Studies document that forest farm households participate in afforestation programs to increase economic well-being. Moreover, higher program subsidies are expected to raise their coverage rate. However, afforestation programs in Taiwan contradict this consensus. Although 71% of forest farm households participate in an afforestation program, participants have lower forest farm incomes compared to non-participants. This study provides an answer for this empirical puzzle by showing that forest farm households not only consider the mean level, but also the stability of the program subsidy in deciding to participate. Using the instrumental variable model, we estimate the causal impact of an afforestation program on forest farm income and income inequality using a population-based sample of forest farm households in Taiwan. Results show that participation in the afforestation program decreases the mean level of forest farm income by 18%. However, it also reduces the variance in forest farm income by 30%. Thus, the high coverage rate of the afforestation program is likely driven by its effect on income stability. We also apply the recently developed profile analysis method to categorize the sample of forest farm households into compliers and non-compliers. The coverage rate of the afforestation program can be improved by targeting forest farm operators that are female or have low levels of education or more children.