"background": "Transport maintenance depots are critical infrastructure for ensuring fleet availability and operational efficiency. In the South African context, systemic inefficiencies within these depots contribute to suboptimal asset yield and increased operational costs, yet robust methodological evaluations of optimisation interventions are scarce. ", "purpose and objectives": "This study aimed to develop and apply a quasi-experimental design to rigorously evaluate the impact of a depot system optimisation programme on yield improvement, measured as the ratio of available vehicle-days to total scheduled vehicle-days. ", "methodology": "A difference-in-differences framework was employed, comparing yield metrics between a treatment group of six depots implementing the optimisation programme and a matched control group of six depots over a comparable period. The core statistical model was Y{it = \0 + \1 + \2 + \3 (\) +, with inference based on cluster-robust standard errors. ", "findings": "The optimisation programme resulted in a statistically significant average treatment effect on the treated (ATT) of 8. 7 percentage points (95% CI: 5. 2, 12. 2) improvement in yield. The primary driver was a reduction in mean workshop turnaround time by approximately 1. 8 days per major service. ", "conclusion": "The quasi-experimental design provided a robust counterfactual analysis, confirming that targeted system optimisation in maintenance depots can substantially improve fleet yield. The methodological approach offers a template for evidence-based engineering management. ", "recommendations": "Transport authorities should adopt similar rigorous evaluation frameworks for infrastructure interventions. Depot managers should prioritise workflow and inventory management reforms, as these were identified as key leverage points within the optimisation programme. ", "key words": "quasi-experimental design, maintenance optimisation, transport infrastructure, yield improvement, difference-in-differences, engineering management",
Botha et al. (Mon,) studied this question.