The land system of state-owned farms in China is different from that in rural areas. Whether the land tenure of state-owned farms can play a role in protecting cultivated land is an important issue for the high-quality development of state-owned agriculture in China. This article develops a dynamic model to examine how land tenure influences farmers' decisions on land improvement. It then analyzes this relationship based on cotton farmers' household-level data from state-owned farms of Xinjiang in China. We applied methods that take into account the possible endogeneity of the land tenure. The results reveal that the stability of land tenure in the past will not affect the current behavior of farmers for they have a relatively stable expectation of current land tenure and a high degree of trust in the government and its policies. The intergenerational transfer of land tenure is not the key factor that affects farmers' land conservation, and the relatively long-term duration of land tenure (possibly five years or more) during their careers is more important. Our findings also reveal that non-property factors, such as government intervention (e.g., technology promotion) that alleviates the limited rationality of farmers, cannot be ignored because they played a crucial role in past land improvement when land tenure was less stable.
Keywords: IV ordered probit; conditional mixed process (CMP); control function (CF); cotton farmers; dynamic model; intergenerational transfer of land tenure; land improvement; land tenure; non-property factors.