27
lowest quartiles, while the effect was statistically insignificant for individuals in the higher two quartiles.
In contrast, estimates for on-farm work suggest that households and individuals in the lowest-asset
quartiles moved away from on-farm work (columns 2 and 4). The magnitude of the coefficient gets
steadily smaller as the level of assets in the quartile categories gets higher (although the increase is not
linear).
We found consistent results when we split the households using Zeldes’ rule into
liquidity-constrained and -unconstrained groups and compared the DID estimates. The DID estimates for
the constrained group was positive and statistically significant both at the household and individual
levels. The DID estimates for the unconstrained group were insignificant.16
In sum, the findings reveal that the less liquidity-constrained a household is prior to the
program the more positive the impact of the Grain for Green program is on its off-farm employment
participation. One way of interpreting this result is that participation in Grain for Green relaxes a
household’s liquidity constraint and that it garners resources the household can use to participate in
off-farm work. Thus, the more constrained the household, the larger is the program’s impact on off-farm
work.
Human Capital
We also are interested in understanding how human capital can influence the program’s effects
among households. Age and education are two fundamental indicators of human capital that affect the
ability of individuals to find off-farm work. Higher education is expected to result in greater rewards
from off-farm labor (Becker, 1993). Education here is defined as the number of completed years of