Table 11 presents F-tests for the first stage of the 3sls procedure46 and shows that our
instruments are closely related to the variables they are instrumenting for, limiting the
potential for weak instruments, especially in the specification with many controls.
____________Table 11__________________
First-stage F-tests for the instruments_______
_________Endogenous variable_________
Temp. Mig. Perm. Mig. Intern. Mig.
F-Stat. 10,97 9,93 24,31
P. Value 0,00 0,00 0,00
With respect to results shown in Table 10, all explanatory variables of the propensity to adopt
modern seeds have the expected sign; in particular, the amount of land operated is positively
correlated with the propensity to adopt, while tenure insecurity (the share of temple and
mortgaged-out land) has a negative impact on that. The percentage of irrigated land and
power means of ploughing have a significant positive sign, whilst the negative sign of the
land-labour ratio variable suggests that HYVs are more labour-intensive crop than traditional
ones47.
The impact of migration on the propensity to adopt high-yielding varieties of seeds depends
on which type of migration households participate in and, in turn, on the determinants of the
migration decision. Our findings show that migration significantly affects agricultural
technology upgrading; however, while having a household member migrated abroad has a
positive effect in fostering household propensity to adopt modern and risky seeds, domestic
temporary and permanent migration have a negative impact on the adoption propensity in
source households. Therefore, if migration is a risk and credit-alleviating strategy, this seems
not to be true for all types of migration in Bangladesh. Explanations for this may lie on both
risk and credit arguments: indeed, overseas economic opportunities are likely to be less
correlated with local earnings and provides much higher returns (remittances) than it is the
case for domestic migration.
In this sense international migration acts as a shelter against local uncertain income prospects,
as predicted by the NELM perspective and shown by our results. However, given that
migration is an endogenously shaped process, this seems not to hold for all farm households.
46 They test for the joint significant of the instruments in regression of the endogenous migration variables on all
exogenous variables in the system.
47 For a deeper discussion on determinants of HYVs adoption in Bangladesh see Mendola, 2003.
33