When estimating the income effects of sustainable agroforestry,
endogeneity becomes a major challenge. However, Dasgupta et al.
(2005) indicate that a model on income effects can be run if the causality
and robustness of the model are given. Since the two stage least
square estimation showed no endogeneity between sustainable behavior
and income per capita,6 an ordinary least square (OLS) model was
run. The results of the OLS estimation are indicated in Table 6 together
with the estimates of the quantile regression for comparison.
The OLS results show that households achieve a higher income per
capita if they extract firewood sustainably. However, the quantile regression
provides more information. Regarding the lowest percentile
the model indicates a change in income impacts. Here, the very poor
farmers (0.05 percentile) achieve a lower income when they extract
wood sustainably (−0.44). In other words, if the very poor extract
more firewood than the growth rate allows, they generate higher income
compared to other farmers included in the lowest percentile,
who harvest firewood sustainably. Considering that the poor are more
dependent on natural resources than the wealthier households, they
tend to extract more natural resources than their growth rates allow.
In this way, the rural poor can sustain their living standard in the short
run, but they face the danger of running into the poverty–environment
trap. As a consequence, their natural resources will decline and it will
become more difficult for them to maintain their living standard in the
long run.
In contrast, the better-off households (0.5–0.75 percentiles) can
afford to extract firewood below the growth rate. For comparison,