Off-farm work decisions of farm couples and land transfer choices in rural China

ABSTRACT

Although previous studies have widely examined the association between off-farm work participation and land use behaviours of rural households, little attention has been paid to the effects of the joint off-farm work decisions of farm couples on land transfer choices. This study investigates the determinants of farm couples’ off-farm work participation, using a seemingly unrelated bivariate probit regression model and survey data collected from rural China. We also estimate the impact of the joint off-farm work decisions of farm couples on land transfer choices by employing a multinomial logit model and controlling for the endogeneity issues of off-farm work variables. The empirical results show that farm couples are jointly making decisions to work off the farm, but their decisions affect household land transfer choices differently. In particular, we show that the husbands participating in the off-farm work are more likely to rent in land, while their wives are less likely to do so. Both the husbands and wives are more likely to rent out land if they work off the farm. Our findings highlight the importance of farm couples’ off-farm work decisions in stimulating the development of rural land rental markets.

KEYWORDS:

• Off-farm work
• farm couples
• land transfer choices
• SUBP
• MNL

JEL CLASSIFICATION:

• R14
• R23
• C81

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ If both $T_{12}>0$ and $T_{22}>0$, Equation (2)(2) $T=T_{i1}+T_{i2}+L_i\left(i=1,2\right)$(2) would imply that both a husband and his wife are working off the farm. However, if either $T_{12}>0$ ($T_{22}=0$) or $T_{12}=0$ ($T_{22}>0$), Equation (2)(2) $T=T_{i1}+T_{i2}+L_i\left(i=1,2\right)$(2) would suggest that either the husband or his wife is working off the farm.

² Following previous studies (Browning 2000; Abdulai and Delgado 1999), we employ a stylized two-member (husband and wife) model in this study to simplify the analysis. We note that there may exist other household members, some of whom may work off-farm. However, we assert that generalizing to additional household members will not affect the primary intuition to be gained from our theoretical model.:

³ We assume that the total land cultivated, B, is composed of owned land ($B_{own}$), land rented-in ($B_{in}$) and land rented-out ($B_{out})$, i.e. $B=B_{own}+B_{in}-B_{out}$. Unlike land rented-in and land rented-out, the owned land itself does not directly change the production costs but we consider it in B to ease our interpretation.

⁴ The reservation wage for off-farm work is the marginal value of the individual’s time when all of it is allocated to either the farm and leisure (Owusu, Abdulai, and Abdul-Rahman 2011; Ma, Abdulai, and Ma 2018).

⁵ The decision whether or not to work off-farm is at the discretion of the husband and wife. This fact leads to issues of endogeneity with ${\theta }_1$ and ${\theta }_2$ in specification (11). As discussed in the section below, this study employs a control function approach to address the endogeneity issues of ${\theta }_1$ and ${\theta }_2$ in our efforts to empirically examine the impact of farm couples’ off-farm work decisions on the household land transfer choices. For clarity, it should be noted that it is possible that endogenous variables exist in a reduced form of the theoretical model.

⁶ Among all sampled households, only 17 of them both rent in and rent out land, and in this case, it is not possible to model it as a separate group. To ease our analysis, we clarified a sample to the rent-in group if the land rented-in is larger than that rented-out by a household, and vice versa.

⁷ The results of the coefficients estimated by MNL are presented in Table A3 in the Appendix.

⁸ In rural China, men migrate at a higher rate to seek high-salary job opportunities than women and women are usually left with responsibility for the family farm. Thus, it is possible that husbands working off the farm increases household income which can then be used to rent in land.

⁹ Although the coefficients of household size variable are not statistically significantly in Tables 3 and 4, we have kept this variable in our model estimations because it is one of the essential variables capturing household-level demographic characteristics.

Additional information

Funding

Xiaoshi Zhou acknowledges the research fund support for Young Teachers of School of Economics and Management, Nanjing University of Science and Technology (No. JGQN2010). Wanglin Maacknowledges the financial support from the Faculty of Agribusiness and Commerce at Lincoln University within the Seed Fund Project (INT5066). Gucheng Li acknowledges the financial support from the National Natural Science Foundation of China (No. 71873050).