panel data with three equations

[James Paul]

Hi all,

I am trying to solve the following equation system. The data is panel and I will include fixed effects (id, time)


depvarA = z x1 x2 i.id i.time (eq.1)
depvarB = depvarA x1 x2 i.id i.time (eq.2)
depvarC = depvarB x3 i.id i.time (eq.3)


If there are only first two equations, I would try the following
xi:xtivreg depvarB x1 x2 i.time (depvarA = z), fe cluster(id)

However, I have no idea when it comes to solving three equations with panel data.
The only thing I know is reg3, which does not allow robust standard errors.

One thing I can think of is combining eq1 and eq2 like
depvarB = z x1 x2 i.id i.time

However, I do want to see how variable z impacts depvarA, which again impacts depvarB, which impacts depvarC.

[Phil Bromiley]

Welcome to Stata list. You will increase your chances of useful answer by following the FAQ on asking questions – provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

You can estimate this model an equation at a time simply by instrumenting the endogenous variable in each equation with exogenous variables. This is not as efficient as three stage least-squares or maximum likelihood on the system, but the estimates in a given equation are not then dependent on all three equations being correctly specified. Also, cmp might do it. Or it can be done explicitly in SEM. There are probably some other estimators as well, but I can't think of them immediately.