Dear Stata users,

I observe many groups of 3 individuals, and I want to estimate an interaction model of the form:

y_1=f_1(x, y-1)+e_1
y_2=f_2(x, y-2)+e_2
y_3=f_3(x, y-3)+e_3

Where x is an exogenous variable and y-i represents the sum of LHS variables of partners, i.e. y-1= (y_2+y_3). Importantly, the functional forms f_1 ... f_3 are non-linear in the parameters. Which is the right Stata/MATA command to estimate the 3 associated parameters (beta_0 for x, beta_1 for y-i, and the variance of the error term)? In my understanding, SEM and NLSUR aren't fit for that,

Many thanks,
M.