Delta method to

[marianne lefebvre]

Hello

I am using stata 14 and the mixlogit command (user install) to analyse data from a discrete choice experiment with 6 attributes. I want to estimate the marginal rate of substitution (MRS) for each of the attributes, using the budget attribute (Cbudg_fa) as the numeraire.

global randvars Cuse_d Cuse_a Cnatu Cfaun_fa Cfaun_fo Cinfo Cwork_d Cwork_a Cbudg_fa Cbudg_fo
mixlogit Choice_long, rand($randvars) group(idChoicenumber) id(id) nrep(1000)

for each attribute and observation, I calculate the individual coefficients:
mixlbeta Cuse_d, nrep(500) saving("${MYPATH}\Resultats\500\newpanel_Cuse_d.dt a") replace

Then I calculate the average marginal rate of substitution
gen MRSabs_`x'=abs(`x')/abs(Cbudg_fa)

Question 1: Is that correct to take the absolute value of the coefficient to calculate the MRSs? I am interested in the weight of each attribute, independantly from the sign of the individual coefficient
Question 2: I would like to estimate the standard errors of the MRSs. How to do that? with the delta method ?

Thanks in advance for your answers.
Best,
Marianne

[Phil Bromiley]

Welcome to Statalist. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stat output, and sample data using dataex.

I'm afraid that I don't use this procedure. With user-written procedures, getting an answer often depends on someone happening to use that procedures. You may need to contact the authors. However, I don't see why you'd want to ignore the signs of coefficients in calculating the MRS. I would think the sign matters.