Effectiveness and Cost‐Effectiveness of Ecosystem‐Based Disaster Risk Reduction Interventions in Low‐ and Middle‐Income Countries: A Rapid Systematic Review

Background: Climate change and widespread environmental degradation have increased the risk of natural hazards in recent decades. Hydrological, meteorological, and climatological disasters have become more frequent globally (Parry et al. 2007; Cavallo and Noy 2009; CRED 2022), affecting over 3.9 billion people since 2000 and causing losses totalling USD 2.2 trillion (CRED 2020). However, the greatest consequences are felt in low and middle income countries (LMICs) where nearly 90% of deaths due to natural hazards between 2000 and 2018 (World Meteorological Organization 2021). Objectives: This rapid systematic review aimed to assess and synthesise the evidence on the effects and cost‐effectiveness of ecosystem disaster risk reduction (Eco‐DRR) interventions in preventing and mitigating hazards and natural disasters, and consequences for natural capital and human development outcomes. Search Methods: We initially screened 529 articles from a prior review (Sudmeier‐Rieux et al. 2021) on Eco‐DRR interventions using a pre‐validated tool in EPPI Reviewer to apply inclusion and exclusion criteria. We also identified project‐level evaluations from organisations such as OECD DEReC, UNDP ERC, and the UNDRR. The search was further enhanced using machine learning in Open Alex to create a citation network, using the studies identified in the initial steps as a training dataset. Selection Criteria: This review includes empirical primary studies measuring the impacts and cost‐effectiveness of Eco‐DRR interventions. We searched for and included studies that evaluated primary outcomes, including hazard prevention, hazard mitigation and natural capital stocks; we also included secondary human development outcomes if reported in these papers. Eligible designs included ex‐post impact evaluations using randomised assignment (randomised controlled trials [RCTs]), quasi‐experimental designs (QEDs) and qualitative impact evaluation designs, and ex‐ante impact evaluations using statistical modelling approaches. Economic and financial evaluations (cost‐effectiveness, cost‐utility and cost‐benefit analyses [CBAs]) were included to measure cost‐ effectiveness. Data Collection and Analysis: Two reviewers were assigned to screen the title and abstract for each study. Potentially relevant studies were then assessed at full text by two independent coders, following which data collection for the included studies was double‐coded. A third reviewer resolved all disagreements. All included studies were assessed using a pre‐validated critical appraisal tool based on the, ‘weakest link’ principle. Findings were synthesised by grouping outcomes into hazard prevention and mitigation, natural capital, and human development outcomes. Results: The review synthesised evidence from 58 studies, of which 28 were effectiveness studies, 20 were economic evaluations, and 10 were modelling studies. The effectiveness studies were all in green infrastructure nearly all concerned forestry and natural land use interventions, primarily protected areas although one study also evaluated geophysical hazards (public investment and early warning about urban landslides). Most evaluations were of interventions in Latin America and the Caribbean (21 studies), although there was a small number in Sub‐Saharan Africa (5 studies) and East Asia and the Pacific (16 studies); five evaluations were done at the global (LMICs) level. The effectiveness studies were assessed as being of moderate confidence (6 studies) or low confidence (21 studies). The synthesis of outcomes, using meta‐analysis of effect sizes (g), found beneficial effects of Eco‐DRR interventions on hazard incidence and exposure and natural capital outcomes. These included large effects on the reduction of forest fire (g = 0.32; 95% confidence interval (CI) = 0.23, 0.48; evidence from 4 studies), and medium‐ and small‐sized beneficial effects on forest cover (g = 0.12; 95% CI = 0.07, 0.18; 16 studies) and vegetation cover (g = 0.06; 95% CI = 0.01, 0.11; 4 studies). The synthesis of secondary human development outcomes was more limited because only a small subset (3) of papers that primarily measured hazard exposure and natural capital also reported these outcomes, which included loss of life, property damage, income, expenditure, and agricultural revenues. There were 10 modelling papers, of which half were for countries in Latin America and the Caribbean, 4 from East Asia, and 1 for South Asia. Five of the papers were concerned exclusively with flooding, two concerned storm surges (and thus also with flooding), and two more included flooding among the hazards assessed. One paper concerned landslides. Each paper used a different model and reported different outcomes. All papers used a range of different data sources, including satellite data, existing meteorological and other data sources, and primary data collection by the study team. None of the papers reported economic analysis. Two of the 10 papers were rated as high confidence in study findings, 5 at moderate confidence, and 3 at low confidence. Twenty economic evaluations assessing the cost‐effectiveness of Eco‐DRR measures were included. All the included papers were CBAs, with two papers accounting for income differences and incorporating equity weights to estimate social welfare benefits. While direct costs for structural measures were widely included, indirect costs and benefits were rarely reported. These indirect costs include the loss of productivity due to evacuations, repairs, and disruptions to economic activity. Similarly, the benefits of non‐structural measures, such as reduced flood damage and avoided societal disruptions, are often difficult to quantify and not reported. Five of the 20 papers were rated as having high confidence in the study findings, 8 were at moderate confidence and the 7 were at low confidence. Most results support the economic effectiveness of Eco‐DRR interventions. BCRs of 4 and above were regularly reported. For example, an extremely high BCR of 1800 was reported for drought risk reduction measures in Sudan for irrigation, early warning systems demonstrating exceptionally high potential financial returns (IRR of 409%). Additionally, the ecological prioritisation of engineering measures was lower compared to EbA measures, as they were primarily focused on addressing water scarcity without the same level of positive ecological impacts. Conclusion: The evidence base on the effectiveness and cost‐effectiveness of Eco‐DRR is compartmentalised. Most studies of the effectiveness of implemented actions have evaluated green infrastructure interventions, particularly protected areas, whereas most of the modelling and cost‐effectiveness studies evaluated blue and hybrid infrastructure technologies. Green infrastructure interventions, such as protected areas, have generally shown large effects on increasing natural capital stocks, especially in forestry reserves, and on reducing the incidence of or exposure to hazards like fire, although fewer studies have measured the latter. In modelling studies, most of the papers concluded that the intervention being assessed should be adopted. But in the absence of economic analysis, such a conclusion cannot be drawn from the demonstrated impact alone. Economic evaluations generally supported the benefits exceeding costs (BCRs > 1) of Eco‐DRR, particularly in reducing flooding, especially structural hybrid interventions, but also capacity building, early warning systems, and mangrove restoration. However, these studies have significant limitations regarding information about distributional effects, non‐monetised values, and indirect costs and benefits. Interventions with the highest BCRs included early warning systems for floods and non‐structural interventions, such as land use planning and capacity building. Future economic evaluations should be more comprehensive, considering direct and indirect costs and benefits. Improvements in understanding different approaches to Eco‐DRR and the relative role of directly targeted versus systemic approaches to risk management are essential. We found very limited research on environmental hazards and geological and geophysical hazards, such as landslides and earthquakes, suggesting that these topics require investigation in LMICs in future primary studies of whatever type (both effectiveness and cost‐effectiveness). More studies that evaluate topics relating to equity, such as the effects on the social welfare of vulnerable populations, are needed. Studies of the effectiveness in these areas might usefully draw on established methods of qualitative impact evaluation. (Summary of findings tables: S1).