<p>Cluster agriculture (CA) is an innovative, government-supported approach to enhance agricultural productivity. However, empirical evidence on its impact mechanisms, particularly on wheat Total Factor Productivity (TFP) remains limited. This study explores the impact mechanism of CA on wheat TFP using survey data from 454 farm households in the West Shewa Zone. TFP is measured as the ratio of wheat output to combined inputs, while mediating variables including agricultural input supply (AIS), market linkages (ML), and knowledge spillovers (KNS) are assessed using Likert scales. To address selection bias and quantify causal mechanisms, we apply endogenous switching regression (ESR), propensity score matching (PSM), and mediation models. The results indicate that participation in CA significantly increases wheat TFP by 0.294 units. The mediation analysis reveals that AIS accounts for 35.9% of this effect through cost-efficient access to quality inputs, followed by ML (13.5%) through improved market access) and KNS (3.2%) through knowledge spillovers. Heterogeneous effects are observed, particularly with districts and government subsidies. Robustness checks confirm the reliability of these findings. CA enhances wheat productivity primarily through input supply systems, with secondary gains from market integration and knowledge diffusion. Policies should prioritize access to input, market integration, knowledge dissemination, and district-specific interventions to maximize TFP gains in cluster agriculture.</p>