where φ(.) is the density function for a normal variable.
In the case of bivariate probit model, several marginal
effects can be computed. We are interested mainly in the
change in probability of continued adoption: Prob(continue=
1)= Prob(continue=1, gap=1)=P11. The marginal effects
are derived using derivative rules of the bivariate probability
function (Baslevent and El-Hamidi, 2009):