The third and fourth columns in Table 7 present the pseudo-
R2 from the propensity score estimation before and after
matching. The likelihood-ratio test of the joint significance
of all the regressors in the probit model of propensity score
estimation before and after matching and their corresponding
p-values are presented in the fifth and sixth columns of the Table.
The corresponding p-values of the likelihood-ratio test
show that the joint significance of regressors on treatment status
could always be rejected after matching. It was, however,
never rejected before matching. The relatively low pseudo-R2
after matching and the p-values of the likelihood-ratio test
of joint significance of the regressors imply that there is no systematic
difference in the distribution of covariates between
both groups after matching.
To gain further insights into the differential impact of adoption
on farmers belonging to different land ownership catego-ries, we also analyzed the causal impacts of adoption on PCE
and poverty status for different categories of land ownership. 15
The nearest neighbor estimates, which are presented in Tables 8
and 9 for Chiapas and Oaxaca, respectively, generally reveal
that even within the different farm size groups, adoption tends
to positively and significantly impact on PCE and negatively on
poverty level. 16 Specifically, the results indicate that adoption
of germplasm exerts a positive and significant impact on PCE
and negative impact on poverty, with declining impact as land
ownership increases in both Chiapas and Oaxaca. This finding
is consistent with the notion that poorer farmers tend to benefit
more from new agricultural technologies.