Many software that handles nested logit regression such as R(mlogit) , stata (nlogit), python (pylogit,biogeme) with the exception of Gauss does not have the option of marginal effect as a post estimation test. However, the marginal effect for similar model such as multinominal logit, ordered logit etc can be executed using margin in R and stata and statsmodel in python. Does this really implies that marginal effect is not a relevant post estimation test for nested logit and if this is, I would love to have suggestion on the preferred pre - estimation and post estimation test for nested logit with references and preferred software of estimation.