In a first approach, the two decisions, adoption and
continued participation, could be specified independently of
each other using two separate probitmodels. Using a utilitarian
framework, the utility of adopting (qgap*) and the utility of
continued adopting (continue*) are dependent on vectors of the
explanatory variables x1 and x2. If we assume a linear model,
this description translates into two linear relationships:
qgap* = x1. β1 + ε1 and continue* = x2. β2 + ε2, where β1
and β2 are the parameters of the linear relationship to be
estimated. However, these utilities are not directly observable,
as the only information available is the decision to adopt (qgap)
and continue participating (continue).With a realistic hypothesis
that farmers will adopt and continue participating when
their decisions translate into positive utility gains and if the
decisions are independent, we need to solve a system of
independent probit models. However, a number of the factors
that influence the continued participation are linked to their
own experience: the longer the farmers participated in the