capital are highly important in shaping household decisions with respect to the subsistence
strategy of migration.
Yet, according to the NELM approach, individual family members’ labour time is
simultaneously allocated between migration and non-migration work so as to maximise
expected utility of the whole household. The interplay between the household decision to
participate in different typologies of migration and the investment behaviour of household
members left behind with respect to the adoption of risky farming technologies is the focus of
the next section.
5.2 Migration effects on source households’ agricultural productivity
Investing in high-productivity technology, such as HYVs of rice, is subject to financial and
risk constraints, which are strictly binding for small farmers living in contexts of missing
credit and insurance markets.
The NELM hypothesis that migration is a subsistence strategy enabling rural households to
overcome investment constraints and achieve the transition to high-productivity agricultural
techniques is tested through a system of equations, in order to address the issue of
simultaneous household decisions towards investment in agricultural activities and alternative
migration destinations. However, given the endogenous nature of the household migration,
instrument variables are needed in order to identify migration equations.
Thus, we estimate the potential impact of different typologies of migration, i.e. temporary (T),
permanent (P) and international (I) migration, on the farm household propensity to adopt
high-productivity seeds (as a proxy for the household risk-management capacity) through a
3sls simultaneous equations linear probability model. For identification purpose, we use the
sample proportions of households in the origin village participating to each separate migration
typology as instrument variables. The latter are proxy variables for external and social
migration networks (as explained above). Different village histories of migration -
specifically for temporary, permanent and international migration - are likely to lower
migration costs and increase the opportunities for village households to send out migrant
members, but should not affect the propensity to adopt new farming technologies39.
Based on the literature, we estimate a system of equations as follows:
Yi=γ0+γ1XiT+γJ2MiJ+εiT
(5.1)
MiJ =β0J +β1JXiM +β2JZJiM +εJiM; J=T; P; I,
39 The same instrument is used in Rozell et al. 1999.
29