists of values outside of -1, 0, and 1.

However, -margins- works when the following variables:
ImporterExchangeRate ExporterExchangeRate GSP LegalOrigin
CommonCurrency Colony, are dropped from the model. The H-matrix now
consists of only -1, 0, and 1. I have read the following thread which
suggests that is appropriate to use the -noestimcheck- option after
-margins- in certain situations:

In my case, -noestimcheck- does yield the necessary marginal effects.
I would very appreciate if someone can kindly comment if it might be
appropriate in my situation to use this option in order to arrive at
the marginal effects.

As usual, thank you very much for all the comments and suggestions!
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index