The symmetry is naive. The only way I'd be completely confident in it is if the branching factors matched, i.e. the distribution of "number of employees per employer" mirrored the distribution of "number of employers per employee".

Absent that, you need to prove assymmetries cancel out, which has a higher burden of proof.