Policy Design for Cash-for-Repair Housing Recovery under High Inflation: Supervision Intensity, Safety Outcomes, and Equity Trade-offs

Yemen’s protracted conflict has devastated urban housing while intensifying socio-economic vulnerability, making reconstruction both a humanitarian imperative and a risk-governance challenge. Focusing on urban Taiz, this study evaluates housing reconstruction modalities in terms of beneficiary satisfaction, perceived structural quality, and equity under severe inflation and market volatility. Using a convergent mixed-methods design, we combine (i) a household survey of 320 assisted households, (ii) 20 semi-structured interviews with institutional and technical stakeholders, and (iii) a transparent scenario-based simulation model that explores coverage–safety trade-offs under a fixed budget and inflation shocks. Survey results indicate that owner-driven cash assistance dominates the intervention landscape and is associated with high satisfaction yet does not reliably align with perceived structural safety in this setting. Given the extreme scarcity of non-ODR cases in the survey, regression estimates for modality effects are treated as exploratory signals and interpreted primarily through triangulation with qualitative mechanisms and simulation evidence. Across strands, engineering supervision emerges as the critical socio-technical mechanism that converts financial inputs into safer reconstruction outcomes, while inconsistent targeting and grant scales risk undermining perceived fairness and social cohesion. The simulation illustrates that low-grant, weakly supervised approaches maximise coverage but can fail to achieve minimum safety thresholds, whereas a medium-grant, strongly supervised hybrid configuration offers a plausible compromise between breadth and depth under high inflation. The study contributes a supervision-centred, simulation-informed policy framing to support donors and coordination actors in balancing agency, safety, and equity in resource-constrained urban recovery.