- Choice Modeling (CM)
- Contingent Valuation
Method (CVM)
The revealed preference method utilizes observed behavior among the individuals that has already occurred and makes use of this to approximate suggested willingness to pay for a change in mortality risk. This method has an advantage over the stated preference approach in that if a person pays a certain amount for a commodity, it is known with conviction that the same person’s WTP for that commodity is at least the amount he/she is willing to pay. The four methods used to reveal preferences include: (a) the hedonic pricing method; (b) the travel cost method; (c) the cost of illness approach; (d) the replacement cost method [ 14 , 15 , 16 ].
In contrast with revealed preference methods, the stated preferences method creates a hypothetical market in a survey. It parallels a market survey and estimates a willingness to pay for hypothetical reduction in mortality risks, since it resembles market behavior. In addition, stated preference methods incorporate both active and passive use of a commodity by the consumer. Direct or active values arise when an individual physically experiences the commodity, while passive or indirect values entail that an individual does not directly experience the commodity. The three methods used for stated preferences include: (a) the contingent valuation method; (b) the choice modeling method; (c) life satisfaction analysis [ 17 , 18 , 19 ].
Non-behavioral methods are not necessarily based on human choices and cognitive biases which affect the choices subconsciously. They include the human capital method (HCM) [ 20 ] and life quality index method (LQI) [ 21 ] to estimate the valuation of statistical life, and they are used to elicit the value of an individual in a society in the absence of a possibility to conduct a survey pre- or post- disaster.
In the selected literature, 7 papers out of 16 used stated preference methods. Within stated preference methods, two papers used choice modeling, while the other five used a contingent valuation method.
Papers using choice modeling method included Bockarjova et al. (2012) [ 22 ] and Rheinberger (2011) [ 23 ]. While Bockarjova et al. (2012) [ 22 ] carried out a choice modeling experiment via an internet-based questionnaire and elicited responses from people living in flood prone areas in the Netherlands in two separate studies, Rheinberger (2011) [ 23 ] undertook a choice experiment by recruiting respondents via a phone call prior to a mail survey.
For contingent valuation method, Leiter et al. (2010) [ 24 ] used face-to-face interviews and elicited people’s willingness to pay to prevent an increase in the risk of dying in a snow avalanche. Similarly, Hoffmann et al. (2017) [ 26 ] used a computerized payment card method to estimate the willingness to pay to reduce mortality risk in Chinese population living in four different cities in China. In contrast, Ozdemir (2011) [ 25 ] used a contingent valuation method as well, but used a mail survey to elicit willingness to pay to reduce the risks from tornadoes in the USA.
For non-behavioral methods, Dassanayake et al. (2012) [ 35 ] used a quality of life index method to evaluate intangible flood losses and integrate them into a flood risk analysis.
Other papers used one or a combination of methods. For example, Porfiriev (2014) [ 31 ] approached the economic valuation of human losses resulting from natural and technological disasters in Russia using the theory of welfare and an international comparative approach. Cropper and Sahin (2009) [ 12 ] used the comparative approach, along with transferring the VSL from USA to a whole list of countries classified by income groups by the OECD to estimate VSL.
There was a wide range of VSL values in the literature, ranging from ISD 143,000 to 15 million for one life [ 12 , 25 ]. Table 2 summarizes the estimated value of statistical lives in the articles included in the review. Disaster types range from natural disasters to technological disasters with some disaster types appearing more often than others in the literature, with earthquakes and floods being the most common. The VSLs appeared to increase over the years: while it was estimated to be USD 0.81 million in 2005 in Switzerland in the context of avalanches [ 34 ], it was evaluated between USD 6.8 and 7.5 million in 2011 [ 23 ].
Estimated values of statistical life in included articles.
Reference | VSL (in Millions USD *) | Countries | Disaster Types |
---|---|---|---|
Cropper and Sahin (2009) [ ] | 0.143 (Low-Income-Country) 4.27 (High-Income-Country) | Not Specified | Not Specified |
Porfiriev (2014) [ ] | 0.19 (International comparison) 0.33 (Welfare method) | Russia | Natural and technological |
Hoffmann et al. (2017) [ ] | 0.61 | China | Not Specified |
Sadeghi et al. (2015) [ ] | 0.73–1.4 | Iran | Earthquakes |
Fuchs and Mcalpin (2005) [ ] | 0.81 | Switzerland | Avalanches |
Daniell et al. (2015) [ ] | 2.2 | Australia, calculations applied to case studies in Turkey and Croatia | Earthquakes |
Cheng (2018) [ ] | 2.34 | Australia | Heatwave |
Leiter et al. (2010) [ ] | 2.3–4 | Austria | Avalanches |
Dassanayake et al. (2012) [ ] | 2.5–9.2 | Germany | Floods |
Zhai et al. (2003) [ ] | 3.3–9.2 | Japan | Floods |
Johansson and Kristrom (2015) [ ] | 5.2–12.8 | USA | Floods and storms |
Rheinberger (2011) [ ] | 6.8–7.5 | Switzerland | Snow avalanche and rockfalls |
Barbier (2022) [ ] | 1.25–7.7 | Italy | Earthquake |
Bockarjova et al. (2012) [ ] | 9.6 | The Netherlands | Floods |
Hammitt et al. (2019) [ ] | 10 | China | Not specified |
Ozdemir (2011) [ ] | 15 | USA | Tornado |
* Values were converted into United States Dollars (USD) in respective years. Source: Authors [ 37 ].
Disasters are complex events, and the assessment of losses they have caused is a compounded task. This review’s exploration of literature estimating the value of statistical life with regard to disasters highlighted the complexity and variability of the estimation of values of statistical life and the methods involved.
The geographical locations of studies included in the review showed the parts of the world where most of the studies were focused. An overwhelming majority of studies estimated the value of statistical life in high-income countries. The main reasons for this are related to the data availability and the investment made by developed countries in research and development for the advancement of science in general [ 38 ]. Low- and middle-income countries often experience several disasters occurring year round, and become trapped in a loop of disaster recovery and management annually. Amid ever-present financial constraints, disaster risk reduction and management planning to deal with disasters and their impact in the country therefore becomes much more demanding [ 39 ].
The estimation of economic damages due to disaster in a low-resource setting can also be challenging. Not all the houses, agricultural land, crops and other assets are insured in low- and middle-income countries. The insurance coverage is relatively small if not non-existent in these countries [ 40 ] and the data to quantify the impacts of disasters, such as the number of deaths, missing, affected population as well as reconstruction costs, are often incomplete and not well recorded. So, the unavailability of appropriate information becomes a big challenge in the first step of conducting research. This might be the reason why low- and middle-income countries are not well represented in studies estimating the value of life in disasters. As a result, the lack of studies in low- and middle-income countries can lead to a certain degree of extrapolation of results found in VSL calculation in high-income-country-based studies.
Furthermore, we note that the majority of articles measuring the value of life were about floods. Floods are indeed the most common type of disasters. In an analysis of disasters recorded in the EM-DAT database from 2000 to 2019, nearly half (n = 3254) of all recorded events (n = 7348) were floods [ 41 ]. However, there are many other types of disasters, and it is important to rely on such studies where those disasters were considered when measuring the value of a statistical life.
Methods used for VSL estimations showed significant diversity among the articles included in this review. Although the stated preferences method is the most frequent, it is closely followed by the adaptation method. There could be various reasons for this difference in methodologies across the literature. For instance, non-marketed good with no complementary or substitute market good may not have readily available individual data, and hence may lead the researchers to undertake stated preference methods with which to elicit people’s willingness to pay to reduce a hypothetical disaster risk through surveys [ 19 ]. The scope of the study and the budgetary constraints may also explain why a researcher chooses one method over the other. Additionally, the characteristics of the survey participants are another important factor, as they influence the type of survey that can be conducted and the methodology adopted. For example, if the target population is old and poor, face-to-face interviews in respondents’ private homes might be more suitable than internet-based questionnaires [ 42 , 43 ].
There was a wide range of monetary values of the VSL in the literature. These differences could be due to the level of income of the country where the disaster occurred [ 40 ]. The method of calculation could be another reason for such differences, for example, as consumers optimize their lifetime utility, thus neglecting intergenerational (long-term) utility, using willingness to pay (WTP) methods for a reduction of risk can often lead to overestimated values [ 44 , 45 ]. It could also simply be due to the differences in cultural norms between countries [ 40 ]. Furthermore, the context and the aim of the research and its evolution over the years might also explain variations across the studies. Further studies are required to establish a concrete cause for this observation. It should also be highlighted that low VSL estimates in low-income countries do not inherently mean that a human life is worth less. It could simply reflect individual income, the cost of commodities and the value of currency [ 8 , 46 ].
This study presents a number of limitations. First, the review only included articles published in English, and some studies may exist in other languages. Second, papers that did estimate a VSL considered a range of different methods, and therefore direct comparison of estimated values was not straightforward. Papers referring to economic impact in terms of natural environment or animals were also excluded, as they do not refer to value of statistical life; however, they can be important for calculating overall economic cost of disasters [ 47 , 48 ].
This study aims to explore literature estimating the value of statistical life with regard to disasters through a systematic review. After applying the inclusion criteria on the 2121 articles found in the initial keywords search, only 27 articles were included for final review. In the included literature, several attempts at estimating the value of statistical lives in disasters were identified; however, there was no consensus on the method used, and few investigations were carried out in a low- and middle-income country context. This review therefore provides a limited view of the value of statistical life calculations in disaster settings, which may become useful when implementing disaster risk reduction policies and calculating global losses incurred due to disasters. It reveals that an agreed, robust and multi-sectoral approach for the disaster and economics community remains to be defined.
For PubMed MeSH, terms such as sanctity of life, life sanctity, life sanctities, respect for life, economic life valuation, life valuation/s, economic valuation/s and economic life were used. In addition to this, the search was performed in SCOPUS and GOOGLE SCHOLAR.
The data extraction form recorded the descriptive aspect of all the studies included in the review, the including methodology used to calculate VSL, results, strengths and limitations. A total of 16 categories of information were extracted:
(1) Author, (2) Title, (3) Year published, (4) Journal, (5) Study location, (6) Aim of the study, (7) Disaster type, (8) Type of study (Theoretical/Empirical), (9) Study data source, (10) Study participants, (11) Method of VSL estimation, (12) VSL if given, (13) Strengths, (14) Limitations, (15) Relevant references and (16) Study design.
We are grateful to the European Commission for providing the Erasmus Mundus Grant for completing the Erasmus Mundus Master Course in Public Health in Disasters (EMPHID). We also thank USAID/DCHA/OFDA [ref no. 72OFDA20CA00072] for funding the research at Centre for Research on the Epidemiology of Disasters at the Universite catholique de Louvain.
A.K.: Formal analysis, investigation, writing—review & editing; S.B.: Formal analysis, investigation, writing—original draft; M.M.d.A.: Conceptualization, methodology, project administration, supervision, writing—original draft, writing—review & editing; R.C.D.: Funding acquisition, supervision, review & editing; P.A.G.: Funding acquisition, supervision, review & editing; S.T.: Conceptualization, funding acquisition, methodology, project administration, supervision, validation, writing—original draft, writing—review & editing. All authors have read and agreed to the published version of the manuscript.
Informed consent statement, data availability statement, conflicts of interest.
The authors declare no conflict of interest.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Chapter: summary.
No person or place is immune from disasters or disaster-related losses. Infectious disease outbreaks, acts of terrorism, social unrest, or financial disasters in addition to natural hazards can all lead to large-scale consequences for the nation and its communities. Communities and the nation thus face difficult fiscal, social, cultural, and environmental choices about the best ways to ensure basic security and quality of life against hazards, deliberate attacks, and disasters. Beyond the unquantifiable costs of injury and loss of life from disasters, statistics for 2011 alone indicate economic damages from natural disasters in the United States exceeded $55 billion, with 14 events costing more than a billion dollars in damages each.
One way to reduce the impacts of disasters on the nation and its communities is to invest in enhancing resilience. As defined in this report, resilience is the ability to prepare and plan for, absorb, recover from, and more successfully adapt to adverse events . Enhanced resilience allows better anticipation of disasters and better planning to reduce disaster losses—rather than waiting for an event to occur and paying for it afterward.
However, building the culture and practice of disaster resilience is not simple or inexpensive. Decisions about how and when to invest in increasing resilience involve short- and long-term planning and investments of time and resources prior to an event. Although the resilience of individuals and communities may be readily recognized after a disaster, resilience is currently rarely acknowledged before a disaster takes place, making the “payoff” for resilience investments challenging for individuals, communities, the private sector, and all levels of government to demonstrate.
The challenge of increasing national resilience has been recognized by the federal government, including eight federal agencies and one community resilience group affiliated with a National Laboratory who asked the National Research Council (NRC) to address the broad issue of increasing the nation’s resilience to disasters. These agencies asked the NRC study committee to (1) define “national resilience” and frame the main issues related to increasing resilience in the United States; (2) provide goals, baseline conditions, or performance metrics for national resilience; (3) describe the state of knowledge about resilience to hazards and disasters; and (4) outline additional information, data, gaps, and/or obstacles that need to be addressed to increase the nation’s resilience to disasters. The committee was also asked for recommendations
about the necessary approaches to elevate national resilience to disasters in the United States.
This report confronts the topic of how to increase the nation’s resilience to disasters through a vision of the characteristics of a resilient nation in the year 2030. The characteristics describe a more resilient nation in which
• Every individual and community in the nation has access to the risk and vulnerability information they need to make their communities more resilient.
• All levels of government, communities, and the private sector have designed resilience strategies and operation plans based on this information.
• Proactive investments and policy decisions have reduced loss of lives, costs, and socioeconomic impacts of future disasters.
• Community coalitions are widely organized, recognized, and supported to provide essential services before and after disasters occur.
• Recovery after disasters is rapid and the per capita federal cost of responding to disasters has been declining for a decade.
• Nationwide, the public is universally safer, healthier, and better educated.
The alternative, the status quo, in which the nation’s approaches to increasing disaster resilience remain unchanged, is a future in which disasters will continue to be very costly in terms of injury; loss of lives, homes, and jobs; business interruption; and other damages.
Building resilience toward the 2030 future vision requires a paradigm shift and a new national “culture of disaster resilience” that includes components of
(1) Taking responsibility for disaster risk;
(2) Addressing the challenge of establishing the core value of resilience in communities, including the use of disaster loss data to foster long-term commitments to enhancing resilience;
(3) Developing and deploying tools or metrics for monitoring progress toward resilience;
(4) Building local, community capacity because decisions and the ultimate resilience of a community are driven from the bottom up;
(5) Understanding the landscape of government policies and practices to help communities increase resilience; and
(6) Identifying and communicating the roles and responsibilities of communities and all levels of government in building resilience.
A set of six actionable recommendations (see Box S-1 at the close of the Summary) are described that will help guide the nation toward increasing national resilience from the local community through to state and federal levels. The report has been informed by published information, the committee’s own
expertise, and importantly, by experiences shared by communities in New Orleans and the Mississippi Gulf Coast, Cedar Rapids and Iowa City, Iowa, and Southern California, where the committee held open meetings.
UNDERSTANDING, MANAGING, AND REDUCING DISASTER RISK
Understanding, managing, and reducing disaster risks provide a foundation for building resilience to disasters. Risk represents the potential for hazards to cause adverse effects on our life; health; economic well-being; social, environmental, and cultural assets; infrastructure; and the services expected from institutions and the environment. Risk management is a continuous process that identifies the hazard(s) facing a community, assesses the risk from these hazards, develops and implements risk strategies, reevaluates and reviews these strategies, and develops and adjusts risk policies. The choice of risk management strategies requires regular reevaluation in the context of new data and models on the hazards and risk facing a community, and changes in the socioeconomic and demographic characteristics of a community, as well as the community’s goals. Although some residual risk will always be present, risk management strategies can help build capacity for communities to become more resilient to disasters.
A variety of tools exist to manage disaster risk including tangible structural (construction-related) measures such as levees and dams, disaster-resistant construction, and well-enforced building codes, and nonstructural (nonconstruction-related) measures such as natural defenses, insurance, zoning ordinances, and economic incentives. Structural and nonstructural measures are complementary and can be used in conjunction with one another. Importantly, some tools or actions that can reduce short-term risk can potentially increase long-term risk, requiring careful evaluation of the risk management strategies employed. Risk management is at its foundation a community decision, and the risk management approach will be effective only if community members commit to use the risk management tools and measures made available to them.
THE CHALLENGE OF MAKING INVESTMENTS IN RESILIENCE
Demonstrating that community investments in resilience will yield measurable short- and long-term benefits that balance or exceed the costs is critical for sustained commitment to increasing resilience. The total value of a community’s assets—both the high-value structural assets and those with high social, cultural, and/or environmental value—call for a decision-making framework for disaster resilience that addresses both quantitative data and qualitative value assessments. Ownership of a community’s assets is also important; ownership establishes the responsibility for an asset and, therefore,
the need to make appropriate resilience investments to prepare and plan for hazards and risks. Presently, little guidance exists for communities to understand how to place meaningful value on all of their assets. Particularly during times of economic hardship, competing demand for many societally relevant resources (education, social services) can be a major barrier to making progress in building resilience in communities.
Accessing and understanding the historical spatial and temporal patterns of economic and human disaster losses in communities in the United States are ways for communities to understand the full extent of the impact of disasters and thereby motivate community efforts to increase resilience. Historical patterns of disaster losses provide some sense of the magnitude of the need to become more disaster resilient. The geographic patterns of disaster losses—e.g., human fatalities, property losses, and crop losses—illustrate where the impacts are the greatest, what challenges exist in responding to and recovering from disasters, and what factors drive exposure and vulnerability to hazards and disasters. Although existing loss databases in the United States are useful for certain kinds of analyses, improvement in measurements, accuracy, and consistency are needed. Furthermore, the nation lacks a national repository for all-hazard event and loss data, compromising the ability of communities to make informed decisions about where and how to prioritize their resilience investments.
MEASURING PROGRESS TOWARD RESILIENCE
Without some numerical means of assessing resilience it would be impossible to identify the priority needs for improvement, to monitor changes, to show that resilience had improved, or to compare the benefits of increasing resilience with the associated costs. The measurement of a concept such as resilience is difficult, requiring not only an agreed-upon metric, but also the data and algorithms needed to compute it. The very act of defining a resilience metric, and the discussions that ensue about its structure, helps a community to clarify and formalize what it means by the concept of resilience, thereby raising the quality of debate. The principles that resilience metrics can entail are illustrated by some existing national and international indicators or frameworks that address measurement of the resilience of different aspects of community systems. The Leadership in Energy and Environmental Design for developers, owners, and operators of buildings is one example. Comparison of the strengths and challenges of a variety of different frameworks for measuring resilience suggests that the critical dimensions of an encompassing and consistent resilience measurement system are
• Indicators of the ability of critical infrastructure to recover rapidly from impacts;
• Social factors that enhance or limit a community’s ability to recover, including social capital, language, health, and socioeconomic status;
• Indicators of the ability of buildings and other structures to withstand earthquakes, floods, severe storms, and other disasters; and
• Factors that capture the special needs of individuals and groups, related to minority status, mobility, or health status.
Presently, the nation does not have a consistent basis for measuring resilience that includes all of these dimensions. Until a community experiences a disaster and has to respond to and recover from it, demonstrating the complexity, volume of issues, conflicts and lack of ownership are difficult. A national resilience scorecard, from which communities can then develop their own, tailored scorecards, will make it easier for communities to see the issues they will face prior to an event and can support necessary work in anticipation of an appropriate resilience-building strategy. A scorecard will also allow communities to ask the right questions in advance of a disaster.
BUILDING LOCAL CAPACITY AND ACCELERATING PROGRESS: RESILIENCE FROM THE BOTTOM UP
National resilience emerges, in large part, from the ability of local communities with support from all levels of government and the private sector to plan and prepare for, absorb, respond to, and recover from disasters and adapt to new conditions. Bottom-up interventions—the engagement of communities in increasing their resilience—are essential because local conditions vary greatly across the country; the nation’s communities are unique in their history, geography, demography, culture, and infrastructure; and the risks faced by every community vary according to local hazards. Some universal steps can aid local communities in making progress to increase their resilience and include:
• Engaging the whole community in disaster policymaking and planning;
• Linking public and private infrastructure performance and interests to resilience goals;
• Improving public and private infrastructure and essential services (such as health and education);
• Communicating risks, connecting community networks, and promoting a culture of resilience;
• Organizing communities, neighborhoods, and families to prepare for disasters;
• Adopting sound land-use planning practices; and
• Adopting and enforcing building codes and standards appropriate to existing hazards.
Community coalitions of local leaders from public and private sectors, with ties to and support from federal and state governments, and with input from the local citizenry, become very important in this regard. Such coalitions can be charged to assess the community’s exposure and vulnerability to risk, to educate and communicate risk, and to evaluate and expand the community’s capacity to handle such risk. A truly robust coalition would have at its core a strong leadership and governance structure, and people with adequate time, skill, and dedication necessary for the development and maintenance of relationships among all partners in the community.
THE LANDSCAPE OF RESILIENCE POLICY: RESILIENCE FROM THE TOP DOWN
Strong governance at all levels is a key element of resilience and includes the making of consistent and complementary local, state, and federal policies. Although resilience at its core has to be carried forward by communities, communities do not exist under a single authority in the United States, and function instead under a mix of policies and practices implemented and enforced by different levels of government. Policies that make the nation more resilient are important in every aspect of American life and economy, and not just during times of stress or trauma. A key role of policies designed to improve national resilience is to take the long-term view of community resilience and to help avoid short-term expediencies that can diminish resilience.
Certain policies of the federal Executive Branch, including Presidential Directives and Executive Orders, policies initiated by federal agencies, and policies of the Legislative Branch can and do function to help strengthen resilience. Presidential Policy Directive-8 (PPD-8) calls upon the Department of Homeland Security to embrace systematic preparation against all types of threats, including catastrophic natural disasters. Because the scope of resilience is sometimes not fully appreciated, some who contemplate national resilience policy think first of the Stafford Act and its role in disaster response and recovery. Although the Stafford Act does provide guidance for certain responsibilities and actions in responding to a disaster incident, national resilience transcends the immediate impact and disaster response and therefore grows from a broader set of policies. Many of the critical policies and actions required for improved national resilience are also enacted and implemented at the state and local levels.
Policies at all levels of governance do exist to enhance resilience; however, some government policies and practices can also have unintended consequences that negatively affect resilience. Furthermore, gaps in policies
and programs among federal agencies exist for all parts of the resilience process—including disaster preparedness, response, recovery, mitigation, and adaptation, as well as research, planning, and community assistance. Although some of these gaps are the result of the legislative authorization within which agencies are directed to operate, the roles and responsibilities for building resilience are not effectively coordinated by the federal government, either through a single agency or authority, or through a unified vision.
Community resilience is broad and complex, making it difficult to codify resilience in a single comprehensive law. Rather, infusing the principles of resilience into all the routine functions of the government at all levels and through a national vision is a more effective approach.
LINKING COMMUNITY AND GOVERNANCE TO GUIDE NATIONAL RESILIENCE
Increased resilience cannot be accomplished by simply adding a cosmetic layer of policy or practice to a vulnerable community. Long-term shifts in physical approaches (new technologies, methods, materials, and infrastructure systems) and cultural approaches (the people, management processes, institutional arrangements, and legislation) are needed to advance community resilience. Resilience to disasters rests on the premise that all aspects of a community—its physical infrastructure, its socioeconomic health, the health and education of its citizens, and its natural environment—are strong. This kind of systemic strength requires that the community members work in concert and in such a way that the interdependencies among them provide strength during a disaster event.
Communities and the governance network of which they are a part are complex and dynamic systems that develop and implement resilience-building policies through combined effort and responsibility. Experience in the disaster management community suggests that linked bottom-up and top-down networks are important for managing risk and increasing resilience. Key interactions within the nation’s resilience “system” of communities and governance can be used to help identify specific kinds of policies that can increase resilience and the roles and responsibilities of the actors in government, the private sector, and communities for implementing these policies. For example, to understand hazards or threats and their processes, research and science and technology policies allow federal and state agencies to coordinate efforts on detection and monitoring activities that can be used by regional and local governing bodies, the private sector, and communities to evaluate and address their hazards and risks. Identifying resilience policy areas, identifying those in community and government responsible for coordinating activities in those areas, and identifying the recipients of the information or services resulting from those activities reveal strengths and gaps in the nation’s resilience “system.”
Advancing resilience is a long-term process, but can be coordinated around visible, short-term goals that allow individuals and organizations to measure or mark their progress toward becoming resilient and overcoming these gaps. However, as a necessary first step to strengthen the nation’s resilience and provide the leadership to establish a national “culture of resilience,” a full and clear commitment to disaster resilience by the federal government is essential.
BUILDING A MORE RESILIENT NATION: THE PATH FORWARD
No single sector or entity has ultimate responsibility for improving national resilience. No specific federal agency has all of the authority or responsibility, all of the appropriate skill sets, or adequate fiscal resources to address this growing challenge. An important responsibility for increasing national resilience lies with residents and their communities. Input, guidance, and commitment from all levels of government and from the private sector, academia, and community-based and nongovernmental organizations are needed throughout the entire process of building more resilient communities. The report frames six recommendations (Box S-1) that can help guide the nation in advancing collective, resilience-enhancing efforts in the coming decades.
BOX S-1 Summary Recommendations
Recommendation 1: Federal government agencies should incorporate national resilience as a guiding principle to inform the mission and actions of the federal government and the programs it supports at all levels .
Recommendation 2: The public and private sectors in a community should work cooperatively to encourage commitment to and investment in a risk management strategy that includes complementary structural and nonstructural risk-reduction and risk-spreading measures or tools. Such tools might include an essential framework (codes, standards, and guidelines) that drives the critical structural functions of resilience and investment in risk-based pricing of insurance.
Recommendation 3: A national resource of disaster-related data should be established that documents injuries, loss of life, property loss, and impacts on economic activity. Such a database will support efforts to develop more quantitative risk models and better understand structural and social vulnerability to disasters.
Recommendation 4: The Department of Homeland Security in conjunction with other federal agencies, state and local partners, and professional groups should develop a National Resilience Scorecard.
Recommendation 5: Federal, state, and local governments should support the creation and maintenance of broad-based community resilience coalitions at local and regional levels.
Recommendation 6: All federal agencies should ensure that they are promoting and coordinating national resilience in their programs and policies. A resilience policy review and self-assessment within agencies and strong communication among agencies are keys to achieving this kind of coordination.
Increasing disaster resilience is an imperative that requires the collective will of the nation and its communities. Although disasters will continue to occur, actions that move the nation from reactive approaches to disasters to a proactive stance where communities actively engage in enhancing resilience will reduce many of the broad societal and economic burdens that disasters can cause.
This page intentionally left blankd.
One way to reduce the impacts of disasters on the nation and its communities is to invest in enhancing resilience—the ability to prepare and plan for, absorb, recover from and more successfully adapt to adverse events. Disaster Resilience: A National Imperative addresses the broad issue of increasing the nation's resilience to disasters. This book defines "national resilience", describes the state of knowledge about resilience to hazards and disasters, and frames the main issues related to increasing resilience in the United States. It also provide goals, baseline conditions, or performance metrics for national resilience and outlines additional information, data, gaps, and/or obstacles that need to be addressed to increase the nation's resilience to disasters. Additionally, the book's authoring committee makes recommendations about the necessary approaches to elevate national resilience to disasters in the United States.
Enhanced resilience allows better anticipation of disasters and better planning to reduce disaster losses-rather than waiting for an event to occur and paying for it afterward. Disaster Resilience confronts the topic of how to increase the nation's resilience to disasters through a vision of the characteristics of a resilient nation in the year 2030. Increasing disaster resilience is an imperative that requires the collective will of the nation and its communities. Although disasters will continue to occur, actions that move the nation from reactive approaches to disasters to a proactive stance where communities actively engage in enhancing resilience will reduce many of the broad societal and economic burdens that disasters can cause.
READ FREE ONLINE
You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.
Do you want to take a quick tour of the OpenBook's features?
Show this book's table of contents , where you can jump to any chapter by name.
...or use these buttons to go back to the previous chapter or skip to the next one.
Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.
Switch between the Original Pages , where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.
To search the entire text of this book, type in your search term here and press Enter .
Share a link to this book page on your preferred social network or via email.
View our suggested citation for this chapter.
Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.
Do you enjoy reading reports from the Academies online for free ? Sign up for email notifications and we'll let you know about new publications in your areas of interest when they're released.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Scientific Data volume 9 , Article number: 572 ( 2022 ) Cite this article
12k Accesses
37 Citations
32 Altmetric
Metrics details
Reliable and complete data held in disaster databases are imperative to inform effective disaster preparedness and mitigation policies. Nonetheless, disaster databases are highly prone to missingness. In this article, we conduct a missing data diagnosis of the widely-cited, global disaster database, the Emergency Events Database (EM-DAT) to identify the extent and potential determinants of missing data within EM-DAT. In addition, through a review of prominent empirical literature, we contextualise how missing data within EM-DAT has been handled previously. A large proportion of missing data was identified for disasters attributed to natural hazards occurring between 1990 and 2020, particularly on the economic losses. The year the disaster occurred, income-classification of the affected country and disaster type were all significant predictors of missingness for key human and economic loss variables. Accordingly, data are unlikely to be missing completely at random. Advanced statistical methods to handle missing data are thus warranted when analysing disaster data to minimise the risk of biasing statistical inferences and to ensure global disaster data can be trusted.
Introduction.
As the global effects of climate change are felt more intensively, so too are the human and economic consequences of catastrophic disaster events. In 2020 alone, disaster events attributed to natural hazards affected approximately 100 million people, accounted for an estimated 190 billion US$ of global economic losses and resulted in 15,082 deaths 1 , 2 . In light of COP-26 and recent topical events including, but certainly not limited to, the Haitian Earthquake (2021), Central European Floods (2020) and Australian Bushfires (2020), a renewed urgency has been granted to the research of, preparedness to and mitigation of disasters attributed to natural hazards.
Comprehensive historical data held in disaster databases are central for numerous purposes across both the public and private domain to inform emergency disaster relief management; configure catastrophe risk assessment models; and conduct cost-benefit analyses of disaster risk reduction policies 3 . However, inconsistencies in the reporting of disaster events and methodological difficulties quantifying their impacts, mean that disaster databases are prone to gaps in data availability 3 , 4 . As a result, the scope and reliability of statistical inferences which can be made from disaster data are reduced 5 . Systematic reporting of disaster events is required to minimise missing data. However, to achieve this is a major challenge. Instead, the use of valid and often complex statistical methods to account for missing data are required 6 , 7 .
Missing data is a common issue across all research areas. Standard practices are adopted in randomised clinical trials to combat patient attrition 8 , 9 and in survey-based observational studies to handle unit- and item-non response 10 , 11 . However, within the disaster literature, there is little insight as to how missing data should be handled.
To date, there are six disaster databases which have global coverage: the Emergency Events Database (EM-DAT); NatCatSERVICE; Sigma; GLIDE; GFDRR; and BD CATNAT Global 12 . This analysis utilises EM-DAT data alone, as it is the only publicly available, global disaster database and is widely cited; an initial search of the terms: ‘EM-DAT’, ‘CRED’, ‘Emergency Events Database’ and ‘International Disaster Database’ in Google Scholar returned 21,000 search results spanning numerous disciplines. EM-DAT was founded in 1988 by the Université catholique de Louvain (Belgium) with support from the United States Agency for International Development (USAID), the World Health Organisation (WHO) and the Belgium Government 13 . Data are collated from sources including the United Nations, reinsurance firms, research institutions and the press. As well as reporting the occurrence of major disastrous events attributed to natural, technological and complex hazards, its meta-data captures the human and economic impacts to assist national and international humanitarian action.
In this study, we conduct a missing data diagnosis to assess the extent of missing data in EM-DAT and to identify observable factors associated with missingness. We restrict our analysis to disaster events attributed to natural hazards occurring between the years 1990 and 2020. In addition, through a review of highly-cited empirical literature utilising EM-DAT data, we illustrate how missing data has previously been dealt with. We conclude by providing a discussion on the potential methods to handle missing data within disaster databases.
The advent of digital technologies in academia and research from the 1980s initiated a proliferation in the collection of data and subsequently, the adoption of data governance and data quality tools to ensure its reliability 14 . In the disaster space, technological advances in disaster surveillance and a progressive global agenda towards standardising disaster reporting, embedded in the 2015–2030 Sendai framework 15 , favour a climate for complete data reporting. Despite this, there was a high proportion of missing data in EM-DAT for disaster events attributed to natural hazards occurring between 1990 and 2020 (Fig. 1 ). This was particularly evident for the reporting of economic losses: data were missing for 96.2% of disaster events on reconstruction costs, for 88.1% on insured damages and 41.5% on total estimated damages. In the three months following a disaster event, when EM-DAT collates the majority of its data, precise information on reconstruction costs and insured damages is likely to be sparse. Reporting of human losses were more complete, with proportions of missing data ranging from 1.3% to 22.3%. Given that the volume of external disaster aid hinges predominantly on death tolls and injury counts, this finding is unsurprising. Of note, there were substantial inconsistencies in the reporting of both human and economic losses, where in both cases, aggregate variables (total deaths and total estimated damages) were better informed. In particular, missing data on total deaths were negligible (1.3%). Hence, in this case, there is little risk of missing data biasing statistical inferences.
Missing data patterns and proportions of missing data for key human and economic loss variables in EM-DAT across all disaster events attributed to natural hazards and occurring between 1990 and 2020. Black shading denotes missing data for one or more disaster events; grey shading denotes observed data. Disaster events are ordered along the x-axis by the STATA default: from the least to the most missing across the variables analysed. The proportion of missing data for each variable, given as a percentage of the total data, is shown to the right-hand side of the figure. Variables are presented in descending order of missingness. Human loss variables include No. of Affected, defined as the number of people requiring immediate assistance during a period of emergency; No. of Missing, defined as the number of people whose whereabouts is unknown and who are presumed dead; No. of Deaths, defined as the number of people who lost their lives as a result of the disaster event; and Total Deaths, defined as the sum of No. of Affected and No. of Deaths. Economic loss variables include Reconstruction Costs, defined as the costs incurred due to the replacement of lost assets and the implementation of disaster mitigation measures; Insured Damages, defined as economic losses born by the insurance sector; and Total Estimated Damages, defined as a value of all economic losses directly or indirectly related to the disaster event.
It is important to distinguish not only the extent, but the pattern of missing data to inform the complexity of missingness 16 . For a small proportion of disaster events, data were missing across all variables analysed, denoted in Fig. 1 by the black, continuous vertical band along the right-hand side. More commonly, data were observed for at least one human loss variable and one economic loss variable. For human losses, missing information may be informed through the interdependence of human loss variables; EM-DAT defines total deaths as the sum of the number of people affected and the number of deaths 17 . Whereas for economic losses, there is no such interdependence across economic loss variables. EM-DAT defines total estimated damages as a value of all the economic losses directly or indirectly related to the disaster event, which does not necessarily include reconstruction costs. Missing data occurred sporadically throughout the dataset for each human loss variable, demonstrating an intermittent missing data pattern. However, across the economic loss variables, missing data patterns were less sporadic. For instance, data on reconstruction costs and insured damages were largely missing throughout, even when data on total estimated damages were informed.
Logistic regression analysis was used to identify whether the probability of data to be missing was dependent on observed data within EM-DAT 5 , 18 . Accordingly, it informs the nature of the missing data mechanism, specifically whether the mechanism of missing data deviates from Missing Completely At Random (MCAR). Data which are MCAR, that is, their probability to be missing is independent of observed and unobserved data, can be excluded from analyses with little risk of introducing bias. Conversely, if the missing data mechanism deviates from MCAR, the deletion of missing values may bias statistical inferences. In this case, more advanced methods are necessary to account for data gaps 6 . Here, we restricted our logistic regression analysis to model the probability of missingness within the variables: total estimated damages, the number of people affected, the number of people missing and the number of deaths (Supplementary Table 1 ). Methods to account for missing data are viable only for partially incomplete data 5 . In the case of reconstruction costs and insured damages, where data were almost entirely incomplete in our dataset (96.2% and 88.1% missing respectively) little can be done during data analysis to account for missingness. Hence, these variables were not investigated in the logistic regression analysis. The opposite was true for total deaths; the proportion of missing data on total deaths was negligible (1.3%). This precludes logistic regression analysis due to insufficient variation in the outcome variable, the probability to be missing. As a result, the variable total deaths was not investigated in the logistic regression analysis.
The observable data partially explained the probability of total estimated damages to be missing, revealed by a sizeable pseudo-R 2 value (pseudo-R 2 = 0.416) (Supplementary Table 1 ). Missing data on the number of people affected and the number of deaths were explained less by the observed data (pseudo-R 2 = 0.206 and pseudo-R 2 = 0.188 respectively) (Supplementary Table 2 ). This agrees with the intermittent missing data pattern identified for these variables (Fig. 1 ). The observed data explained much of the probability to be missing for the variable, the number of people missing (pseudo-R 2 = 0.621) (Supplementary Table 2 ). However, a large number of disaster events (n = 6,218) were omitted from the analysis by STATA, by default, due to occurring in years which perfectly predicted the probability to be missing. As a consequence, the precision of the parameter estimates is reduced, which requires the results to be interpreted with caution.
A number of observable factors in EM-DAT were statistically significantly (p < 0.05) associated to the probability of total estimated damages to be missing: the year the disaster occured, the income-group classification of the affected country, disaster severity and disaster type (Supplementary Table 1 ). Disaster events occurring after the year 2002 and in lower-income countries were positive predictors of the probability of data on total estimated damages to be missing. In addition, disaster severity, estimated by the natural logarithm of total deaths, and disaster types having prolonged effects including droughts, epidemics and extreme temperature events were also positive predictors of missingness.
The observed predictors of missingness differed for human loss variables. In contrast to missingness within the variable total estimated damages, the probability of data to be missing for the variables: the number of people affected and the number of deaths were statistically significantly (p < 0.01) lower for disasters which occurred in lower-income countries relative to high-income countries. Lower-income countries are the predominant recipients of international disaster aid. This result suggests that disaster aid incentivises more complete reporting of human losses. The influence of disaster type on the probability to be missing was largely heterogeneous across each variable analysed. In contrast, associations between the year the disaster occurred and the probability of data to be missing were similar across human and economic loss variables, with a higher probability of missingness for disaster events occurring after 2002.
The absence of published, standard procedures to account for missing data in disaster databases has set a precedent for a general lack of consideration of the issue in the empirical literature. We exemplify this by reviewing the top-20 most cited empirical studies utilising EM-DAT as a primary or secondary data source (Supplementary Table 3 ).
Eight studies neglected to mention missing data. Of the studies that did, missing data were often considered more broadly under the remit of data availability. With the exception of Brooks, Adger and Kelly 19 who assessed missing data in-depth in an accompanying paper, mention of missing data or data availability more generally, was limited to one or two sentences.
All disaster events meeting at least one of the following inclusion criteria are included in EM-DAT: i) ten or more people reported dead; ii) 100 or more people affected, injured or homeless; iii) a declaration of a state of emergency by the affected country, or an appeal for international assistance. Accordingly, for disaster events which meet these criteria, data on disaster occurrences are entirely observed in EM-DAT. As such, the extent to which missing data need be considered is dictated by the type of EM-DAT data utilised in empirical studies, namely whether data are utilised on disaster occurrence alone, or on the consequences.
Statistical methods utilised during data analysis typically presume complete information for all variables specified in the model. Therefore, assumptions on the mechanism of missing data are made, whether explicitly or not. Neither the publication year, nor the number of times a paper had been cited affected the approach taken to handle missing data. Of the studies reviewed, half attempted to account for missing data. However, the approaches employed were mostly ad-hoc with no statistical grounding. In six studies, authors restricted their analysis to variables or subsets of data with improved data availability. In addition to restricting the scope of their analysis, Brooks, Adger and Kelly 19 imputed zero values in place of missing values. However, this approach may heavily skew the distribution of data. Dilley et al . 20 aggregated mortality estimates over all disaster events attributed to a natural hazard between 1981 and 2000. This method intends to reduce the relative impact of missing data. Nevertheless, it also compromises the precision of analyses. Barredo 21 and Doocy et al . 22 supplemented EM-DAT data with data from alternative sources in an attempt to generate a more complete dataset. However, merging of datasets can be time-consuming and can itself introduce ‘file matching’ biases if datasets are heterogeneous in their structure or content 16 . Yang 23 utilised mean imputation whereby missing values are imputed with a single, unconditional mean of the observed values. Nevertheless, this method is generally not recommended, as the uncertainty in the predicted values due to missing data are not adequately reflected 24 .
Identifying a suitable approach to handle missing data will depend on the characteristics of the dataset, the variables of interest and the intended data analysis. Regardless, the choice of missing data method should be grounded on plausible assumptions about the missing data mechanisms as informed through a diagnosis of the missing data. In Supplementary Table 4 , we describe conventional and advanced missing data methods which are relevant to handling missing data within disaster databases. An exhaustive review of the various missing data methods is beyond the scope of this paper and can be found elsewhere, see Allison 24 , Graham 25 and Graham, Patricio and Allison 26 .
Two broad approaches to handle missing data exist: deletion and imputation. Deletion methods, which are so-called ‘quick-fix’ methods, include column deletion, Complete Case Analysis (CCA) and Available Case Analysis (ACA). However, these methods can exclude a large proportion of the original dataset when a high proportion of missing data exists. In addition, deletion methods generally rely on missing data to be MCAR. Hence, when the mechanism of missingness deviates from this assumption, these methods can do more harm than good, biasing statistical inferences reported by studies 7 . Imputation-based methods are of two main varieties: single imputation and multiple imputation. Single imputation methods, such as mean imputation and regression-based imputation replace missing values with a single, predicted value estimated from the observed data. In contrast, multiple imputation generates a range of predicted values through multiply imputed datasets and thus, adequately reflects the uncertainty in the predicted values due to missing data 26 . Other advanced missing data methods include inverse probability weighting, maximum likelihood and Bayesian simulation. In addition, missing data methods can be combined. For instance, Bayesian techniques can be incorporated into the first step of multiple imputation.
Neglecting the issue of missing data exposes studies which inform regional, national and international disaster policies to biases and compromises their effectiveness. The complete, peer-reviewed reporting of disaster events through strengthened field-level data quality remains the eventual goal. In the meantime, given the data available, we argue that to achieve reliable evidence-based disaster policy, it is crucial that missing data is appropriately diagnosed and accounted for in the empirical literature.
This analysis is one of the first to reveal deviations from the assumption of MCAR for missing data on key human and economic loss variables in EM-DAT. Consequently, methods to handle missing data which are MAR should be considered. Deviations from the assumption of MAR, namely that missing data are Missing Not At Random (MNAR) and are thus dependent on unobserved data, cannot be explicitly tested. Instead, sensitivity analyses should be performed to test the robustness of parameter estimates to different assumptions about the mechanisms of missing data. Advanced missing data methods such as multiple imputation, maximum-likelihood and Bayesian simulation can be adapted to account for missing data which are MNAR.
In addition, this analysis highlights key limitations of EM-DAT data. Despite technological advances in disaster surveillance and general progress in data collection, an increase in the proportions of missing data since 2002 suggest shortfalls in current data quality procedures. More-so, the large disparities in the reporting of human losses compared to economic losses points to a wider issue in the types of data collected. Count data in physical units (number of deaths, number of people affected etc.) is principally used to portray human losses, whereas monetary estimates (reconstruction costs, insured damages etc.) are used to denote economic losses. Direct inference of monetary estimates is not straightforward. A two-stage valuation approach may therefore be more suitable and limit the extent of missing data on economic losses. In this case, count data, for instance the number of buildings damaged, the number of days without electricity, would primarily be used to evaluate economic losses in disaster databases. Standard economic valuation methods could then be applied, where necessary, to transform these values into monetary estimates.
By examining the global disaster database EM-DAT, this paper highlights shortfalls in the quality and use of disaster data which are rarely considered. Limitations in the quality of disaster databases arising from data collection procedures have been explored in previous literature 3 , 4 , 27 , 28 . In particular, limitations in the availability of data on low-intensity disaster events due to the inclusion criteria specified by CRED are well known. Instead, this paper contributes a different perspective to the existing literature; given the data collected, to what extent is there an issue of missing data and how should it be handled to ensure global disaster data can be trusted? In this respect, we aim to prompt the reader to apply a critical eye when assessing global disaster data, rather than implicitly trust its reliability.
The extent, complexity and mechanisms of missing data will vary according to the data analysed. Because of this, we are unable to prescribe a single approach to handle missing data within disaster databases. Instead, we advocate for analysts to consider missing data on an individual study basis. Although this article is a step in the right direction, a considerable research effort is still required to inform how missing data within disaster databases should be handled. In particular, the mechanisms of missing data in EM-DAT, and other disaster databases, should be more clearly identified. This would allow data limitations to be better informed and the performance of statistical methods, grounded on plausible assumptions about the mechanisms of missing data, to be tested.
As per procedures initially employed by Faria et al . 18 and Gabrio et al . 29 , but adapted to the context of disaster databases, this missing data diagnosis comprised three steps; i) a description of the proportions of missing data; ii) an analysis of the patterns of missing data; and iii) a logistic regression analysis to assess associations between the probability to be missing and observed values in the dataset. The dataset was restricted to disasters attributed to natural hazards and occurring between the years 1990 and 2020. Data analysis was carried out in STATA (version 16.1) and graphics were produced using the scheme ‘cleanplots’ 30 . The code used for the data analysis can be found in the supplementary material ( Supplementary Material ).
The proportion of missing data was computed across all variables using the command ‘mdesc’ 31 . Missing data patterns were produced using the command ‘misspattern, novarsort’ 32 . Due to computational limits attributed to a large number of observations for each variable, the option ‘noidsort’ could not be applied. Accordingly, observations were arranged by the default: from the least to the most missing across the variables of interest. Missing data patterns were produced for all human loss variables in the dataset (number of people affected, number of people missing, number of deaths and total deaths) and select economic loss variables (reconstruction costs, insured damages and total estimated damages). While individual direct, indirect, sectoral and infrastructural losses were included in the dataset, these were omitted from visualisations of missing data patterns as they were rarely informed.
Statistical analyses, including the t-test, Pearson’s chi-squared test, Little’s MCRA test and binary probability models, are used to test the assumption that missing data is MAR. That is, missing data can be explained by observed values in the dataset 18 . Logistic regression analysis was used in this study to model the probability of missingness for a given disaster event i , conditional on a vector of observable predictors of missingness ( X i ) (Eq. 1 ). The outcome variable ( y i ) was therefore a dichotomous missing data indicator taking value 1 if data were missing, or 0 if data were informed. Missing data indicators were generated for the variables: total estimated damages, number of people affected, number of people missing and number of deaths.
with ε it denoting the error term and F (·) denoting the cumulative distribution function (cdf) which takes a standard logistic distribution (Eq. 2 ) in logistic regression analysis.
Observed predictors of missingness ( X i ) included disaster type, a categorical variable denoting the type of natural disaster, omitting the disaster type flood as the reference to control for multi-collinearity; income-classification of the affected country, omitting high-income as the reference category; and the year the disaster occurred, between 1990 and 2020, included in the model as a categorical variable with the year 1990 omitted as a reference. Reference categories were selected due to being the most frequently observed in the dataset. The year the disaster occurred was included in the model as fixed effects to assess both the trend in missing data over time and the individual threshold effects across each year. Disaster severity, estimated by the logarithm of total deaths, was also included in the model specification when the probability of total estimated damages to be missing was assessed. Otherwise, disaster severity was excluded from the model due to being highly correlated with the variables: number of people affected, number of people missing and number of deaths. Total deaths was log transformed to account for the left-skewed, binomial distribution, arising from the high proportion (60%) of zero reported deaths in the dataset. Observed predictors of missingness were selected based on a step-wise approach.
Logistic regression analysis was estimated by maximum likelihood estimation (Eq. 3 ) using the command ‘logistic’ for all disaster events having complete information on the predictors of missingness. Marginal effects were computed at the mean using the command ‘mfx’ to enable interpretation of the magnitude of effect in the probability scale.
The robustness of our results was assessed using a probit model and a conditional (fixed-effect) logit model. The latter was used to test for potential ‘incidental parameter’ bias arising from the inclusion of individual fixed effects.
Disaster databases are highly prone to missingness. Despite this, little insight exists as to how missing data should be handled in this context. To inform previous approaches to handling missing data in disaster databases, empirical studies utilising EM-DAT as a primary or secondary data source were reviewed. As this review was conducted for illustrative purposes only, we sought only prominent, highly-cited studies since these typically shape the standards for subsequent research. Studies were identified through electronic database searches of Web of Science, Scopus, PubMed and Google Scholar on 01/06/2022 using the key search terms: ‘EM-DAT’, ‘Emergency Events Database’, ‘International Disaster Database’ and ‘CRED’. Only papers considered to be empirical in nature and published between 1990 and 2022 were considered. No language restrictions were applied. Of the initial 2,127 search results, 421 papers were deemed to meet the inclusion criteria. Of these, the top-20 most cited papers were identified. For each paper, data were extracted on the use of EM-DAT data, the consideration given to missing data and the approach taken to handle missing data 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 .
The data utilised in this analysis were acquired from the Emergency Events Database (EM-DAT). Data from EM-DAT is publicly and freely available to download from: https://public.emdat.be/ . Restrictions apply to the dataset used in this analysis, which included additional data from that freely available. This was acquired by the authors on a direct request to the Centre for Research on the Epidemiology of Disasters (CRED). Data is however available from the corresponding author upon reasonable request and with the permission of CRED, which can be obtained via written request to Ms. Regina Below, the EM-DAT database manager, at [email protected].
The code used to generate the results of this analysis are available as supplementary material. Data analysis was carried out in STATA (version 16.1).
International Federation of Red Cross and Red Crescent Societies, National Disaster Reduction Centre of China & Academy of Disaster Reduction and Emergency Management. 2020 Global Natural Disaster Assessment Report . https://www.preventionweb.net/publication/2020-global-natural-disaster-assessment-report (2021).
Bevere, L. Natural catastrophes in 2020. Swiss RE sigma https://www.swissre.com/institute/research/sigma-research/sigma-2021-01.html (2021).
Wirtz, A., Kron, W., Löw, P. & Steuer, M. The need for data: natural disasters and the challenges of database management. Nat. Hazards. 70 , 135–157 (2012).
Article Google Scholar
Guha-Sapir, D. & Below, R. Quality and accuracy of disaster data: A comparative analyse of 3 global data sets. CRED Work. Pap . 1–18 (2002).
Rubin, D. B. Inference and Missing Data. Oxford Univ. Press 63 , 581–592 (1976).
ADS MathSciNet MATH Google Scholar
Harrell, F. E. Regression Modeling Strategies 2nd edn (Springer International Publishing, 2015).
Schafer, J. L. & Graham, J. W. Missing Data: Our View of the State of the Art. Psychol. Methods. 7 , 147–177 (2002).
Schulz, K. F., Altman, D. G. & Moher, D. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. PLoS Med. 7 , 1–7 (2010).
Little, R. J. et al . The Prevention and Treatment of Missing Data in Clinical Trials. N. Engl. J. Med. 367 , 1355–1360 (2012).
Article CAS Google Scholar
Young, R. & Johnson, D. R. Handling Missing Values in Longitudinal Panel Data With Multiple Imputation. J. Marriage Fam. 77 , 277 (2015).
Vandenbroucke, J. P. et al . Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): Explanation and Elaboration. PLOS Med. 4 , 297 (2007).
Mazhin, S. et al . Worldwide disaster loss and damage databases: A systematic review. J. Educ. Health Promot. 10 , 329 (2021).
Article PubMed PubMed Central Google Scholar
Cummings, W. S. In Economic Effects of Natural Disasters (ed. Chaiechi, T.) Ch.1 (Academic Press, 2020).
Foote, K. A Brief History of Data Quality. DATAVERSITY https://www.dataversity.net/a-brief-history-of-data-quality/# (2021).
United Nations Office for Disaster Risk Reduction (UNDRR). Sendai Framework for Disaster Risk Reduction 2015–2030. United Nations https://www.undrr.org/publication/sendai-framework-disaster-risk-reduction-2015-2030 (2015).
Little, R. J. A. & Rubin, D. B. Statistical analysis with missing data (Wiley, 2002).
Centre for Research on the Epidemiology of Disasters (CRED). EM-DAT Glossary https://www.emdat.be/Glossary (2021).
Faria, R., Gomes, M., Epstein, D. & White, I. R. A Guide to Handling Missing Data in Cost-Effectiveness Analysis Conducted Within Randomised Controlled Trials. Pharmacoeconomics. 32 , 1157–1170 (2014).
Brooks, N., Adger, W. N. & Kelly, P. M. The determinants of vulnerability and adaptive capacity at the national level and the implications for adaptation. Glob. Environ. Chang. 15 , 151–163 (2005).
Dilley, M., Chen, R. S., Deichmann, U., Lerner-Lam, A. L. & Arnold, M. Natural Disaster Hotspots: A Global Risk Analysis. World Bank https://openknowledge.worldbank.org/handle/10986/7376 (2005).
Barredo, J. I. Major flood disasters in Europe: 1950-2005. Nat. Hazards. 42 , 125–148 (2007).
Doocy, S., Daniels, A., Murray, S. & Kirsch, T. D. The Human Impact of Floods: a Historical Review of Events 1980-2009 and Systematic Literature Review. PLoS Curr . 5 (2013).
Yang, D. Coping with disaster: The impact of hurricanes on international financial flows, 1970–2002. B.E. J. Econ. Anal. Policy . 8 (2008).
Allison, P. D. In The Sage handbook of quantitative methods in pyschology (ed. Millsap, R. E.) Ch. 4 (Springer Publications Ltd., 2009).
Graham, J. W. Missing Data Analysis: Making It Work in the Real World. Annu. Rev. Psychol. 60 , 549–576 (2009).
Graham, J. W., Cumsille, P. E. & Shevock, A. E. In Handbook of Psychology Vol. 2 (ed. Weiner) Ch. 4 (Wiley, 2012).
Osuteye, E., Johnson, C. & Brown, D. The data gap: An analysis of data availability on disaster losses in sub-Saharan African cities. Int. J. Disaster Risk Reduct. 26 , 24–33 (2017).
Lin, Y. C., Khan, F., Jenkins, S. F. & Lallemant, D. Filling the Disaster Data Gap: Lessons from Cataloging Singapore’s Past Disasters. Int. J. Disaster Risk Sci. 12 , 188–204 (2021).
Gabrio, A., Mason, A. J. & Baio, G. Handling Missing Data in Within-Trial Cost-Effectiveness Analysis: A Review with Future Recommendations. Pharmacoeconomics. 1 , 79–97 (2017).
Mize, T. D. Source code for: Cleanplots: Stata graphics scheme. Trenton Mize https://www.trentonmize.com/software/cleanplots (2018).
Medeiros, R. A., Blanchette, D., Medeiros, R. A. & Blanchette, D. Source code for: MDESC: Stata module to tabulate prevalence of missing values. Statistical Software Components https://econpapers.repec.org/software/bocbocode/s457318.htm (2011).
White, I. Network meta-analysis. Stata J. 15 , 951–985 (2015).
Lesk, C., Rowhani, P. & Ramankutty, N. Influence of extreme weather disasters on global crop production. Nature. 529 , 84–87 (2016).
Article ADS CAS Google Scholar
Kahn, M. E. The Death Toll from Natural Disasters: The Role of Income, Geography, and Institutions. Rev. Econ. Stat. 87 , 271–284 (2005).
Noy, I. The macroeconomic consequences of disasters. J. Dev. Econ. 88 , 221–231 (2009).
Alcántara-Ayala, I. Geomorphology, natural hazards, vulnerability and prevention of natural disasters in developing countries. Geomorphology. 47 , 107–124 (2002).
Article ADS Google Scholar
Toya, H. & Skidmore, M. Economic development and the impacts of natural disasters. Econ. Lett. 94 , 20–25 (2007).
Jonkman, S. N. Global Perspectives on Loss of Human Life Caused by Floods. Nat. Hazards. 34 , 151–175 (2005).
Strömberg, D. Natural Disasters, Economic Development, and Humanitarian Aid. J. Econ. Perspect. 21 , 5–222 (2007).
Kovács, G. & Spens, K. Identifying challenges in humanitarian logistics. Int. J. Phys. Distrib. Logist. Manag. 39 , 506–528 (2009).
Barredo, J. I. Normalised flood losses in Europe: 1970-2006. Nat. Hazards Earth Syst. Sci. 9 , 97–104 (2009).
Nadim, F., Kjekstad, O., Peduzzi, P., Herold, C. & Jaedicke, C. Global landslide and avalanche hotspots. Landslides. 3 , 159–173 (2006).
Kellenberg, D. K. & Mobarak, A. M. Does rising income increase or decrease damage risk from natural disasters? J. Urban Econ. 63 , 788–802 (2008).
Deressa, T., Hassan, R. M. & Ringler, C. Measuring Ethiopian farmers’ vulnerability to climate change across regional states . IFPRI Discussion Paper 806 (International Food Policy Research Institute, 2008).
Raschky, P. A. Institutions and the losses from natural disasters. Nat. Hazards Earth Syst. Sci. 8 , 627–634 (2008).
Mohapatra, S., Joseph, G. & Ratha, D. Remittances and natural disasters: Ex-post response and contribution to ex-ante preparedness. Environ. Dev. Sustain. 14 , 365–387 (2012).
Kvaløy, B., Finseraas, H. & Listhaug, O. The publics’ concern for global warming: A cross-national study of 47 countries. J. Peace Res. 49 , 11–22 (2012).
Download references
The authors would like to thank the Université catholique de Louvain for their support on this research paper and the University of York who permitted R.L.J. to undertake part of this research as part of her Master’s dissertation. We would also like to thank the reviewers and editors at Scientific Data, who helped to shape the final version of this paper.
Authors and affiliations.
Institut de Recherche Santé Société (IRSS), Faculté de Santé Publique, Université catholique de Louvain, 1200 Woluwe-Saint-Lambert, Brussels, Belgium
Rebecca Louise Jones & Sandy Tubeuf
Institut de Recherches Économiques et Sociales (IRES), Université catholique de Louvain, 1348 Louvain-la-Neuve, Belgium
John Hopkins Bloomberg School of Public Health, Baltimore, Maryland, 21205, USA
Debarati Guha-Sapir
You can also search for this author in PubMed Google Scholar
The study was conceptualised by S.T. and D.G.S. Data analysis was performed by R.L.J. under the supervision of S.T. All authors contributed to the writing and revision of the manuscript.
Correspondence to Sandy Tubeuf .
Competing interests.
DGS founded the Emergency Events Database (EM-DAT) and contributed her expert knowledge on EM-DAT procedures and mechanics. RLJ and ST declare no competing interests.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary material table of contents, 41597_2022_1667_moesm2_esm.pdf.
Results of logistic regression analysis to test associations between the probability of data to be missing on total estimated damages and observable data in the Emergency Events Database (EM-DAT)
Results of logistic regression analysis to test associations between the probability of data to be missing on select human loss variables and observable data in the Emergency Events Database (EM-DAT)
Reference list, glossary of conventional and advanced missing data methods, rights and permissions.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .
Reprints and permissions
Cite this article.
Jones, R.L., Guha-Sapir, D. & Tubeuf, S. Human and economic impacts of natural disasters: can we trust the global data?. Sci Data 9 , 572 (2022). https://doi.org/10.1038/s41597-022-01667-x
Download citation
Received : 08 April 2022
Accepted : 26 August 2022
Published : 16 September 2022
DOI : https://doi.org/10.1038/s41597-022-01667-x
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Searching for a sustainable solution to increasing chemical pollution.
Environmental Science and Pollution Research (2024)
Natural Hazards (2024)
Theoretical and Applied Climatology (2024)
Landslides (2024)
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
You are here.
In this lesson, daily learning targets, ongoing assessment.
Universal design for learning, closing & assessments, you are here:.
Order printed materials, teacher guides and more.
How to order
Tell us how the curriculum is working in your classroom and send us corrections or suggestions for improving it.
Leave feedback
These are the CCS Standards addressed in this lesson:
Agenda | Teaching Notes |
---|---|
A. Reflecting on Module Guiding Questions (5 minutes) B. Reviewing Learning Target (5 minutes)
A. Developing Research Questions (10 minutes) B. Choosing Expert Groups (10 minutes) C. Expert Group Work: Videos of Natural Disasters (15 minutes)
A. Launching Independent Reading (15 minutes)
A. Accountable Research Reading. Select a prompt and respond in the front of your Independent Reading journal. | ). requires students to gather information from print and digital sources. As such, this lesson is designed for students to use internet sources to watch a video. Ensure the technology necessary for students to complete the research is available. ). Consider using the Independent Reading: Sample Plans if you do not have your own independent reading review routines (see the ).
). ).
|
Levels of support
For lighter support:
For heavier support:
Key: Lesson-Specific Vocabulary (L); Text-Specific Vocabulary (T); Vocabulary Used in Writing (W)
credible, affect, experience, relevant (L)
New materials.
Each unit in the 3-5 Language Arts Curriculum has two standards-based assessments built in, one mid-unit assessment and one end of unit assessment. The module concludes with a performance task at the end of Unit 3 to synthesize their understanding of what they accomplished through supported, standards-based writing.
Opening | Meeting Students' Needs |
---|---|
and remind students that in the previous lesson they were introduced to the guiding questions for the module. Review the anchor chart. and briefly review the characteristic of respect.
| |
Work Time | Meeting Students' Needs |
---|---|
and focus students on the question at the top, telling them that it will be the focus of their research in this unit: and invite students to add them to their (to cause a change in or have an impact on) (to live through)
| |
and the grouped around the room. Read each sign aloud. 1.Move to the part of the room labeled for the natural disaster you would like to study. 2.Once there, share with the group why you chose that natural disaster. | |
. |
Closing | Meeting Students' Needs |
---|---|
to launch independent reading. |
Homework | Meeting Students' Needs |
---|---|
Copyright © 2013-2024 by EL Education, New York, NY.
Help us improve our curriculum..
Tell us what’s going well, share your concerns and feedback.
Terms of use . To learn more about EL Education, visit eleducation.org
Discover the world's research
IMAGES
COMMENTS
Use notes to write report. Using their outline and notes, students now actually draft their report, including introduction, body paragraphs, and conclusion. While students may be used to writing from beginning to end, many writers find writing the body paragraphs first and then writing the introduction and conclusion a useful process. Edit report.
Floods are a type of disaster that people need to be aware of because they can have physical and psychological impacts. One of the impacts of the community on the occurrence of floods is the ...
This chapter and the preceding one use the conceptual model presented in Chapter 1 (see Figure 1.1) as a guide to understanding societal response to hazards and disasters.As specified in that model, Chapter 3 discusses three sets of pre-disaster activities that have the potential to reduce disaster losses: hazard mitigation practices, emergency preparedness practices, and pre-disaster planning ...
In an effort to provide the structure needed to develop the science associated with disaster health, four frameworks have been proposed in these Guidelines: (1) conceptual; (2) structural; (3) operational; and (4) scientific. The first three frameworks provide the common language and the deconstruction of the elements involved in or present in ...
In this article, we recall the United Nations' 30-year journey in disaster risk reduction strategy and framework, review the latest progress and key scientific and technological questions related to the United Nations disaster risk reduction initiatives, and summarize the framework and contents of disaster risk science research. The object of disaster risk science research is the "disaster ...
Download Report (PDF | 4.39 MB) Executive Summary. Compared to the average over the last 30 years (1991-2020), the total frequency of global natural disasters in 2021 was 13% higher, with 81% ...
Natural Disasters: A Resource Guide. Since the beginning of time, earthquakes, floods, tornadoes, tsunamis, wildfires, and other natural disasters have affected our entire planet. This guide provides a wide variety of online and print resources for researching all types of natural disasters. Author: Nathan Smith, Reference and Research ...
'Natural disaster' is a predefined category and is divided into six subgroups and a further 20 subtypes. All categories and definitions are predetermined by the EM-DAT as per their glossary, and the classification system by the database follows the Integrated Research in Disaster Risk Peril Classification and Hazard Glossary.
However, the instrumental use of research is easier to measure and is therefore a more obvious component of the value of research [12]. As specific natural disaster outcomes include loss of life, injury, destruction of property and disruption to infrastructure together with generalised social, economic and environmental impacts, any evaluation ...
Between 1998 and 2017, 1.3 million people were killed and another 4.4 billion were left injured, homeless, displaced, or in need of emergency assistance due to climate-related and geophysical disasters. A risk analysis of natural disasters is helpful not only for disaster prevention and reduction, but also in reducing economic and social losses. Currently, there are many methods for natural ...
The value of one life varies significantly ranging from USD 143,000 to 15 million. While an overwhelming majority of studies concern high-income countries, most disaster casualties are observed in low- and middle-income countries. Data on the human impact of disasters are usually available in disasters databases.
Di (Damage) = Vi x Li (3) The Impact of Natural Disasters: Simplified Procedures and Open Problems. 119. V is the value of the damaged element, ranging from 1 to 10 in an arbitrary scale (Figure 3 ...
Pointing the finger at natural causes creates a politically convenient crisis narrative that is used to justify reactive disaster laws and policies 9. For example, it is easier for city ...
In the model setup, this shock will immediately result in two key changes in the epicenter region; (i) a sudden rise in food prices due to lower output, and (ii), a fall in wages, due to lower ...
es in laws and regulations.This report is a resource to help community association leaders prepare for natural disasters, before an. after the disaster occurs. It provides basic information, training resources, checklists, sample documents, and relevant case studies that address dis.
Disaster assessment is the gathering and analysis of information pertinent to disasters and disaster response. The scope of the information required covers factual details of the hazard event causing the disaster, the needs of those affected, and the available resources for responding to those needs.
Summary. No person or place is immune from disasters or disaster-related losses. Infectious disease outbreaks, acts of terrorism, social unrest, or financial disasters in addition to natural hazards can all lead to large-scale consequences for the nation and its communities. Communities and the nation thus face difficult fiscal, social ...
Reliable and complete data held in disaster databases are imperative to inform effective disaster preparedness and mitigation policies. Nonetheless, disaster databases are highly prone to missingness.
Disaster management is a critical area that requires efficient methods and techniques to address various challenges. This comprehensive assessment offers an in-depth overview of disaster ...
Expert Group Natural Disaster signs by writing the name of each expert group natural disaster on a piece of paper: earthquakes, hurricanes, tornadoes, volcanoes, and tsunamis. Post in separate areas of the room. Group the Infer the Topic Resources as follows and post by the Expert Group Natural Disaster signs: Earthquakes: Resources 4, 5, 6, 17,
This research gives a real-life example of how a teacher can teach and embed the awareness about natural disasters in an uncomplicated and fun way, by using the disaster mitigation model of teaching.
Reduction (UNDRR) Human Cost of Disaster Report 2020, "In the period 2000 to 2019, there were 7,348 major recorded disaster events claiming 1.23 million lives, affecting 4.2 billion people, and
The research addressed both natural and man-made disasters. ... Identifying the effective components in the disaster risk communication gives a more comprehensive view of risk communication to the ...