Spatial Patterns and Determinants of Agricultural Resilience: Evidence From Senegal

ABSTRACT The undesirable consequences of climate change on crop yields threaten the resiliency of farmers' livelihoods in climate‐vulnerable regions. Assessing the resilience of agrifood systems to climate and non‐climate hazards helps identify solutions for ensuring the sustainability of farming households. The literature review indicates that a knowledge gap remains in interpreting outputs generated by procedures under various study‐specific conditions. A review of selected articles from 1547 documents on resilience among Senegalese farmers identified relevant indices representing farmers' resilience from nine studies, resulting in 83 observations for the resilience index and control variables. This study utilized spatial meta‐data and survival regression analysis to examine the effects of regional interactions, shock types, and factor selection on measured resilience through the following phases: (1) Organizing the meta‐data, (2) specifying eight meta‐regression models to assess bias from regional data variations and the interaction effect of sample size, (3) converting meta‐data to survival data to analyze resilience failure exposure and time‐to‐event failure, and (4) regressing the shock types and agroecological zone conditions on the outcomes from phase three. The results indicated that the “climate hazard” shock, “COVID‐19” shock, and “seed diversity effect” were the primary contributors to the highest failure of resilience capacity. The spatial lag significantly affected resilience magnitude. Accounting for the spatial lag changed the negative effect to a positive effect for variables representing different shock types. For example, when accounting for the spatial lag, the impact of “climate hazard” and “other shock sources” shifted compared to the “COVID‐19” shock, indicating that their influence on resilience capacity changed direction. The effect of shock‐type variables on resilience failure exposure was significant, regardless of whether the shock sources remained constant or changed. The findings emphasize the need for policy considerations regarding measurement procedures, regional factors, and shock‐specific interventions to avoid overestimation or underestimation of resilience. For instance, resilience measurement procedures should be improved by distinguishing between permanent and temporary shocks, as well as by considering the vulnerability of interacting regions in comparison to isolated regions. Failure to incorporate these factors may result in an overestimation of resilience for “non‐climate” shocks.