Heterogeneous effects of livelihood strategies on household well-being: An analysis using unconditional quantile regression with fixed effects

Using a household panel dataset for the 2008-2016 period, we analyze the heterogeneous effects of livelihood change on household well-being in rural Vietnam. We use an unconditional quantile regression (UQR) model with fixed effects to control for unobservable time-invariant household characteristics. We find that when a fixed-effects estimator is employed, households switching from a crop livelihood to any non-crop livelihood (e.g., livestock, wage-earning, nonfarm or private transfer livelihoods) increase their per capita income and food consumption. However, the results from the UQR with fixed effects reveal a significant variation in the effect of such a switch in livelihood across various quantiles of well-being distribution, with a larger effect for poorer households. The income effect, however, tends to decline with higher quantiles and even turns negative with a switch to a wage-earning or public transfer livelihood for the better off. Notably, our study confirms the advantage for the poor of changing livelihood from crop to non-crop activities in rural Vietnam. Our research results also suggest that a mean regression approach, that often assumes a homogeneous/mean effect of livelihoods on well-being, may miss some heterogeneity that is useful to researchers and policy makers.