Livelihood diversification among second-generation transmigrants: a compositional analysis

Purpose. This study examines the composition of household income across on-farm, off-farm, and non-farm sectors among second-generation transmigrant households and analyses the factors associated with the relative distribution of income across these livelihood sectors. Particular attention is given to the roles of natural capital, human capital, and demographic characteristics in shaping diversification pathways. Methodology. A quantitative approach using Compositional Data Analysis (CoDA) is employed to account for the interdependence among income shares. Income compositions are transformed using Additive Log-Ratio (ALR) and Isometric Log-Ratio (ILR) techniques. Two Ordinary Least Squares (OLS) models are estimated for the off-farm/on-farm and non-farm/on-farm income ratios, with ILR models used as robustness checks. The analysis is based on primary data from 303 second-generation transmigrant households in Sungai Bahar Subdistrict, Jambi Province, Indonesia, drawn from Mekar Sari Makmur, Marga Mulya, and Berkah, which represent short, medium, and long distances, respectively, from the subdistrict centre. Results. Land ownership and education emerge as the most systematic correlates of household income composition. Land ownership is associated with a stronger on-farm orientation, reflected in lower off-farm/on-farm and non-farm/on-farm income ratios. Higher education is associated with a lower off-farm/on-farm income ratio but a higher non-farm/on-farm income ratio, indicating distinct diversification pathways between the two non-agricultural sectors. Lower household-head age is associated with a higher off-farm/on-farm income ratio, whereas household size and village distance show more model-specific associations. The consistency of the main patterns across the ALR and ILR specifications supports the robustness of the findings. Originality. This study contributes to the livelihood diversification literature by analytically distinguishing off-farm from non-farm income rather than treating them as a single non-agricultural category. It also applies a compositional framework that explicitly accounts for the interdependence of household income shares and provides new evidence on how current land access, education, and household characteristics are associated with livelihood trajectories among second-generation transmigrant households. Practical implications. The findings support policies that strengthen education, vocational training, and skill development to facilitate access to non-farm livelihood opportunities. They also highlight the continuing importance of productive access to agricultural land for households that depend on farming. More broadly, livelihood policies should distinguish between off-farm and non-farm pathways because they are associated with different household characteristics and livelihood conditions.