Cholesky decomposition with equal exogeneity

[Zachary Bartsch]

Say that I have 3 variables (or more), x1, x2, & x3 and I want to conduct a VAR with some lag length (say 3).
Say that I know, thanks to theory, that x3 is endogenous to both x1 and x2.
Say that I don't know whether x1 or x2 is relatively more exogenous.
Solutions that I can find online suggest running multiple VARs with different cholesky orders and examining the robustness of results.

*But*, what if I have good reason to believe that x1 and x2 are equally exogenous?
Is it possible to order the cholesky decomposition with an option of something like:

irf set filename
var x1 x2 x3, lag(1/3)
irf create irfname, order(x1 & x2 x3)

Clearly, as typed, that last line won't work.
Could I adjust the cholesky endogeneity order manually?
I'm using Stata 15. Thanks!

[Justin Niakamal]

I don't think you can do that. But do you get different results when x1 and x2 are swapped? Some software have GIRFs, which circumvent the need for ordering.