<![CDATA[Achieving SDG 8, Decent Work and Economic Growth, in the agricultural sector requires competent and adaptive human resources, particularly in agricultural extension institutions that directly support farmer empowerment, productivity improvement, and local economic development. However, empirical evidence explaining how sustainable skills training contributes to SDG achievement through human resource competence remains limited, especially in developing-country extension contexts. This study examines the influence of Sustainable Skills Training on SDG 8 Achievement through Human Resource Competency Improvement among agricultural extension workers at Agricultural Extension Centers (BPP) in Indragiri Hilir Regency, Indonesia. A quantitative approach using Structural Equation Modeling - Partial Least Squares (SEM-PLS) was applied to data collected from 104 agricultural extension workers through saturated sampling. The measurement model showed acceptable internal consistency and convergent validity, with Cronbach’s Alpha values above 0.85, Composite Reliability values above 0.90, and AVE values above 0.73. The structural model demonstrated strong explanatory power, with R² = 0.893 for SDG Achievement and R² = 0.837 for HR Competence Improvement. Predictive relevance was indicated by a manually calculated Q-square value of 0.983, while the SRMR value of 0.100 indicated marginal model fit. The findings show that Sustainable Skills Training strongly contributes to HR Competence Improvement, and HR competence functions as the main mechanism linking training to SDG 8 Achievement. However, the path coefficient for HR Competence Improvement → SDG Achievement exceeded the conventional standardized range, indicating the need for cautious interpretation due to potential construct overlap and multicollinearity. The novelty of this study lies in demonstrating that HR competence serves as a central pathway through which sustainable skills training supports SDG 8 achievement in local agricultural extension institutions. Future studies should refine the measurement model, validate predictive relevance using Blindfolding or PLSpredict, and test alternative model specifications to improve construct discriminability.]]>
