Self-selection bias in estimating the determinants of landowners' Re-enrollment decisions in forest incentive programs

Despite increasing attention in recent years, only a limited number of studies have investigated the determinants of landowner re-enrollment intentions in conservation incentive programs. However, none of these studies controlled for the potential self-selection of participants. The concern for a self-selection bias is policy relevant because researchers and policymakers investigate the determinants of re-enrollment not only to predict the retention rate of participants but also to promote the long-term success of conservation programs. This paper uses data on eligible landowners, consisting of both participants and non-participants, from a forest incentive program in Japan to examine the determinants of the participants' re-enrollment decision, controlling for a rich set of observable landowner attributes, and conditioning on the unobserved participants' attributes which are identified by modeling the re-enrollment decision jointly with the decision to participate. The empirical results indicate that the unconditional marginal effects from the separate re-enrollment model are biased by selection and underestimate the effects by between 12% and 48%. The results also show that the observable factors that attract landowners to participate also tend to encourage participants to remain in the program. This implies that interventions directed at increasing initial participation are also likely to increase re-enrollment.