26
VI. Heterogeneous Program Effect on Off-Farm Labor
In the previous section, we found—with at least some degree of confidence—that the Grain for
Green program has led to an increase in off-farm labor participation. The DID estimates, however, do
not allow us to understand how the program affects off-farm labor or which types of farmers are
participating. In fact, we are interested in understanding how these changes occur. In particular, based on
the stylized conceptual model, we want to understand the role of two factors when households make
off-farm labor-participation decisions: physical capital and human capital. In this section, we test
whether the program has heterogeneous effects on off-farm labor that depend on the availability of
physical and human capital to the households before the program. To do so, we estimate equation (2).
Liquidity Constraint
We find that the effect of the program on off-farm labor is clearly larger for households that had
less liquid assets prior to the program (Table 7, columns 1 and 3). For households belonging to the
quartile of households with the lowest level of assets, the program increased off-farm work by an
average of 0.52 persons (column 1). Intuitively, this means that one adult member in one out of every
two liquidity-constrained participating households started to work off-farm after joining Grain for Green.
In contrast, although the program had a positive effect on off-farm employment decisions by less
liquidity-constrained household in the other three quartiles, the estimated coefficients are mostly
statistically insignificant. Estimates of the coefficients at the individual level are consistent and show
even stronger results compared to the household-level findings (column 3). The program increased the
probability of a household member starting an off-farm job by 20 percent for households in the two