The effect of agricultural productivity on poverty: what does meta-regression analyses reveal?

It is unlikely that there will be other development interventions than enhancing agricultural productivity growth to steadily alleviate poverty in many developing countries. The progress observed has been uneven, influenced by many factors. Although the literature has extensively investigated the linkages, the discrepancies in empirical results present challenges to scientific understanding and policymaking. This paper aims to estimate the effect sizes drawn from 40 eligible econometric studies conducted over a three-decade period. A random-effects meta-regression model is employed to explore what determines heterogeneities in the genuine empirical effects of the hypothesized linkages. The finding reveals that agricultural productivity has a medium- and high-level effect on poverty alleviation, as indicated by (un)transformed partial correlations. Likewise, the study finds evidence of publication bias towards a positive impact of agricultural productivity on poverty. When accounting for these biases, the genuine empirical effects vary, attributable to several factors, including the empirical setup, research design, data and publication characteristics, trajectories of key variables, the study duration, and the Cobb-Douglas production function used. Therefore, to unlock agriculture’s potential to alleviate poverty in developing countries, research should prioritize more holistic productivity measures, longitudinal data, and parametric functions, as they capture broader macroeconomic impacts and help understand the complex, dynamic, and often nuanced relationships between agricultural productivity and poverty.