Back to cover

Reducing Disaster Risks from Natural Hazards

Chapter 4 | Effectiveness of the World Bank’s Disaster Risk Reduction Activities

Most disaster risk reduction (DRR) operations are not providing sufficient information to understand the level of DRR being achieved, which inhibits an understanding of DRR contributions to development impacts. Most resilient infrastructure investments lack information on resilience standards; many development policy operations lack evidence on the results of policy changes.

The World Bank is increasingly identifying and addressing the needs of some groups that are disproportionately impacted by disasters; however, for other groups, there is slow progress and limited reporting on DRR benefits.

DRR investment projects often effectively support infrastructure construction, but they do not explicitly address operations and maintenance that are required for long-term resilience. This shortcoming is more evident in core disaster projects mapped to the Urban, Disaster Risk Management, Resilience, and Land Global Practice than in sectoral infrastructure projects.

The World Bank has been more effective in developing early-
warning systems infrastructure than in delivering early-warning system services (for example, forecasting and community-
preparedness activities).

Disaster insurance activities have had a limited impact because insurance programs have had difficulty in reaching scale. However, these activities have made progress in awareness raising, capacity building, and product development.

DRR policy reforms have achieved about three-quarters of their indicators, but just more than one-third of indicators capture evidence on downstream effectiveness. Development policy operations have provided valuable disaster-contingent financing.

The World Bank has been highly effective on DRR when it has had sustained engagement using multiple instruments and interventions and when it has deliberately engaged to achieve replication by others.

This chapter assesses the extent to which the World Bank articulates and measures disaster risk–related outputs and outcomes, as well as the results and factors of effectiveness for key DRR approaches. It does this first by assessing the articulation of intended DRR results and how these results are measured in the DRR lending portfolio. It also includes an analysis of DRR project beneficiaries, with a particular focus on how projects identified, addressed, and tracked benefits for disaster-vulnerable groups. Second, this chapter assesses results and generates lessons on factors of effectiveness for four key approaches in the DRR portfolio: disaster-resilient infrastructure, EWSs, disaster insurance, and DRR policy reforms. Third, this chapter presents findings from an analysis of cases in which World Bank DRR activities had highly successful results.

Identifying and Measuring Disaster Risk Reduction Results and Outcomes

Most DRR operations are not providing sufficient information to understand the level of DRR being achieved (for example, reduced exposure and vulnerability), which inhibits an understanding of DRR contributions to development impacts (reduced economic loss and mortality). Although most DRR IPFs have outcome-oriented objectives, such as to build resilience or reduce exposure or vulnerability to disasters, many of these projects do not provide sufficient information to determine whether these goals are being met. Although 89 percent of all IPFs have some DRR indicators in their project documentation, for 61 percent of these, indicators are often articulated only at an output level. For example, for projects that seek to achieve DRR through resilient infrastructure, the most frequently occurring indicator is the length of infrastructure or protective works constructed. Outcome indicators for IPF that support resilient infrastructure could include, for example, the standard to which the infrastructure was built and the contribution of that infrastructure to reduced exposure and vulnerability. For DPOs, only 35 percent of operations had at least one outcome indicator. Box 4.1 includes examples of relevant outcome indicators to assess World Bank IPFs and DPOs’ contributions to development impacts.

While the share of projects that include DRR outcome indicators is rising, so is the share of projects that have no DRR indicators, which is associated with the increased number of sectoral operations that include mainstreamed DRR activities. Figure 4.1 shows that 28 percent of the DRR projects approved in the second half of the evaluation period included DRR-related outcome indicators, as compared with the 24 percent of DRR projects approved in the first half. From the first half of the evaluation period to the second half, the share of projects with DRR activities but without any indicators of DRR results rose from 16 to 24 percent. This decline in the inclusion of DRR indicators was mainly found in mainstreamed sector projects whose objectives often included aims to enhance the climate resilience of infrastructure or to build resilience through sustainable land and water activities. Some core projects that support post-disaster emergency operations also lacked DRR indicators.1

Box 4.1. Good Practice Outcome Indicators for Disaster Risk Reduction in World Bank IPFs and DPOs

For investment projects that seek to build resilience to flood risks, informative disaster risk reduction indicators include the following: the number of people protected from a flood of a certain recurrence interval due to the infrastructure having been built to a resilient standard (as in the Dar es Salaam Metropolitan Development Project); an estimated increase of per capita farmer income or a reduction in economic losses occurring from floods due to resilient infrastructure (as in the Huai River Basin Flood Management and Drainage Improvement project); or a decrease in the number of days of interrupted traffic due to flooding (as in the Jamaica Disaster Vulnerability Reduction Project). For investment projects that address drought risks, such as the Shire River Basin Management Program or the Strengthening Climate Resilience Project in Zambia, good indicators included an estimated change in risk levels of disaster-vulnerable households (due to improved water management) and the increased level of household incomes derived from diversified and disaster-sensitive sources. Other good outcome metrics include decreased time for the restoration and resumption of use of assets and services (for example, restoration of economic activities, access to health care and schools). For development policy financing operations, useful outcome indicators included improvements in lead time for flood operations (Mozambique), the percentage of smallholder agricultural land covered by disaster insurance (Mexico), and an increase in the share of new buildings that comply with building standards (Grenada).

Source: Independent Evaluation Group.
Note: DPO = development policy operation; IPF = investment project financing.

Figure 4.1. Outcome Indicators for Disaster Risk Reduction Investment Projects (FY10–20)


Figure 4.1. Outcome Indicators for Disaster Risk Reduction Investment Projects (FY10–20)

Source: Independent Evaluation Group.

Note: DRR = disaster risk reduction; FY = fiscal year.

DRR results reporting significantly varies by hazard: the reduction of flood risk is more likely to be measured than the reduction of drought risk. Most projects that aim to reduce the risks of flood and storm (84 percent of 407 flood projects) articulate and include indicators that measure some level of DRR result. However, almost one-third of projects that address drought risks (31 percent of 174 projects) do not measure drought-related results (figure 4.1). These projects are often multihazard projects that include metrics for flood-related events but not for drought. Outcome-level reporting for seismic risk is also rare, as it is challenging to measure or predict a reduction in risk related to earthquake impact. As compared with other hazards, earthquakes are infrequent, even in areas where there is significant risk. Only 9 percent of the DRR portfolio that addressed seismic risks (8 projects) used outcome indicators in their disaster risk–related reporting (figure 4.1). Projects in St. Vincent and the Grenadines and Grenada are good examples, as they included estimates for the number of beneficiaries with reduced risks due to the enhanced resilience of public buildings.

Targeting and Tracking Disaster-Vulnerable Groups

People living in poverty, who are more likely to live and work in areas with high disaster exposure, face high risks from disasters. Efforts to mitigate poverty and disaster risks are complementary, as poor populations are highly affected by disasters (Hallegatte et al. 2017). Disasters can lock already poor individuals into poverty traps because their assets are too minimal to recover even in the long term (Hallegatte, Rentschler, and Walsh 2018).

There are populations that are disproportionately negatively affected by disasters because they are both highly exposed to disaster risk and extremely vulnerable (most susceptible to disaster impacts). The World Bank is producing a body of analytics that considers the needs of disaster-vulnerable persons (Erman et al. 2021; Krylova, Sirker, and Haile 2021; Williams 2020; World Bank 2022b, 2022c). To determine how DRR projects have identified, addressed, and tracked benefits for groups that disproportionately suffer from disasters, the evaluation conducted a content analysis of 135 closed projects and compared these with the content of 82 projects approved in FY20 (see appendix A). The most-referenced disaster-vulnerable groups for which this evaluation could assess trends include women and girls, children and youth, persons with disabilities, and the elderly (other disadvantaged groups may include ethnic minorities or migrants, for example). The disaster-related vulnerabilities of these groups are described in box 4.2.

The World Bank is increasingly identifying and addressing the needs of some groups disproportionately impacted by disasters; however, for other groups, there is slow progress and limited reporting on DRR benefits. Findings for the most frequently referenced disaster-vulnerable groups are discussed in the following sections.

Box 4.2. Marginalized and Disadvantaged Groups at Risk of Suffering the Most from Disasters

Women and girls. Disasters disproportionately impact the life expectancy of women. A study of 141 countries found that disasters lower women’s life expectancy more than men’s due to higher morbidity and more severe economic impacts, including higher rates of unemployment. Female-headed households are more exposed—for example, in Bangladesh, there are three times as many female-headed households in flood-prone zones as in noneroded ones. The risk of gender-based violence and child marriage is also prevalent after disasters.

Children and youth. Children compose one-third of the global population and one-half of the extremely impoverished population. Disasters affect household welfare in ways that reduce children’s access to nutritious foods and health services, leading to permanent stunted growth of children, and their access to education, contributing to declining enrollment rates and an increase in the number of dropouts. In Africa, enrollment rates declined 20 percent in drought-affected regions, with similar post-disaster impacts reported elsewhere.

Persons with disabilities. More than 1 billion people live with disabilities, 80 percent of whom reside in developing countries. The United Nations estimates that 20 percent of the world’s poorest populations have a disability. Limited mobility, discrimination, and other barriers increase vulnerabilities during a disaster. For instance, the fatality rate for persons with a disability in Japan after the 2011 Tohoku earthquake and tsunami was four times higher than that of the general population.

Elderly individuals. Studies indicate that by 2050, one in five people in developing nations will be more than 60 years old. The elderly are particularly vulnerable to hazards due to lack of mobility, preexisting health issues, nutritional needs not considered during emergencies, and isolation from families and social services.

Sources: Arnold et al. 2018; De Silva and Burton 2008; Erman et al. 2021; Hallegatte, Rentschler, and Walsh 2018; Shetty 2012; UN 2021; World Bank 2020a.

Although women and girls are increasingly being integrated into the design and tracking of DRR activities, significant gaps remain in integrating them into DRR planning processes, particularly at local levels. Sixty-three percent of FY20 projects versus 21 percent of closed DRR projects integrate gender considerations in DRR activities and increasingly measure gender-related DRR outcomes, including by focusing on women’s agency in decision-making. Sixty-five percent of FY20 projects versus 29 percent of closed DRR projects integrated indicators to track gender-related DRR benefits.

The World Bank is increasingly identifying and addressing the vulnerability of children and youth in DRR contexts but often not tracking these disaggregated results. Twenty-eight percent of FY20 projects versus 10 percent of closed DRR projects integrate youth considerations in DRR activities. However, only 39 percent of the FY20 projects and 43 percent of the closed DRR projects that integrate youth tracked their results, except for safer school and emergency shelter activities (where awareness was increased, exposure was decreased, or education services were resumed).

There is very slow progress in integrating the needs of persons with disabilities and the elderly into DRR activities, and no DRR disaggregated results have been tracked. Eighteen percent of projects approved in FY20 versus 4 percent of all closed DRR projects integrate considerations regarding persons with disabilities in DRR activities; the elderly are integrated in 12 percent of FY20 projects and 7 percent of closed DRR projects. However, only one project tracked results for persons with disabilities, and none tracked results for the elderly. Projects supporting persons with disabilities and the elderly included accessibility standards, mobility, and building-use considerations, but few integrated these groups into DRR decision-making (except community-based DRM committees in Vietnam and Indonesia). The Istanbul Seismic Risk Mitigation Project developed guidance titled “First 72 Hours for the Individual and Family in an Earthquake” and effectively used accessibility standards in the construction of public facilities. A noteworthy trend is that projects approved in IDA countries in FY20 include these groups, whereas for closed projects, these groups were only cited in IBRD projects.

Results of Key Approaches and Explanatory Factors of Effectiveness

This evaluation reviewed results achieved and factors of effectiveness for four key activities. It was infeasible for the evaluation to carry out a systematic review of results given the large and heteronomous nature of the DRR portfolio. Instead, the evaluation reviewed key activities based on portfolio size and consultations with World Bank management. These key approaches included (i) disaster-resilient infrastructure, (ii) EWSs, (iii) disaster insurance, and (iv) DRR policy actions. For each activity, the evaluation reviewed completion reports for all closed projects that featured this activity and identified results achieved or not achieved, as well as factors of effectiveness.

Disaster-Resilient Infrastructure

The World Bank has successfully built or strengthened disaster-resilient infrastructure or protective works in two-thirds of projects that had this aim, whereas significant shortfalls in other projects were mainly associated with engineering capacity, procurement issues, and delays due to working in areas at high risk for disaster. There were 60 closed IPFs with disaster-related objectives that included construction or rehabilitation of resilient infrastructure, including irrigation and drainage, roads, schools, and hospitals or protective works. These projects included 174 indicators relating to resilient infrastructure. Two-thirds of these projects fully or mostly achieved their infrastructure targets. Another quarter failed to deliver planned infrastructure due to weak engineering designs, procurement issues, and delays caused by hazards. For example, in India, a coastal protection project achieved only half of its planned embankments due to delays caused by storms. The project had difficulties in obtaining bids for small civil work projects in high-cost remote areas, and it underestimated the time needed to obtain environmental permits for work. A project in Mozambique rehabilitated only half of the planned dikes and levees because of cost overruns and procurement delays. In Haiti, less than half of the planned multihazard-resilient evacuation shelters were reconstructed due to insufficient coordination unit capacity, delays in procurement, insufficient resources for resettlement compensation, and financing delays from high turnover in ministers.

Enhancing design standards is critical for achieving resilient infrastructure, but project documents do not provide sufficient information on the standards applied to validate probable DRR results. Design standards that contemplate an appropriate level of resilience are important in both the construction and rehabilitation of infrastructure and assets. Incorporating the evolution of hazards over time due to climate change, deforestation, and urbanization is also essential. If projects use a probabilistic risk estimation, they can calculate the cost implications of different levels of resilience, enabling risk-informed decisions on standards. In the absence of an estimation, asset owners may be taking on hidden contingent liabilities, which will demand future investments. In most of the projects reviewed (92 percent), the introduction of design standards was used to strengthen the resilience of infrastructure assets. However, explicit references to standards used for resilient infrastructure (for example, that infrastructure is resilient to a 1-in-50-year flood rather than a 1-in-20-year flood) were included in only 35 percent of the completion reports for these projects. Other projects referred to a general intent to build resilience or build back better, or they provided anecdotal evidence that the strengthened resilience was effective, normally when a disaster occurred before project completion. Furthermore, only 7 projects out of 60 indicated that the resilience standards proposed were informed by a fully probabilistic risk estimation—one that not only informs the resilience standards adopted but also can be used to calculate future expected losses using conventional financial risk metrics, such as average annual loss and probable maximum loss. None of the projects indicated whether climate change was integrated into probabilistic risk metrics. Twenty-two projects did not refer to risk identification or estimation at all.2

Efforts to identify and address territorial or system-level risks amplified the effectiveness of infrastructure activities to achieve wider resilience aims. Twenty projects out of 60 (primarily in the Water and ENB GPs) included activities that addressed underlying risk drivers, such as strategies for water rights management or forest management or inadequately planned urban development. For example, in Xinjiang, China, a strategy for water rights management complemented water storage infrastructure improvements to address drivers of environmental degradation and falling water tables and reduce drought risk. In Dakar, Senegal, a flood reduction project included both stormwater drainage infrastructure and the integration of flood risk consideration into urban planning, including solid waste management.

Most resilient infrastructure projects did not explicitly address deficiencies in O&M, which is likely to undermine the long-term resilience of built or rehabilitated infrastructure. A justification for many resilient infrastructure projects is that infrastructure has deteriorated over time due to deficient O&M, to the point where a new project is required to rehabilitate the infrastructure. Moreover, for resilient infrastructure, this deterioration is often associated with exposure to hazards. Unless budgets and arrangements for O&M are adequate, a cycle of build-neglect-repair/replace is likely to continue, which undermines resilience. Out of 60 projects, 16 directly addressed O&M deficiencies through measures such as the creation of new budget lines, institutional reforms, or design of new revenue or pricing mechanisms for infrastructure services. Of these, only 1 project at closure raised concerns about the sustainability of O&M. In contrast, 19 of the 30 projects that did not directly address O&M deficiencies raised concerns about effectiveness and sustainability of O&M at closure (in 4 projects, documentation was insufficient to draw a conclusion). These concerns included the weak institutional capacity of agencies responsible for planning and implementing O&M, the absence or insufficiency of existing budget lines for O&M, a lack of clarity over institutional responsibilities (leading to weak accountability), and so on. Although in some projects funds were provided for O&M during the project, Independent Evaluation Group validations often raise concerns that this is unlikely to solve underlying deterioration. Efforts to address O&M were more common in sectoral projects such as those mapped to the Water or Transport GPs, while only 2 of the 29 resilient infrastructure projects mapped to the Urban, Disaster Risk Management, Resilience, and Land GP directly addressed O&M deficiencies. This may be because it is more difficult to address sectoral infrastructure issues such as O&M in multisector disaster operations, and sector specialists prioritize these issues to a greater degree than disaster specialists.

Community engagement in the development of local resilient infrastructure systems is associated with the systems’ effective construction or rehabilitation and likelihood of sustainability. Local infrastructure systems provide essential services to individuals, households, communities, and businesses and can include water, drainage, sanitation, local road, health, and education facilities. There are 26 closed and evaluated DRR projects that used community approaches to help build resilient local infrastructure systems. All 26 projects met their infrastructure targets; evidence attributes results to the enhanced relevance of the infrastructure and ownership by communities for their O&M. Two-thirds of the projects went deeper by engaging communities in consultations on DRR activities and delegating decision-making responsibility to community organizations. Engaging community organizations to discuss the most appropriate approaches and solutions to local infrastructure systems was an explicit component of these projects and was undertaken at the planning stage. These projects addressed the sustainability of local infrastructure systems (including ecosystem services) on which local livelihoods and welfare depend. Community ownership of local infrastructure systems and services strengthened sustainability and facilitated longer-term O&M of the infrastructure assets, a necessary factor for achieving sustained disaster resilience.

Early-Warning Systems

The World Bank often supports the development or strengthening of EWSs to help reduce disaster risks in advance of hazard events. EWSs are defined as “integrated systems of hazard monitoring, forecasting and prediction, disaster risk assessment, communication and preparedness activities, systems, and processes that enable individuals, communities, governments, businesses, and others to take timely action to reduce disaster risks in advance of hazardous events” (UNDRR 2016). The World Bank has helped develop or strengthen EWSs in 150 projects across 69 countries since FY10, of which 35 projects (32 IPF and 3 DPF) are closed and evaluated and are the subject of this assessment. (There were also 47 nonlending projects covering EWSs outside the scope of this effectiveness assessment.)

Effective EWSs require support for four interlinked elements. These elements are derived from a World Meteorological Organization checklist (WMO 2018). First, EWSs require the collection and curation of disaster risk information—“comprehensive data collection, mapping and analyses of all dimensions of disaster risk, including hazards, exposure, vulnerability, and capacity, related to persons, communities, organizations, and countries and their assets.” Second, they require equipment, technologies, and capacity building for multihazard detection, monitoring, analysis, and forecasting of the hazards and potential consequences. Third, they need support for communication and dissemination to ensure people receive timely warnings and to facilitate coordination and information exchange. Fourth, they require support for disaster preparedness, public awareness, and response capabilities—institutions and people that are enabled to act early and respond to warnings through risk education. The following sections elaborate on the effectiveness of World Bank support in achieving these interlinked elements.

Disaster Risk Information

The World Bank’s support for hazard mapping, a critical component of disaster risk information systems, has enabled clients to map the probable location and intensity of disasters but often not the vulnerability levels of affected populations or their capacity to respond. In the EWS portfolio, hazard mapping included the development of hydrometeorological models, flood risk and landslide susceptibility maps, cyclone hazard risk atlases, digital elevation maps, aerial photography, and geographic information systems. In all but three EWS projects, support for hazard mapping successfully helped clients (for example, hydrometeorological and DRM agencies) identify where hazards are likely to occur, as well as their likely intensity and frequency. However, the same projects reported much less on how they assessed the potential vulnerability of affected groups. Two examples where vulnerability was assessed were in Rio de Janeiro and in Togo. In Rio de Janeiro, the World Bank supported hazard mapping in 15 municipalities with landslide risk, resulting in the identification of 16,000 inhabitants living in high-risk areas. In Togo, the World Bank helped update an urban topography map, establish a database on settlements, and assess the population’s vulnerability to floods.

Detection, Monitoring, Analysis, and Forecasting

Although EWS infrastructure has been put in place, intended forecasting capabilities were only successful in half of all projects with this aim. The World Bank built or repaired hazard detection, monitoring, analysis, and forecasting infrastructure and technologies in three-quarters of closed EWS projects. EWS forecasting infrastructure includes meteorological, hydrological, and seismic monitoring stations; weather stations; and radar systems. All but two projects achieved their infrastructure aims. However, efforts to enhance the accuracy, availability, and timeliness of forecasting through technological innovation and capacity building were successful only in half of these projects. For example, in Mozambique, hydrometeorological monitoring stations were transmitting hourly data for forecasting at project close. However, projects in Brazil, Peru, Chad, and Haiti failed to meet their forecasting goals. In Brazil, data collection stations for precipitation and water levels were off-line at project close, and while equipment was acquired, software to process EWS information was not installed. In Peru, none of the EWS subprojects were completed due to cumbersome review and clearance procedures required by the government. In Chad, network issues prevented the transmission of data from the stations. In Haiti, the hydrometeorological system was not operational at project close. In Madagascar and Somalia, where the World Bank partnered with the United Nations and humanitarian agencies, there was a lack of reporting on forecasting capabilities.

A World Bank stocktaking of southern African EWSs shows how a lack of sustainable funding and, relatedly, the lack of maintenance of observation infrastructure undermine gains made in forecasting capacity. A stocktaking of EWSs in 16 southern African countries found that significant portions of their observation infrastructure, including sensors, radars, and weather stations, were no longer operational. According to the agencies interviewed, this is due to a lack of sustainable funding for transport, spare parts, and staff to repair and calibrate the infrastructure. Rainfall radar, for example, was operational in only 3 of the 16 countries; many other types of infrastructure had fallen into disrepair (World Bank 2021f).

Although EWSs work best when they include impact-based forecasting, only half of the projects articulate this intent, with most achieving this aim. Impact-based forecasting targets specific vulnerable groups, generating targeted alerts associated with required actions (for example, making disaster-related decisions or enabling evacuation). Of the 35 closed EWS projects, 17 supported impact-based forecasting, and 14 achieved this aim. In India, Mozambique, Moldova, and São Tomé and Príncipe, impact-based forecasting helped fishermen decide whether to go to sea. In India, impact-based forecasting activities in a cyclone risk mitigation project enabled a local government to suspend fishing activities and evacuate 200,000 fishermen ahead of the landfall of Cyclone Fani. In Moldova, Nepal, and Vietnam, farmers were provided agrometeorological information and weather forecasts to enable decisions about planting; in Nepal, this resulted in improved productivity and reduced crop losses during floods. In Malawi, Togo, and China, impact-based forecasts helped authorities change the water discharge volume of dams and reservoirs ahead of extreme rainfall events to reduce flood impacts.

A key element of successful impact-based forecasting is the targeting that is enabled through participatory design. Engaging directly with communities enables experts to identify the types of decisions that can be informed by forecasting, the information needed to make those decisions, and the communication style to use, adapted to the capacity of communities (Baudoin et al. 2016; Sufri et al. 2020). Participatory approaches were often key elements of success in implementing forecasting. They enabled continuous improvement of design and trust-building—a critical factor in whether communities choose to heed a warning. In São Tomé and Príncipe, the project team engaged with fishing communities to understand how they make decisions about when to go to sea. Having learned that the fishermen placed trust in traditional forms of weather detection and community information exchange, the project enhanced the utility of its forecasting and alert system. Furthermore, the trust established through participatory approaches enabled the project team to have effective discussions with communities on relocation.

Almost one-quarter of the closed portfolio achieved expanded lead time through enhanced forecast capacity, enabling communities to better prepare and respond to imminent hazard risks. The expanded lead time for severe weather warnings refers to the increased amount of time that communities are given to prepare for and respond to sudden onset hazards. One-quarter of closed EWSs aimed to increase lead time or timeliness, and all but one reported positive outcomes. For example, in Moldova, the lead time for severe storm weather warnings was expanded from three to six hours. In Vanuatu, the increased lead time of seismic monitoring helped the government issue evacuation orders after the occurrence of a 7.6 magnitude earthquake in 2018.

Coordination, Communication, and Dissemination

The degree of coordination among relevant agencies was a common marker of success in EWS projects. Effective EWSs require timely communication and information exchange among agencies that manage meteorological, hydrological, climate, and disaster response. All but two EWS projects with interagency coordination achieved their data analysis and hazard forecasting aims. The Mekong Integrated Water Resources Management Project demonstrates that “it is essential to identify roles, responsibilities, and coordination mechanisms from the project outset. Without coordinated procedures across the agencies responsible for forecasting, hydrology, and risk communication, it is not possible to have a functional EWS” (World Bank 2020i). Likewise, the successful Moldova EWS project shows that “the multiple systems needed for hydrometeorological services delivery—including observation, data management, meteorology, hydrology, climatology, visualization, forecasting, and dissemination—must communicate effortlessly with each other” (World Bank 2017h).

Effective emergency communications in times of disaster require building the redundancy and robustness of telecommunication and electric power infrastructure. The need for redundancy in disaster communication and dissemination systems was a lesson of effectiveness in EWS projects in India, São Tomé and Príncipe, Togo, and Mozambique. In these projects, the World Bank supported multiple communication channels among populations that had uneven access to technologies. These included horns, radio warnings, cell broadcasts, or interactive voice-response technologies fused into a multichannel warning system.

Disaster Preparedness, Public Awareness, and Response Capabilities

Enhancing communities’ disaster preparedness is key to maximizing the effects of upstream DRR investments—and saving lives—but few projects demonstrate that such preparedness activities will be effective in the face of disaster. Investments in EWS communication systems, shelters, and evacuation routes do little to reduce vulnerability if communities cannot react to warnings promptly. Mock drills, for example, can increase awareness about safe evacuation procedures and save lives. Of the 20 closed EWS projects that sought to strengthen disaster preparedness and response capacity, 60 percent did not report on community-preparedness results and another 10 percent included claims not backed by evidence. Only three projects included verifiable evidence on the effectiveness of preparedness activities. In India, the training of 535 village DRM task forces and the conduct of mock drills facilitated timely evacuation and strengthened local search and rescue capacity. In Mozambique, a beneficiary survey found that daily forecasts and impact-based warnings changed behavior.

Sustainability of Early-Warning Systems

Only a few EWS projects—all of which exclusively support EWSs—incorporate the factors that enable the O&M of EWS. The O&M of networks of sensors, weather stations, alert equipment, and other technology is key to operating EWSs. Only 4 of the 35 closed projects—in India, São Tomé and Príncipe, and Togo—supported EWS maintenance. These were “stand-alone” projects; EWSs were the main aim. However, projects that included an EWS as a component did not focus on O&M. A World Bank stocktaking of EWSs in African countries found that a lack of sustainable financing is responsible for limited EWS service provision. In these countries, river gauge and hydrometeorological infrastructure is often in disrepair (World Bank 2021f). Power and connectivity are key issues, adversely affecting telemetry networks and communications among stations, central offices, and end users.

Disaster Insurance

Disaster insurance is an important part of the World Bank’s work on disaster risk finance. The World Bank’s support to clients on disaster risk finance includes mechanisms such as reserves, contingent credit, catastrophe bonds, disaster funds, adaptive social protection schemes, parametric insurance, insurance for public assets, and promotion of catastrophe risk insurance markets. This evaluation focuses on disaster insurance as a key activity because there is a body of evidence on results; many other aspects of disaster risk finance have been supported primarily through analytics and advisory services, which have not been evaluated, or are part of more recent projects that have not yet been completed.

Disaster insurance works to increase the preparedness of affected public or private entities by transferring disaster risk. Insurance schemes collect premiums and make payouts should a disaster occur and thus can improve preparedness and enhance resilience by reducing the financial impact of disasters. However, the potential of disaster insurance to achieve DRR impacts depends on the size and scale of its coverage: if the number of participants or share of assets covered is too low, then the payout will be too limited to meaningfully offset the negative impacts of disasters. The World Bank has approved 46 lending operations (23 IPF and 23 DPF) with disaster risk insurance in 30 countries, of which 20 are closed and have an Implementation Completion and Results Report, supporting 13 insurance programs. These operations have supported disaster and asset data collection, modeling and risk assessments, market research, adoption of authorizing laws or policies, development of insurance products, awareness raising, and the financing of premiums. Insurance is offered in different forms for sovereigns, businesses, or households for indemnity or index-based approaches, and it covers a range of hazards. There are also 40 disaster insurance nonlending activities outside the scope of this assessment.

Disaster insurance activities targeting businesses or households have had difficulty reaching scale and sometimes lack evidence on coverage. Of eight insurance schemes of this type supported by closed operations, only half have evidence of insurance uptake targets being achieved, and one was a small pilot. For the other four cases, schemes failed to achieve their targets or provided inadequate evidence. For example, in Kerala, India, an agricultural insurance program achieved two-thirds of its intended increase in the number of farmers covered. In Mongolia, a herder insurance scheme reached only 11 percent uptake despite years of awareness raising. Herders often insured only small parts of their herd to access subsidized loans, and the program was ill attuned to traditional risk mitigation strategies (World Bank 2020b). Several disaster insurance projects may report on awareness, access, or payouts but not on insurance coverage. In these cases, claims are made about the high level of insurance access enjoyed by affected populations because products have been developed where none existed before. In Sri Lanka, an operation supporting a national household disaster insurance scheme reported on total payouts but not penetration rates or the number of beneficiaries.

Insurance programs have struggled to achieve high penetration rates in part because clients have found it difficult to make insurance mandatory without public subsidy. World Bank efforts to go straight to a mandatory catastrophe insurance program have usually been unsuccessful; the World Bank has been more successful when it has built disaster insurance demand incrementally, including by focusing on submarkets to demonstrate success. Although mandatory insurance is desirable for its ability to maximize the size of the insurance pool and avoid adverse selection problems, it may not be achievable without significant prior experience with voluntary programs, and it is politically challenging to implement. In the case of the South East Europe and Caucasus Catastrophe Risk Insurance Facility, the governments of North Macedonia, Serbia, and Kazakhstan expressed strong political reservations against making catastrophe insurance compulsory. Consequently, penetration rates were much lower than planned: in North Macedonia, these rates were only 1 to 2 percent, compared with the planned 10 to 15 percent. Some projects have stepped back from efforts to support mandatory insurance, instead focusing on alternative market-based solutions. For example, in Kazakhstan, when the authorities postponed the implementation of a compulsory catastrophe insurance program, they launched insurance products in the agriculture sector and weather-risk market. The work featured extensive consultations with farmers to assess affordability and set premiums suitable to local conditions. Targeted coverage rates were achieved for agricultural insurance programs in Mexico and Colombia that covered low-income smallholders, but these schemes required tight targeting and public subsidies.

Projects supporting sovereign insurance have often been only partially successful because many countries have achieved only moderate coverage or have dropped coverage completely. Often, the small size of sovereign insurance contracts means that while they provide valuable liquidity to support disaster response, they cover only a small portion of the cost of disasters to governments. For example, the World Bank played a critical role in establishing the Caribbean Catastrophe Risk Insurance Facility, a multicounty parametric insurance pool that provides insurance coverage to governments for cyclones, earthquakes, and storms at a relatively low cost. The World Bank has been successful in supporting the expansion of participating countries, and the program has widened the hazards it offers coverage for, but countries have not significantly increased coverage levels. Across sovereign insurance programs, governments also often signal a desire for insurance coverage for frequently occurring disaster events but find the cost of insuring such events too high. Several countries have discontinued their use of World Bank–supported sovereign disaster insurance. Some Pacific Island countries dropped participation in a regional parametric insurance program because the disasters they faced were not trigger events. The Solomon Islands dropped its insurance coverage because it did not receive a payout to cover earthquake or storm damage, because the earthquake damage was not sufficient to meet the selected threshold, and the wind speed of the storm fell short of the cyclone level needed to trigger payments, despite significant flooding. Similarly, Vanuatu and the Marshall Islands dropped coverage because the insurance covered cyclones but not damages from drought (experienced by the Marshall Islands) or from volcanoes (experienced by Vanuatu). Other countries, however, have continued their coverage, such as Samoa and Tonga. In the Philippines, the government chose not to renew the parametric disaster insurance program supported by a World Bank DPO, finding that a catastrophe bond better met their disaster risk financing needs.

Despite their limitations, disaster insurance activities have made progress on awareness raising, capacity building, and product development and have mobilized private capital. These are important building blocks for future progress on insurance market development and the broader ability of governments to manage their financial disaster risks. For example, in the Marshall Islands and Vanuatu, experience with insurance projects and associated policy dialogue increased government capacity to consider financial disaster risk and made the countries more sophisticated consumers of financial DRM products, including noninsurance approaches. In the Philippines, risk modeling developed for a discontinued parametric insurance program helped enable the creation of a catastrophe bond. Disaster insurance and other catastrophe risk instruments have also served to mobilize private capital, which is critical because public funds alone cannot offer sustainable disaster risk finance solutions.

Disaster Risk Reduction Policy Reforms

The World Bank has used policy lending instruments to engage on DRR policy reform and to provide disaster-contingent credit. The World Bank approved 84 DPOs since 2010 (14 percent of the DRR portfolio) that include DRR policy actions. An important subset (27 operations) are catastrophe deferred drawdown options (CAT DDOs), which provide a contingent credit line that can be accessed after a disaster, rather than disbursement at the time of approval, and have policy actions related to improving disaster policy. Of the 84 operations, 33 are closed and have completed self-evaluations, which are the subject of this effectiveness analysis.

DPOs supporting DRR policy reforms have achieved nearly three-quarters of their DRR indicator targets, but only a small number of these indicators captured downstream results. Many DPOs with DRR actions have broad objectives that are not closely related to disasters, so achievement of indicator targets is a more useful metric of success than achievement of objectives. DPOs have supported the adoption of disaster and emergency strategies and plans, disaster mainstreaming in public investment plans and sectors, resilient infrastructure, and disaster risk finance. Across the evaluated DPO portfolio, there were 119 DRR indicators. All but five of the evaluated operations included relevant DRR indicators, and 72 percent achieved their targets. However, a majority (61 percent) of indicators captured upstream measures such as the issuance of regulations or approval of frameworks, while a smaller percentage of them (39 percent) captured downstream measures such as tracking implementation of policy measures at a subnational level, operationalization of new institutions, or changes in behaviors.

Factors of Effectiveness for Disaster Risk Reduction Policies

DPOs are most effective when they include a strong policy matrix and monitoring framework with indicators that can show tangible progress of risk reduction actions, yet many DRR prior actions have been very process oriented. A significant share of DPOs’ prior actions (28 percent) was excessively process oriented and did not give confidence that a policy change would be achieved. For example, in Panama, a prior action required only that an agency submit to the executive branch a proposal for enabling the agency to design, develop, and implement financial protection measures. In Sri Lanka, a prior action required cabinet approval to establish a steering committee to monitor a program for sharing spatial data. In Honduras, a prior action defined responsibilities for local emergency plans, not their adoption. Stronger operations included substantive prior actions: in the Philippines, operations included a requirement for direct budget allocations for risk reduction programs, implementation of a risk layering strategy by setting up new financial instruments, and the operationalization of an earthquake resilience program.

CAT DDOs have provided a timely and important source of post-disaster financing that meets government needs. An advantage of the instrument has been its soft trigger mechanism, which has enabled governments to access funds when needed based on a declaration of emergency and to finance emergency response and recovery. This avoids a problem faced by parametric financial mechanisms in which financial support is unavailable if a disaster occurs that does not precisely fit the parametric trigger (for example, a cyclone that causes severe flooding but does not meet a wind speed trigger). Recently approved CAT DDOs have included public health emergencies as a triggering event, and nearly all CAT DDOs were triggered in 2020 to support COVID-19 response activities: the outstanding balance of CAT DDOs fell from roughly $2 billion to $100 million in 2020. Countries that chose not to trigger their credit line either did not yet have a severe COVID-19 situation or waited to trigger until they had passed through cyclone season. An important factor for successful use of contingent credit lines is the need for clarity and understanding on trigger conditions to ensure they are used to improve financial DRM and not merely as easily accessible budget support. For example, in the first generation of CAT DDOs, some clients did not understand the expectations for trigger and saw it as general budget support: in the first Philippines CAT DDO, the government triggered the $500 million credit line only a week after it became effective, for a relatively minor disaster (World Bank 2017i). These issues have been largely resolved through World Bank dialogue with the client, greater familiarity with the instrument, and World Bank analytics that help the government optimize the timing of drawdowns. In Serbia, improved government understanding of the instruments’ procedures reduced the disbursement lag from 28 days for the first withdrawal to 5 days for their third withdrawal.

A key advantage of DPF including CAT DDOs has been their use as a platform to engage ministries of finance and budget and economic planning agencies. As discussed in chapter 3, these agencies have the influence and leverage to allocate resources to DRR and to influence line agencies to act. DPFs that engage on disaster risk finance have been a useful entry point for engaging finance ministries, as it touches on their core business. For example, in Peru, the first CAT DDO was able to engage the Ministry of Economy and Finance on a financial protection strategy against disasters, including through the use of a range of instruments: a fiscal stabilization fund, sovereign bonds, contingent finance, and catastrophe bonds. Yet, in countries where substantial DRR engagements already exist, DPF operations have been most effective when they leveraged analytical work and relationships built through other instruments. In Grenada, a multisector DPO on broader fiscal risk management made progress on regulatory systems for physical planning (though not on compliance with building codes) by building on government relationships and technical knowledge developed through DRR investments.

DPF has been most effective at catalyzing DRR implementation when combined with other lending instruments and supported by complementary technical assistance. Policy lending operations have often been most effective when combined with investment lending. In the Philippines, DPF and IPF generated mutual leverage, with a CAT DDO providing the policy framework for disaster-sensitive community-driven development and conditional cash transfers, while these mechanisms were implemented through IPF. In Sri Lanka, DPF supported overall resilience policy, while IPF supported the implementation of resilient infrastructure design for roads. Nearly all evaluations of DPF highlight the role of technical assistance in building the capacity needed to operationalize and implement policy reforms. Adopting international best practices on DRR is relatively new in many countries, so civil services often do not have the necessary skills. For example, in Sri Lanka, DPF helped achieve the adoption of a national disaster management plan and a national spatial data infrastructure concept that was lagging for years, while technical assistance supported implementation of the plan and data infrastructure.

The absence of sufficient engagement with subnational governments has meant that DRR and DPF have not always achieved desired downstream results. Engaging on a subnational level is important for DRR, as many DRR responsibilities are carried out by regional, municipal, or local governments, so policy changes at the national level must be implemented and operationalized through local authorities. However, many DPF programs have had difficulty achieving downstream results because of the focus of the instrument and World Bank team engagement on national government counterparts, as well as due to lower technical capacity and coordination challenges at the local level. For example, a CAT DDO in Sri Lanka supported a national policy and plan for disaster management, but only one of nine provinces adopted a standard bylaw issued to regulate and supervise its implementation, and only two of nine provinces completed basin-wide risk mitigation investment plans required under the policy.

Achieving Highly Successful Disaster Risk Reduction Results

The World Bank has sometimes been highly effective in achieving disaster risk reduction results that go beyond the direct effects of project interventions. This evaluation identified what has worked to achieve DRR results in client countries in highly successful cases. An activity was defined to be highly successful if it effectively addressed a major developmental challenge (relevance), addressed root causes to support a change in trajectory (depth of change), or is leading to large-scale impacts (scale of change). These highly successful results exist on a pathway toward achieving transformational change. The evaluation conducted four case studies on specific DRR engagements deemed to be highly successful (see appendix A). Box 4.3 describes the cases, and the following section describes success factors.

Box 4.3. Examples of Highly Successful Disaster Risk Reduction Results

Flood preparedness and risk mitigation in Bihar, India. In Bihar, the World Bank has contributed to flood mitigation and preparedness at scale. Bihar is India’s most flood-prone state, with 76 million people facing recurring flood threats. Devastating floods in 2008 affected more than 3.3 million people. The World Bank supported flood disaster risk reduction through $470 million in lending for the Bihar Kosi Flood Recovery Project (2010–18) and the Bihar Kosi Basin Development Project (2015–). These projects supported structural improvements for 70 kilometers of embankments in the Kosi river basin, decentralized approaches to embankment monitoring and maintenance, flood forecasting, and early-warning systems that have been scaled to other basins. Flood forecast systems have improved to the point of providing 90 percent accuracy in forecasts at a lead time of 72 hours, and work is being done to expand this to five days. This enables vulnerable people to evacuate before floods hit, reducing deaths and losses. The World Bank is continuing to help expand these measures to the entire state.

Resilient schools in Mozambique. Improving the resilience of schools in Mozambique has been critical because the schools’ structural weakness and exposure to disasters have resulted in an average of 550 classrooms being destroyed annually by cyclones and floods. The World Bank achieved success by using advisory services and analytics to develop structurally resilient school building designs and standards, supporting policy changes to adopt these standards for all new construction, financing an initial set of school retrofits through a Program-for-Results, and working with partners to establish a sectorwide funding platform for school reconstruction. Since 2016, all newly constructed classrooms have followed resilient design standards, and 5,762 classrooms were constructed after the standards (as of 2019). All schools constructed under the standards survived the severe cyclones in 2019.

Integrated flood management in Metro Manila, the Philippines. Manila, a megacity with a population of 13.5 million, faces severe flood risks. A 2009 typhoon caused flooding that submerged 80 percent of the city, killed hundreds of people, and caused damage equivalent to 2.7 percent of the national gross domestic product. The flood management system was haphazard and reactive, based on administrative boundaries. Since then, the city has changed its trajectory to adopt an integrated, coordinated, and long-term approach to flood management through a master plan.

The plan development required building evidence-based consensus among agencies responsible for flood management and all metro-area mayors. The World Bank supported a Post-Disaster Needs Assessment that convened stakeholders in relation to recovery and disaster risk reduction. It also used consultative analytical work to conduct a detailed risk assessment, leading to the design and approval of the Flood Management Master Plan in 2012, laying out flood mitigation works that would cost approximately $7.5 billion over a 20- to 25-year period. The World Bank then supported feasibility and design studies to operationalize the plan. In 2018, the World Bank approved $500 million for the Metro Manila Flood Management Project, which finances the plan’s implementation. Once complete, the plan will have dramatically reduced urban flood risk.

Seismic risk mitigation in Istanbul, Türkiye. Istanbul is a megacity of 15 million people and the nation’s economic engine, but it faces severe vulnerability from earthquakes. The probability of a major earthquake between 2004 and 2034 was estimated at 62 percent, with damage of $20 to 60 billion. It has achieved a major reduction in earthquake vulnerability through a program that improved the resilience of public buildings. The World Bank provided $563 million to the Istanbul Seismic Risk Mitigation and Emergency Preparedness Project (2005–15) that financed many earthquake risk mitigation measures, including emergency communication and information management systems, emergency response capacity, public awareness and training, and retrofits and reconstruction of public buildings. The project’s subnational place-based multisector model and its creation of a highly effective project coordination unit established a platform capable of attracting significant financing. As of 2021, the program had attracted €2,219 billion from eight donors (including the World Bank), conducted resilient reconstruction for 430 buildings, and retrofitted 1,105 buildings, focusing on hospital and schools. These upgrades represent 83 percent of all of Istanbul’s vulnerable schools and 53 percent of vulnerable hospitals. The upgrades will reduce damage and save lives: an economic analysis for the World Bank financing only (a quarter of the program) estimated that damage to the improved buildings would be reduced from 40 to 5 percent and that at least 3,000 lives would be saved in the event of an earthquake.

Sources: Independent Evaluation Group case studies; World Bank 2017f.

Analytical work that brings international best practices and robust technical assessments has often been integral to catalyzing and designing highly successful interventions. In Bihar, the World Bank supported increased state-level flood resilience by providing international knowledge on improved embankment designs and building government acceptance for these designs through consultation and trainings. The World Bank also made available experts to help develop and test flood models and drew on regional work on inundation mapping and flood risk forecasting to support systems and knowledge development. In Mozambique, analytical work that included risk assessments and a catalog of resilient architecture and construction convinced government officials to adopt new models of school construction with higher up-front costs to reduce reconstruction costs. In Manila, extensive flood risk assessments coupled with feasibility and design studies of priority flood protection measures contributed to the city’s change trajectory toward developing a citywide flood management system. In Istanbul, ASA contributed to a rank-ordered list of investment priorities based on vulnerability. This analysis helped the city achieve a significant reduction of seismic risk in its public buildings by helping it balance competing priorities, insulating the project from political pressure, and helping the project retain a focus on earthquake risk mitigation rather than other activities. The rank-ordered list of investment priorities also made scale-up easier.

Highly successful DRR results were achieved through adaptive approaches that prioritized the development needs of the client. In Bihar, the World Bank accepted the clients’ need to prioritize reconstruction, with an understanding that over time the balance could be adjusted toward risk reduction. In Mozambique, the World Bank’s approach allowed for some unconventional designs—for example, schools constructed by communities that would not meet formal design standards but would be safer than the status quo and feasible to implement. In Manila, parts of the master plan addressed chronic solid waste issues that compromised the integrity of pumping mechanisms.

Highly successful results have often been achieved when the World Bank has deliberately engaged in supporting replication by others. In Bihar, the World Bank initially focused on piloting improved designs for embankment strengthening, which were applied to the most degraded embankments, and these techniques were replicated and scaled up through subsequent investments. Flood forecasting work initially focused on a single river basin, but additional World Bank support—including technical assistance to enable flood forecasting systems to use public domain software, as well as other institutional- and capacity-strengthening efforts—helped expand the system across the entire state. In Mozambique, an aspect of success was that the World Bank was able to promote the new standards through an Education Sector Support Fund that pooled all donor financial assistance for school construction. By providing training, fiduciary supervision, and construction oversight, the World Bank reduced other donor concerns about corruption risks so that they were willing to contribute to the fund. In Manila, convening other development partners in relation to a common vision helped attract trust fund support and cofinancing of infrastructure plans. In Istanbul, the World Bank helped establish a strong platform, including implementation arrangements and financial, procurement, and monitoring and evaluation systems. Based on this well-functioning and transparent system and confidence in World Bank standards, other international finance institutions added their support, substantially increasing the scale of results achieved. However, the project model of subnational multisectoral approach to earthquake risk reduction has not been replicated elsewhere in Türkiye because its modality required exceptional features and specific enabling legislation and did not follow the preferred centralized approach of the national government.

Given its highly technical nature, achieving highly successful DRR results has required a strong focus on institutional strengthening and awareness raising through demonstrations and trainings to shift mindsets. In Bihar, developing capacity to reliably forecast and disseminate flood events required sustained effort to strengthen institutions and establish new ones with the requisite technical skills to use and maintain the new systems. The World Bank helped shift mindsets through study tours for senior officials, knowledge exchange through workshops, and the embedding of expert consultants in new flood centers. Acceptance from engineers came from growing familiarity and recognition that new designs were lower cost and more sustainable. In Mozambique, continuous institutional support and capacity building in a low-capacity context helped overcome complex procurement challenges to achieve more resilient schools. The World Bank helped improve site selection and construction practices through both on-the-job training for contractors and communities and quality control processes. In Istanbul, the World Bank’s support for creating a semiautonomous, highly capable, professional project coordination and implementation unit was critical for achieving the significant level of progress on seismic risk reduction.

In all case studies, success has required sustained engagement over a decade or more and the use of multiple instruments and interventions. In Bihar, the World Bank had a 14-year engagement and has used two main investment lending operations plus trust-funded technical assistance. In Manila, long-term engagement since 2009—buttressed by an extensive trust-funded technical assistance program—has been key to building political and technical buy-in, and investment financing was necessary to operationalize the plan. In Istanbul, sustained World Bank support provided technical knowledge, credibility to decision makers that proposed solutions met international good practice standards, and confidence that tenders were competitive and fair. The project also built on relationships, trust, and preparatory work carried out under a prior earthquake reconstruction project.

  1. Examples of emergency projects include investment operations that seek to restore agricultural (including fishing) livelihoods and devastated rural and urban infrastructure in small island developing states such as Dominica and Samoa, as well as in Haiti and Zimbabwe, but that do not measure anticipated risk reduction effects associated with flood or drought.
  2. Although this may be because risk estimations were carried out in other studies or projects not mentioned in the project documentation, in these cases, it is not possible to verify how effective the standards were in strengthening asset resilience. Similarly, it is difficult to ascertain from the project documentation reviewed how effectively standards were implemented in practice in the field.