households that own less paddy land, own more bullocks and have a male head of household are more
likely to be sharecroppers. The probability of fixed rent tenancy is also positively related to the ownership
of machines and the presence of male adults. The tobit estimates in the third and fourth columns show
similar results.
The different effects of bullock and machine ownership confirm that each contract is used to
adjust different inputs.31 Apart from these differences, the results for share-rents and fixed-rents are
qualitatively similar. As expected, the ownership of land has a significantly negative effect on both types of
leasing. This confirms that households lease-in land in order to achieve the optimal operational scale
consistent with its characteristics. The ownership of highland lands, which are used mainly for the
cultivation of non-seasonal crops, vegetables and dry grains, also has a negative but considerably smaller
effect on leasing. This can be explained by the fact that highland cultivation is an imperfect substitute for
rice cultivation in utilizing some unmarketable resources of the household, especially labor and
management, but not others that are specific to the wetlands (e.g. bullocks). In addition, due to the greater
intensity of input use in rice farming, the elasticity of substitution of inputs between highlands and paddy
lands is likely to be less than unity. The labor endowment of the household, measured by its size, has a
positive effect on both types of leasing indicating the household labor is an unmarketable resource.
However, the gender composition of the labor endowment has different effects on each type of contract.
Among other household characteristics, education of the head, the proportion of students and engagement
31 A potential criticism is that family size and the ownership of assets may be endogenous, i.e. those who lease-in
land purchase more inputs. In the absence of convincing instruments, we rely on the empirical observation that many
of these inputs, especially family labor and farm animals, are largely exogenous in the leasing model. Although the
choice of family size is theoretically endogenous in the household’s decision making process [Evenson and
Roumasset 1986, DeSilva 1998], the large gestation period and other considerations associated with fertility (both
economic and social) would make the adjustment of family size by and large exogenous to the land leasing process.
Similarly, it has been documented that bullocks are usually transferred within families and seldom sold. The
endogeneity of farm machinery is a more likely problem. Here, we rely on the argument that the purchase of large
assets (e.g. tractors, vehicles) is unlikely to depend on leasing choices because the extent of land leasing-in is quite
small (on average 0.35 acres, or about 30% of the average ownership holding). Therefore, we include only relatively
large and lumpy assets as exogenous regressors, leaving out small farm implements such as mammoties, sickles and
water pumps which are much more likely to be endogenous.
36