Smallholders in the Global South are often excluded from their land rights by exclusionary policies of conservation, which are intensifying in the face of a changing climate. Such conflicts are often the most problematic where they overlap with the extant political-marginalisation of the smallholder, articulated in ethnic terms. The world’s Indigenous Peoples, being among the most vulnerable to dispossession while also tending to inhabit forested spaces, are of central importance to these debates. South Asia comprises a “natural laboratory” of different institutional arrangements that mediate the relationship between the State and the land and forestry rights of Indigenous Peoples. This study uses ethnographic methods, remote sensing analysis of forest cover change and archival research to test whether this institutional variation leads to differences in dispossession and deforestation, and offer conclusions as to why this is the case. The Garo community inhabits both Meghalaya state in India and the Madhupur region of Bangladesh, with strong and weak legal protections over lands and forests respectively. The study finds that dispossession and deforestation outcomes are worse in Bangladesh than in India. This difference results institutionally from a lack of recognition of the legitimacy of Garo land rights in Madhupur which in turn stems from the local, ethnically-bounded political-economy, the “Predatory State”. This comprises an assemblage of local political elites and State bureaucrats for whom both the Garo smallholding and the original forest represent untapped resources that dispossession releases into the dominant political-economy. In contrast, Garo land rights are afforded full legal recognition in Meghalaya, resulting in low levels of dispossession and a conservation regime that is robust because ownership of protected areas remains vested in the community. This situation is deteriorating as land values increase, due to an “Absentee State” that does not provide robust and accessible dispute resolution processes, resulting institutionally in community rights being only partially embedded in the wider administrative structure. The main underlying cause is that institutional dysfunction was deliberately designed into the system to allow the State to “recentralise while decentralising”.