Just so this is fully clear to me: The model you want has 4 equations
ln w_a = b1x1 + e1 wage-offer abroad
ln w_f = b2x1 + e2 wage-offer home
ys1= b3x2 + e3 participating in labour market if ys1>0
ys2= b4x2 + e4 migrate if ys2>0
with (e1,e2,e3,e4) distributed multivariate normal.
Am I correct? This is similar in spirit to Heckman and Sedlacek (1986)
model of static sectoral choice. It has been estimated by Nielsen and
Westergaard-Nielsen (2001?) and Paul Glewwe (1999) in his book on
school quality in Ghana.
I estimated such a model in my MA thesis on return to human capital in
rural Pakistan, and have coded it for Stata's ML command if your are
interested.
