Answers are grounded in this paper and cite their sources.
Pakistan's 2022 monsoon floods caused unprecedented losses, and renewed flooding in 2025 struck while recovery was still underway; little is known about recovery inequality under repeated flood exposure. Hence, we assessed flood impacts and recovery inequality across Pakistan by combining grid-level propensity score matching (PSM) with satellite indicators of socioeconomic activity (VIIRS nighttime lights, NTL), vegetation (MODIS NDVI), and population exposure. From 85,533 cells on the Web-Mercator level-14 grid (≈2.1-km ground spacing; mean cell area ≈4.4 km 2 ), 15,794 heavily flooded cells (flood fraction > 20%) were matched to non-flooded controls, reducing mean absolute standardised differences from 0.76 to 0.09; a stricter calliper–Mahalanobis specification balanced every covariate below |SMD| = 0.10. In the shock year, flooded cells showed a modest NTL-growth decline (matched difference [MD] = −0.83 percentage points [pp], p = 0.02) that attenuated to non-significance under the stricter specification, whereas vegetation loss was severe (+29.0 pp). NTL trajectories exceeded those of matched controls in 2023–2024 (+2.9 and +4.0 pp), but weakened in 2025: event-specific estimates indicated a 2.1–2.9 pp reduction in 2024→2025 NTL growth, depending on the control definition, alongside a further +10.9 pp vegetation loss. Recovery was strongest in Punjab and weakest in Sindh. Linking outcomes to the Relative Wealth Index, poorer cells lost more and recovered more slowly (Spearman ρ = 0.167 in the shock year; 0.06–0.08 thereafter). As pre-flood trends are non-parallel, we interpreted all estimates as covariate-adjusted MDs rather than causal effects — a replicable framework for monitoring disaster recovery inequality in data-scarce regions.
No takes yet. Share an insight, caveat, or question.
Sughra et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: