Estimating the causal effects of multiple treatments of crop commercialization on multidimensional poverty in observational studies

• We assessed the causal impact of different crop commercialization groups (subsistence, semi-commercial, and commercial) on multidimensional poverty among rural households. • Of the rural households surveyed, 8.1% were classified as semi-commercially oriented and 12.4% as commercially oriented. • Rural households in the subsistence group exhibited the highest unadjusted multidimensional poverty rate at 48.4%. • Compared to subsistence-oriented households, commercial-oriented households demonstrated a 3.56 percentage point reduction in multidimensional poverty. • Commercial-oriented households experience a 2.48% reduction in multidimensional poverty relative to semi-commercial households. Poverty remains a critical global challenge, particularly in countries like Ethiopia, where agriculture dominates, and rural areas experience high levels of multidimensional poverty. Understanding the relationship between agriculture and multidimensional poverty is thus vital. This study utilizes data from the Ethiopian Socioeconomic Survey to examine the causal impact of crop commercialization on multidimensional poverty among rural households. The analysis employs generalized propensity score methods to estimate multiple treatment effects. To address selection bias and covariate imbalance, several estimators were applied, including inverse probability weighting (IPW), overlap weights, matching weights, and their augmented forms. Overlap weighting yielded the best covariate balance and most efficient causal estimates, making it especially suitable for evaluating multiple treatments with moderate overlap. Among rural households, 8.1% were classified as semi-commercial oriented and 12.4% as commercial oriented. Results indicate that crop commercialization significantly reduces multidimensional poverty, particularly when extreme propensity scores are taken into account. Specifically, commercial-oriented rural households face a 3.56% lower risk of multidimensional poverty compared to subsistence-oriented households (risk difference = -0.035608 [-0.035609, -0.035607]), and a 2.48% lower risk compared to semi-commercial-oriented households (risk difference = -0.02477 [-0.02477, -0.02476]). These findings underscore the significance of agricultural commercialization in reducing multidimensional poverty and improving rural welfare. Policymakers should prioritize directing resources and infrastructure toward subsistence groups to improve their access to livelihood strategies and reduce disparities in multidimensional poverty.