computing marginal effects using logit fixed effects model

[Floris bedkom]

Dear Statalist,

My dataset is crossectional, but I want to include industry and year fixed effects. My dependent variable is binary, hence I use a logit model to compute coefficients. However, when I include these fixed effects in the model, I cannot compute the marginal effects anymore. For example, when I only include industry fixed effects, using the following code:

logit DIMPR Eindex _asize _aTobinsQ_W aleverage_W aFCF_W _tsize _relsize_W _tcash_W _tleverage_W allstock percentagecash dealsize2_W industryrelatedness _hightech i.aIndustry48, robust

mfx compute

Stata then gives the following statement:

default predict() is unsuitable for marginal-effect calculation
r(119);

Similarly, when I use the following code:

logit DIMPR Eindex _asize _aTobinsQ_W aleverage_W aFCF_W _tsize _relsize_W _tcash_W tleverage_W allstock percentagecash dealsize2_W industryrelatedness _hightech i.aIndustry48, robust

 margins, dydx(Eindex _asize _aTobinsQ_W aleverage_W aFCF_W _tsize _relsize_W _tcash_W tleverage_W allstock percentagecash dealsize2_W industryrelatedness _hightech) atmeans

Stata then gives the following statement:

estimates post: matrix has missing values
r(504);

Am I wrong for including fixed effects using a logit model? If so, what is a good alternative to fixed effects in this setting?

Thank you very much for you answer.

Floris