Overlooked: Examining the impact of disasters and climate shocks on poverty in the Europe and Central Asia region MARCH 2021 The International Bank for Disclaimer Acknowledgments Reconstruction and Development/ This document is the product of work This report was prepared as part of The World Bank performed by World Bank staff and a regional technical assistance titled consultants. The findings, interpretations, ‘Strategies and Options for Scaling 1818 H Street, N.W. and conclusions expressed in this Up Disaster Risk Management in ECA Washington, D.C. 20433, U.S.A. document do not necessarily reflect Countries’, supported by a grant from the views of the Executive Directors of the Global Facility for Disaster Reduction March 2021 the World Bank or the governments and Recovery (GFDRR). The preparation they represent. The World Bank does of this report was led by Yann Kerblat, not guarantee the accuracy of the data Ali Arab, Brian Walsh, Alanna Simpson, included in this work. The boundaries, and Stephane Hallegatte. The team colors, denominations, and other would like to thank David Sislen and information shown in any map in this work Christoph Pusch for overall guidance and do not imply any judgment on the part insightful contributions provided by peer of The World Bank concerning the legal reviewers: Javier Baez, Sarah Coll-Black, status of any territory or the endorsement and Zoe Trohanis. or acceptance of such boundaries. The authors wish to recognize the work Rights and Permissions of many colleagues who contributed The material in this work is subject to and provided valuable comments and copyright. Because the World Bank contributions, including Julia Biau, encourages dissemination of its Alexandru Cosmin Buteica, Laurent knowledge, this work may be reproduced, Debroux, Tafadzwa Dube, Simon Ellis, in whole or in part, for noncommercial Salih Bugra Erdurmus, Ayse Erkan, purposes as long as full attribution to Abdulaziz Faghi, Chris-Philip Fischer, this work is given. Any queries on rights Elena Daniela Ghita, Corina Grigore, and licenses, including subsidiary rights, Ellen Hamilton, Carlos Hernandez, should be addressed to the Office of the Amparo  Samper Hiraldo, Niels Publisher, The World Bank: Holm‑Nielsen, Ahmet Kindap, Daniel 1818 H Street, N.W. Kull, Andrea Liverani, Michael Lokshin, Washington, D.C. 20433, U.S.A. Catalin Pauna, Joanna Mclean Masic, Fax: 202-522-2422; Nora Mirzoyan, Alice Mortlock, Ana Maria Email: pubrights@worldbank.org. Oviedo, Nicolas Pondard, Paul Prettitore, Tatiana Proskuryakova, Jamele Rigolini, Ana Simundza, Tatiana Skalon, Zuzana Suggested Citation Stanton‑Geddes, Gallina Andronova World Bank. 2021. Overlooked: Examining Vincelette, and Christoph Ungerer. the impact of disasters and climate shocks All errors, interpretations, and conclusions on poverty in the Europe and Central Asia are the responsibility of the authors. region. Washington, DC: World Bank. The report was edited by Anne Himmelfarb Cover photo: Young boy outside a and designed by Jon Walton and damaged home near Spitak, Armenia. Paperplane Pilots Pte. Ltd. Credit: Razmik Zakaryan Table of Contents 4 | Executive Summary Executive Summary When a disaster strikes, it affects not only households’ physical assets but also their income levels and ability to contribute to the local economy. These multidimensional impacts depend on both the physical characteristics of the hazard event itself (seismic magnitude, flood levels, and so on) as well as the socioeconomic characteristics of affected populations. For example, wealthier households might have access to a wide variety of coping mechanisms (for example, financial savings, housing insurance, and timely access to early warnings) which are not necessarily accessible for many poor households. Critically, these socioeconomic disparities shape not only the severity of shocks on household-level economies but also the duration of subsequent recovery and reconstruction efforts. This is because disaster effects can persist long after the physical hazards recede—for example, by forcing affected households to manage difficult trade-offs between regular expenditures (for example, food, education, health care, and so on) and longer-term costs linked to the replacement or reconstruction of assets. The effects of severe (or successive) disasters on vulnerable populations become visible when the recovery of housing structures is delayed, when school enrolment rates suddenly drop, when there is a higher exposure to public health risks, or when significant (or recurring) debts constrain reconstruction efforts. LEFT A flood defense embankment built from sandbags stands near the banks of the Sava river in Šabac, Serbia in May 2014. Credit: Srđan Popović Executive Summary | 5 Figure 1: The conventional asset-focused risk assessment approach compared to the proposed ‘Unbreakable’ model, which focuses on socioeconomic resilience. 04 01 02 03 WELL-BEING LOSSES ASSET LOSSES HAZARD EXPOSURE VULNERABILITY SOCIOECONOMIC RESILIENCE Source: Hallegatte et al. 2017. Socioeconomic characteristics assessment approach adds the The results of this analysis indicate may therefore help determine in dimension of socioeconomic that the economic well-being of advance which households (or resilience to the conventional risk citizens in these countries is affected communities) are more likely to assessment framework, which far more than the estimated cost recover faster and which are in need traditionally consists of hazard, of physical damages to buildings of external assistance to accelerate exposure, and vulnerability and public infrastructure. In the reconstruction efforts. (figure 1). Socioeconomic resilience eight countries that were analyzed, represents the ability of affected earthquakes and floods could This report explores the links households to cope with and have the most significant impacts between disaster impacts and recover from disasters. To better on 2016 poverty levels in Yerevan poverty levels in select countries understand the economic impacts (Armenia), Tbilisi (Georgia), and in the Europe and Central Asia of disasters, this report presents Bucharest (Romania). Overall, region—Albania, Armenia, a new disaster risk assessment this report demonstrates that Bulgaria, Croatia, Georgia, Greece, model that can identify the most the recovery and reconstruction Romania, and Turkey. These eight vulnerable socioeconomic groups process not only depends on the countries were selected because and assess their ability to withstand extent of physical damages due to they are all characterized by (a) disaster-related shocks, helping disasters but is also significantly relatively high levels of income develop modern, inclusive, and shaped by the economic context inequality, (b) relatively high cost-effective resilience solutions of each country and the level of levels of exposure to floods and that include investments and socioeconomic resilience of their earthquakes, and (c) significant programs outside conventional citizens. existing engagements in disaster DRM instruments such as adaptive risk management (DRM) with the social protection, financial inclusion, World Bank and other development and formal risk pooling. partners. This new disaster risk 6 | Abbreviations Abbreviations AAL Average Annual Losses AHE Aggregate Household Expenditures CRRA Constant Relative Risk Aversion DRM Disaster Risk Management ECA Europe and Central Asia ECAPOV Europe and Central Asia Poverty Data ECATSD Europe and Central Asia Team for Statistical Development ESA European Space Agency EU European Union EU-SILC European Union Statistics on Income and Living Conditions FIES Food Insecurity Experience Scale GDP Gross Domestic Product GMD Global Monitoring Database GNI Gross National Income GFDRR Global Facility for Disaster Reduction and Recovery GRADE Global Rapid Post‑disaster Damage Estimation NUTS Nomenclature of Territorial Units for Statistics PDNA Post-disaster Needs Assessment PML Probable Maximum (Asset) Loss PPP Purchasing Power Parity ROI Return on Investments UNDP United Nations Development Programme UNDRR United Nations Office for Disaster Risk Reduction CHAPTER 1: The Europe and Central Asia region faces growing disaster and climate risks The Europe and Central Asia (ECA) region1 is vulnerable to a variety of hazards, including floods, earthquakes, droughts, landslides, and wildfires. In the past 30 years, 500 significant floods and earthquakes across the region have led to roughly 50,000 fatalities and more than US$80 billion in damages (World Bank 2017). Some 20 ECA countries are estimated to have a 10–20 percent chance of being affected by a major earthquake during the next 50 years, and they could face economic losses equivalent to more than 20 percent of their gross domestic product (GDP) (World Bank 2017). In Turkey alone, more than 110,000 deaths, 250,000 hospitalizations, and 600,000 destroyed housing units were recorded as a result of earthquakes in the 20th century (Erdik et al. 2003). Floods pose the highest risk for the Baltic states, the Central European states, and the Russian Federation, whereas the Caucasus states, Southeast European states, and Central Asia tend to be more prone to earthquakes. The Europe and Central As shown in figures 2A and 2B, several major ECA cities were nearly or completely destroyed by earthquakes or floods, including Dubrovnik, Croatia Asia region is vulnerable (in 1667); Ashgabat, Turkmenistan (in 1948); Skopje, North Macedonia (in to a variety of hazards, 1963); and Bucharest, Romania (in 1977), among others. Over time, close including floods, to 30 percent of capital cities across Europe and Central Asia have been destroyed by earthquakes or floods at some point in their history (World earthquakes, droughts, Bank 2017). landslides, and wildfires. From a governance perspective, a significant number of countries in the region have faced political, economic, and social disruptions in the past 30 years. As a result, the lack of clarity on institutional roles and sectoral responsibilities has constrained the decision-making process for disaster risk management (DRM). In addition, a series of recent disasters exposed the vulnerability and lack of preparedness in the region, confirming the need for further investment in disaster risk reduction and emergency preparedness (Pollner, Kryspin-Watson, and Nieuwejaar 2008). In May 2014, for example, Bosnia and Herzegovina as well as Serbia experienced the heaviest rainfall ever recorded in the last 120 years. These massive floods in both countries resulted in devastating consequences for a number of important economic sectors, and economic damages were estimated to be up to €3.5 billion (Stadtherr et al. 2016). In Serbia, these floods caused damages and losses 1 The World Bank’s Europe and Central Asia region consists of 28 countries. In 2016, almost 489 million people lived in the region, generating an estimated regional gross domestic product of US$1.15 trillion. 8 | The Europe and Central Asia region faces growing disaster and climate risks Figure 2A: A chronology of significant earthquakes across selected countries in the ECA region over time. Year Country Earthquake name and magnitude Human and Economic Impact 1667 Croatia Ragusa Earthquake – M7.4 • 3,000–5,000 fatalities • 50 years of economic downturn • US$7.5 billion in damage 1920 Albania Tepelene earthquakes – M6.3 • 200 fatalities • 15,000–30,000 homeless 1928 Bulgaria Chirpan earthquake – M6.9 • 120–150 fatalities Popovitsa earthquake – M7.1 • 1,800+ injured • Loss of US$1.4 billion (2020 US$), equal to 7% of GDP at the time 1939 Turkey Erzincan earthquake – M7.7 • 30,000–40,000 fatalities • 100,000 injured • Over US$2 billion (2020 US$) in damage 1940 Romania Vrancea earthquake – M7.7 • 593 fatalities, • 1,271 injured 1953 Greece Great Kefalonia Earthquake – M6.8 • 445–800 fatalities • US$1.6 billion in damage (2020 US$) 1977 Romania  Vrancea Earthquake – M7.2 • 1,570+ fatalities • 11,221 injured • 200,000 homeless • US$10 billion (2020 US$) in economic losses 1978 Greece Thessaloniki Earthquake – M6.2 • 45–50 killed • 220+ injured • US$250 million–US$1 billion in damage 1988 Armenia Spitak Earthquake – M6.8 • 25,000+ fatalities • 130,000 injured • 514,000 homeless • Economic losses equivalent to US$2.7–4.8 billion (in 2020 US$), close to 1 year’s GDP at the time • 40% of Armenia affected 1991 Georgia Racha earthquake – M7.0 • 270+ fatalities • 160,000 homeless • US$7.6 billion (2020 US$) in damage 1999 Greece Athens Earthquak – M6.0 • 143 fatalities • 1,600 injured • 50,000 homeless • US$3–4.2 billion in damage 1999 Turkey Marmara earthquak – M7.6 • 17,000+ fatalities • 250,000 homeless • 43,959 injured • US$5 billion in damage 2003 Turkey Bingöl earthquak – M6.4 • 177 fatalities • 520 injured The Europe and Central Asia region faces growing disaster and climate risks | 9 Year Country Earthquake name and magnitude Human and Economic Impact 2011 Turkey Van earthquakes – M7.2–5.6 • 604 fatalities (due to several earthquake tremors) • 4,152 injured • 60,000 homeless 2019 Albania Albania earthquake – M6.4 • 51 fatalities • 913 injured • 11,490 housing units heavily damaged or destroyed • €985.1 million in damage and losses 2020 Croatia Zagreb March Earthquake – M5.5 • 1 fatality • 26 injured • 26,000+ damaged buildings (including 24,000+ damaged buildings in housing sector) • €11.3 billion in damage and losses (PDNA) 2020 Turkey and Aegean Sea earthquake – M7.0 • 116 fatalities (in Turkey) Greece • 1,035 injured • US$907 million in damage and losses (estimates) 2020 Croatia Croatia December Earthquake – M6.2 • 7 fatalities and 15 injured in Petrinja and surrounding villages • 43,000+ damaged buildings • €4.8 billion in damage and losses (PDNA) estimated at 5 percent of its GDP, (of which 1,900 are unusable) and a intrinsic link between poverty pushed 125,000 people under the total of 27 people were injured, of impacts and socioeconomic poverty line, and ultimately led to whom 1 succumbed to her injuries vulnerability, with the lack of local the loss of 52,000 jobs (Government (Bogaerts 2020). The total cost from capacity to manage disaster risks of Serbia et al. 2014). In November this earthquake is estimated at €11.3 in a sustainable manner. 2019, Albania was struck by a billion (World Bank 2021c). strong 6.3 magnitude earthquake, The COVID-19 pandemic which resulted in €844 million Such shocks not only roll back exacerbated the effects of in estimated damages as well as development gains and disrupt natural disasters in 2020, by 51 fatalities and displacement of livelihoods of vulnerable residents degrading micro-economic and 17,000 people. The earthquake in hazard-prone areas, they also macroeconomic coping mechanisms ultimately affected more than 1.9 marginalize vulnerable communities and responses. With the direct million (out of a total population of by pushing them into deeper effects of the pandemic still 2.8 million) and caused extensive poverty and exacerbate existing largely unresolved, it is difficult to damage across 11 municipalities, income inequalities in affected estimate its indirect consequences including the two most populous countries and territories. These on contemporaneous floods and and urbanized cities, Tirana and recent disasters confirm that many earthquakes, but what is already Durrës (Government of Albania countries in the ECA region are known is significant. Clearly, the et al. 2020). More recently, in inadequately equipped to manage pandemic complexified coordinated March 2020, Zagreb, the capital disasters of such magnitude and responses at all scales and may and largest city of Croatia, was hit highlight the undeniable link have reduced the surge capacity by a 5.5 magnitude earthquake, between poverty and vulnerability of some local health care systems the strongest earthquake to hit on the one hand and poor governance in their disaster responses. In the city since 1880. More than in disasters on the other (UNDP countries that did not institute 26,000 buildings were damaged 2016). They also reemphasize the robust emergency cash transfers, 10 | The Europe and Central Asia region faces growing disaster and climate risks Figure 2B: A chronology of significant floods across selected countries in the ECA region over time. Year Country Period Human and Economic Impact 1926 Romania January 1926 ~1,000 fatalities 1970 Romania May–June 1970 200–215 fatalities; at least US$3 billion in damage 1992 Albania November 1992 11 fatalities; US$12 million in damage 1994 Greece October 1994 11 fatalities; US$700 million in damage 1997 Georgia April 1997 2 floods caused 7 fatalities and over US$40 million in damage 1997 Armenia June 1997 4 fatalities; US$12 million 1997 Romania June–August 1997 US$310 million in damage 1998 Romania June 1998 US$150 million in damage 1998 Turkey August 1998 1 million people affected; over US$1 billion in damage 2002 Albania September 2002 1 fatality; US$23 million in damage 2003 Greece January–February 2003 US$800 million in damage 2005 Bulgaria August 2005 30 fatalities; US$600 million in damage 2005 Romania August 2005 3 floods caused 60 fatalities; almost US$2 billion in damage 2006 Romania February–April 2006 economic damage over 1% of GNP; 160 localities affected 2009 Turkey September 2009 31 fatalities; US$600 million in damage 2010 Albania January 2010 US$51 million in damage 2010 Romania June–July 2010 23 fatalities; US$1 billion in damage 2012 Georgia May 2012 100,000 people affected; US$3 million in damage 2013 Georgia July 2013 Close to 25,000 people affected 2014 Greece December 2014 24 fatalities 2014 Bulgaria May 2014 15 fatalities; US$400 million in damage 2014 Croatia May 2014 3 fatalities; 9,000 people affected 2015 Georgia June 2015 19 fatalities; over US$20 million in damage The Europe and Central Asia region faces growing disaster and climate risks | 11 widespread, severe, and sustained shocks to labor earnings depleted many households’ precautionary savings. Due to evictions and foreclosures, fears of loss of housing are widespread, leaving populations more exposed and vulnerable to the elements. The COVID-19 pandemic exacerbated the effects of natural disasters in 2020, by degrading micro- and macroeconomic coping mechanisms and responses. At a macroeconomic scale, substantial social assistance packages, productivity and tax A health official in Bucharest, Romania treats COVID-19 patients in June, 2020. Credit: M. Moira losses during initial quarantines, and subsequent rapid shifts in consumer behavior have depleted examines not only their respective of these analytics is to (a) inform many sources of contingent level of exposure to earthquake DRM interventions that are financing and increased deficits, and floods but also their ability socially inclusive and that reflect leading to larger debt and lower to cope with such shocks, that in-country needs and context in ability to respond to future shocks. is, their level of socioeconomic the coming years and (b) facilitate The considerations described in resilience. These eight countries the prioritization of sub-national this report affect the efficiency of were selected because they are measures and related investments compounding crisis responses, all characterized by (a) relatively that aim to strengthen the level of targeting poverty reduction high levels of income inequality, socioeconomic resilience across to hasten recovery and shared (b) relatively high levels of exposure the region. prosperity. to floods and earthquakes, and (c) significant existing engagements Given the growing influence of in DRM with the World Bank and other disaster and climate shocks on development partners. socioeconomic activities in the ECA region, this report presents a Building on the Unbreakable disaster risk assessment model to model (Hallegatte et al. 2017), illustrate how earthquake and flood this disaster risk assessment risks affect different sub-national approach therefore adds a new entities and income groups. The dimension— socioeconomic analysis focuses on eight countries resilience—to the conventional risk in the ECA region—Albania, assessment framework—which Armenia, Bulgaria, Croatia, Georgia, consists of hazard, exposure, and Greece, Romania, and Turkey—and vulnerability. The overall objective 12 | The Europe and Central Asia region faces growing disaster and climate risks Search and Rescue efforts in Izmir (Turkey) during the October 2020 Aegean Sea earthquake. Credit: hasanucarphotography/ Shutterstock.com CHAPTER 2: Climate change, urbanization, and socioeconomic vulnerability are the s main risk drivers region’ 2.1 Climate change projections for Europe and Central Asia Climate change by raising temperatures and changing hydrology is expected to produce more frequent, intense disaster events in the ECA region. Increased aridity and widespread wildfires will result in the loss of vegetation, while prolonged periods of intense rainfall and snowmelt will increase the frequency of massive mudslides, rockslides, and debris flows. In parts of Central Asia and the Western Balkans specifically, unprecedented heat extremes could occur in over 60 percent of summer months and drought risk could increase by 20 percent in a 4°C warmer world. At the same time, projections suggest an increase in riverine flood risk, mainly in spring and winter, due to more intense snowmelt in spring and heavier rainfall in the winter months. Reduced water availability in some places will become a threat as increases in temperatures head toward 4°C (World Bank 2014a). Recent destructive floods, Climate change by worsening wildfires, and devastating heat waves (box 1 and figure 3) are raising temperatures already becoming more common throughout the region, illustrating to policy makers the need to prioritize DRM efforts. and changing hydrology, is expected to produce Climate change impacts will vary from region to region, but Central Asia, the Western Balkans, and the South Caucasus are often identified as the more frequent, intense most vulnerable hotspots in the region, yet they are least ready to adapt to disaster events in the consequences of climate change (World Bank 2014a). Climate shocks will ECA region. have an adverse effect on the GDP of each country and affect a significant number of sectors, most notably agriculture and forestry. More importantly, they will disrupt the lives and livelihoods of people. Melting glaciers and warming temperatures will shift the growing season and the flow of glacier- fed rivers further into spring in Central Asia, while in the Western Balkans and the South Caucasus, worsening drought conditions will put crops at risk, with potential declines for urban health and energy generation. Decaying infrastructure, unsustainable and inefficient land and water management practices, and aging populations will increase sensitivity to climate change, while socioeconomic vulnerabilities and weaker institutions will put pressure on local capacity to adapt to increasing climate impacts (World Bank 2014a).2 2 Today, urbanization in Eastern European and Central Asian countries is profoundly affected by demographic transition. Compared to the rest of the world, countries in the region have much lower population growth rates and are among the only countries experiencing a decline in both their total population and their urban population (Restrepo Cadavid et al. 2017). 14 | Climate change, urbanization, and socioeconomic vulnerability are the region’s main risk drivers Figure 3: A heatmap of Europe for July 25, 2019. Climate change is increasing constructed before modern building the frequency and severity of extreme weather events across Europe, such as codes. Moreover, disaster risks are this recent summer heatwave breaking temperature records in 7 countries. also exacerbated by the ongoing impacts of climate change. While poorly maintained flood protection infrastructure and insufficient weather forecasting systems mean that even small‑scale events such as river floods or landslides can result in major disasters,4 the concentration of assets, services, and populations in urban areas also increases exposure to weather extremes such as floods or heat waves.5 For instance, informal settlements on floodplains and steep hillsides in the Western Balkans have been severely affected by floods and landslides in recent years (World Bank 2014a). Old building stocks create a particular challenge, especially in high-density cities exposed to seismic risk. In many ECA countries, one of the key housing-related Source: ESA 2019. © ESA. Contains modified Copernicus Sentinel data (2019), processed by ESA. challenges is the seismic threat Licensed under CC BY-SA 3.0 IGO. faced by certain building types— for example, pre-1990s multifamily apartment buildings—which may be Recent decades have also seen record-breaking heatwaves) caused beyond their design lifespan but are record-breaking meteorological economic losses in excess of €511 not being replaced due to limited events in the ECA region such billion (71 percent of which were demand for urban renewal.6 The as heat extremes (particularly in uninsured) and more than 91,000 engineering problem of building urban areas) or unseasonal rainfall casualties over the last four decades resilience into public and private patterns (see table 1). Examples (EEA 2019). 3 infrastructure is also layered with include the Western European heat complex political, economic, social, wave during the 2003 summer, legal, and financial realities, and with a death toll estimated at 2.2 Deficient the sheer scale of the undertaking around 70,000 (see box 1), or the infrastructure and urban explains why it has been so difficult heat wave of the 2010 summer in settlements in transition to address since the last century. Eastern Europe and Russia, with While many poor people will be an estimated death toll of 55,000. In Europe and Central Asia, living in isolated, rural areas, More recently, in 2017 and 2019, disaster risks are mainly due to the Mediterranean region also aging and poorly maintained public 4 Floods and landslides are a significant endured severe heat waves with infrastructure such as roads, schools, problem in almost all countries in the region. record-breaking temperature peaks and hospitals, which were typically 5 As recently seen with the unprecedented heat waves, Europe is the region with the in July and August (Guerreiro et al. highest average temperature for 2011–2015. 2018). These patterns also confirm 6 Another important challenge is the assumed that climate and disaster risks have 3 In the EU member states (EU-28), disasters excessive cost that is associated with seismic caused by weather- and climate-related strengthening (or rebuilding and relocation) been on the rise (see table 1). In extremes accounted for some 83 percent and the complex systems that regulate with the European Union (EU), disasters of the monetary losses over 1980–2017. multi-unit dwellings (and related insurance Weather- and climate-related losses policies), this includes for example the weak and climate shocks (including amounted to €426 billion (at 2017 values). institutionalization of with Homeowner See EEA (2019). Associations (Mathema and Simpson 2018). Climate change, urbanization, and socioeconomic vulnerability are the region’s main risk drivers | 15 Table 1: Selection of record-breaking meteorological between 2000 and 2015, their societal impacts, and confidence level that the meteorological event can be attributed to climate change. Region (Year) Meteorological record-breaking event Impact, costs Level of confidence in attribution to climate change England and Wales (2000) Wettest autumn on record since 1766; £1.3 billion Medium several short-term rainfall records Europe (2003) Hottest summer in at least 500 years Death toll exceeding 70,000 High England and Wales (2007) May to July wettest since records Major flooding causing £3 billion Medium began in 1766 Southern Europe (2007) Hottest summer on record in Greece Devastating wildfires Medium since 1891 Western Russia (2010) Hottest summer since 1500 500 wildfires around Moscow, Medium crop failure of ~25%, death toll estimated at 55,000, US$15 billion in economic losses France (2011) Hottest and driest spring on record Grain harvest down by 12% Medium in France in France Western Balkans (2014) Heaviest rainfall recorded in last €3.5 billion in damages in Serbia Medium 120 years and in Bosnia and Herzegovina; at least 86 casualties Adapted from World Bank 2012 and Stadtherr et al. 2016. increased rural-to-urban migration damage or collapse. For example, and continued urban expansion Because employment in many EU countries, housing into hazard-prone areas in several inequality is a key driver of uneven countries in the ECA region means is often the main climate impacts—and that applies as that a growing proportion of urban source of income and much to heating (box 1) as to cooling populations will be at risk of climate- (box 6). The devastating impact of precautionary savings related extreme events and will face hazard and climate events is not just rising food and energy prices (box 1; for emergencies, access the result of the presence of risks; table 1), thus increasing poverty to employment is a key it also depends on the capacity of levels among urban population individuals to cope with such shocks groups (World Bank 2014a). factor when considering and the capacity of surrounding vulnerability of the communities to respond or mitigate their effects (UNDP 2016). Given 2.3 Sources of affected population. their limited financial resources, socioeconomic households living in poverty are is, economically vulnerable likely to have difficulty in coping with vulnerability people who do not have access to sudden heat waves or droughts, The adverse impacts of disasters adequate income and whose human relocating or evacuating in the event and climate change in the ECA development opportunities are of a torrential flood, or absorbing region mostly affect poor and lower than average—are poor not unseasonal spikes in the price of low-income communities, given only in terms of basic resources, food or energy. their increased exposure and they also tend to live in hazardous their underlying socioeconomic areas or in poorly constructed or ill- Because employment is often vulnerability. Poor people—that maintained houses that easily suffer the main source of income A girl cools off in fountain during a heatwave in Athens, Greece on Aug. 8, 2017. Credit: Alexandros Michailidis. Box 1: Recent heat waves had disproportionate impact on poorer, socially isolated, and aging population groups in the EU. With a death toll estimated at 70,000 socioeconomic vulnerabilities enabled Older tower blocks, which typically and losses estimated to exceed €13 decision-makers to come up with house low-income tenants in the billion, the heat wave of 2003 was targeted solutions by strengthening United Kingdom, France, and Eastern one of the 10 deadliest disasters in health surveillance (monitoring Europe, are particularly prone to Europe in the last 100 years and the admissions in emergency wards) overheating. Compared to low-income worst recorded in the last 50 years.a and environmental surveillance and vulnerable households, wealthier Yet heat waves (periods of anomalous (meteorological data) and also by population groups are better protected warmth) do not affect everyone reviewing national and local action from extreme weather because they in the same way: vulnerable and plans to clearly identify agencies have better access to clean water low-income population groups will responsible for heat wave issues. supplies, cool environments, and air experience more of their effects than conditioning equipment and are able other groups due to social isolation, In general, poorer neighborhoods and to temporarily relocate. Several studies the presence of elderly populations, informal settlements are more likely also demonstrate that the poorer a preexisting health conditions, to be exposed to heat waves and other household or a neighborhood is, the lack of geographical mobility, and environmental risks than better-off more exposed it will be to urban heat economic disadvantage. For example, and formal settlements due to their island effects (World Bank 2020a).b in France, both older people from lower-quality infrastructure, unsuitable poorer socioeconomic groups and building designs, and lower access to Sources: Chakraborty et al. 2019; Euronews 2019; Joyce 2019; McGregor et al. 2007; Michelon, people with more limited social essential services (health infrastructure Magne, and Simon-Delavelle 2005; Saheb 2018, networks were disproportionately and so on). Europe’s poor have 2019; UNEP 2004 ; World Bank 2020a. impacted by the 2003 heat wave. historically been concentrated in older, a. In the summer of 2003, the unprecedented This health crisis, without precedent cheaper, and poorly built housing, heat wave that affected Western Europe raised summer temperatures 20–30 since the Second World War, has such as small units in inner‑city percent higher than the seasonal average in had serious repercussions and has apartment buildings. Because they Celsius degrees. led the French government to take have less choice in where they can b. Urban heat islands are city areas where various steps to limit the effects of any buildings and sealed surfaces trap live, they tend to rely on homes and reradiate heat so that nighttime future heat waves on public health. close to workplaces and affordable temperatures do not drop as they do in rural A better understanding of underlying and green areas. amenities, often in industrial areas. Climate change, urbanization, and socioeconomic vulnerability are the region’s main risk drivers | 17 A community’s level of resilience (or its lack of resilience) depends on a range of structural vulnerabilities inherent within social groups, which can affect access to protective assets, to essential emergency information, and to strong social networks. and precautionary savings for emergencies, access to employment is a key factor when considering vulnerability of the affected population. Unemployment continues to be a problem in many countries in the ECA region. The absence of jobs also means that social programs are often the only source of income and thus the ultimate safety net for entire households. The unforeseen loss of employment Strengthening the structure of Sisli Vocational High School in Istanbul, Turkey to be more and the absence of temporary earthquake resistant. Credit: Simone D. McCourtie opportunities can also marginalize vulnerable groups into deeper poverty and affect their financial disparity generate resilience on the effectiveness of policies resilience. For instance, in the barriers and are likely to make and measures usually put in place aftermath of the Balkan floods communities and social groups by local authorities (risk-informed of 2014, people with disabilities significantly more vulnerable to land use planning, building code could not return to their previous shocks over time such as disasters. enforcement, and so on). With jobs and faced difficulties in This is true for both sudden-onset limited access to formal insurance, accessing workplaces given that and slow-onset disaster events. lower incomes, lower-quality recovery efforts were delayed in A community’s level of resilience assets, and lower levels of human certain areas, and displaced Roma (or its lack of resilience) also development, poor households have populations lost their informal and depends on a range of structural to manage such shocks under highly seasonal job opportunities (street vulnerabilities inherent within social constrained conditions. Damages vending, for example) and were groups, which can affect access to induced by disasters therefore not eligible for social transfers due protective assets (flood-protection exacerbate the preexisting to the absence of documentation barriers, earthquake-resilient socioeconomic inequalities, and preexisting socioeconomic schools), to essential emergency disrupt livelihoods, and have the marginalization (UNDP 2016). information (evacuation orders, potential to roll back long‑term early warning notifications) and to development gains. Qualitative evidence within the strong social networks (remittances region suggests that persistent coming from abroad, financial poverty and entrenched income reserves). Finally, resilience depends A man in Serbia wades through floodwaters to assist his neighbors. Credit: Vladm CHAPTER 3: Disaster effects can be better managed with a resilience-informed analysis looking at households The proposed risk assessment framework builds on the model from the Unbreakable report (2017) and seeks to provide more in-depth insights into second-order effects of disasters than traditional risk assessment. In particular, it shows how regions (or specific geographical areas) that are prioritized for DRM interventions can differ depending on which risk metric is used since each metric informs a different set of policy objectives. Moreover, focusing on average losses or aggregate post-disaster damage figures does not illustrate the distributional effects of disasters, especially given the significant level of socioeconomic inequalities in the ECA region. Instead a focus on factors of poverty increases (such as well-being losses) provides a more granular perspective by broadening the analysis to better represent the interest of all affected income groups and yield benefits with regard to poverty reduction, shared prosperity, and other sustainable development goals. To enable a more socially inclusive approach to disaster risk management, such The proposed risk post-disaster second-order effects therefore need to be aptly estimated and better incorporated in current DRM practices. assessment framework builds on the model Disasters have complex and diverse consequences that can be measured and, increasingly, anticipated. Relevant metrics in this study include recovery from the Unbreakable time, economic losses at household level (which include both income and consumption), poverty incidence, and welfare (or well-being losses). Each of report (2017) and these metrics provides a different perspective on the costs induced by a given seeks to provide more disaster. In contrast to risk analyses centered on direct damages, many of in‑depth insights into these metrics show how disasters disproportionately affect poor households because such income shocks constrain the poor to make difficult decisions second‑order effects of between immediate needs (such as food and health care) and recovery needs. disasters than traditional As a result of these trade-offs, poor households take longer to recover from disasters and are more likely to face long-term consequences. Such costs risk assessment. are not included in traditional risk assessments that measure the severity of disasters through their direct damages or by looking at the replacement cost of assets damaged or destroyed by a disaster. Moreover, looking at aggregated data or average post-disaster damage figures may not be insightful in certain instances given the significant level of inequalities in the region. Other side effects resulting from disasters—such as the impacts on health, education, or quality of life—have been more difficult to incorporate into disaster loss estimates or cost-benefit analysis of possible disaster risk reduction interventions. 20 | Disaster effects can be better managed with a resilience-informed analysis looking at households One implication of the use of asset prioritization makes sense from a tools (for example, social transfers, losses as a measure of disaster purely monetary perspective, asset- formal and informal post-disaster impacts is that DRM strategies tend focused strategies disincentivize support, savings, insurance, and to favor the wealthy population attractive investments in the access to credit). Households groups, central business districts, poorest areas, even when small that lack access to these tools will and other clusters of valuable interventions could significantly struggle to cope with shocks and assets. Risk analysis solely based reduce the effects of disasters on could fall into chronic poverty as a on aggregated asset losses tends the economic well-being of the result (Carter and Barrett 2006). In to drive risk reduction investments population. short, a complete focus on asset toward zones where infrastructure, losses obscures the role of poverty facilities, and private assets are At a macroeconomic level, asset reduction as a tool to reduce concentrated. Yet, that type of losses obscure the relationship disaster impacts and reduces the analysis does not demonstrate between vulnerability and impact of DRM strategies when how asset losses disrupt people’s development. Economic growth considering larger development livelihoods and living conditions tends to increase disaster losses agendas (Hallegatte et al. 2017). at household level, especially for as measured in asset losses the poorest income groups and (Hallegatte et al. 2017; Kahn 2005; To provide a more comprehensive the most vulnerable.7 While this Schumacher and Strobl 2011), but measure of disaster impacts, development and higher incomes the World Bank’s Unbreakable also make people more resilient: report introduced (Hallegatte 7 In addition, such second-order effects of the long-term impacts of disasters et al. 2017)—and subsequent disasters—particularly the socioeconomic consequences on poor populations—are on communities’ well-being and studies (for example, Hallegatte, often not incorporated as part of the recovery prospects depend not only on direct Rentschler, and Walsh 2018; process or considered in cost-benefit analysis to calculate the return on investments of impacts (asset losses) but also on Markhvida et al. 2020) developed— complex infrastructure projects (for example). the accessibility of existing financial the concept of well-being losses Figure 4: The conventional asset-focused risk assessment approach compared to the proposed ‘Unbreakable’ model, which focuses on socioeconomic resilience. 04 01 02 03 WELL-BEING LOSSES ASSET LOSSES HAZARD EXPOSURE VULNERABILITY SOCIOECONOMIC RESILIENCE Source: Hallegatte et al. 2017. Disaster effects can be better managed with a resilience-informed analysis looking at households | 21 and socioeconomic resilience. To consider the different abilities of the poor and the nonpoor to cope with the asset losses in the aftermaths of a disaster, the effect of asset losses on income (accounting A complete focus on asset losses obscures the role of poverty reduction as a tool to reduce disaster impacts and reduces the impact of DRM strategies when considering larger development agendas. for capital productivity) and then on consumption (accounting for diversification of income, social protection, and post-disaster transfers) are modelled. Well- A woman sits by the side of the road outside of Skopje, FYR Macedonia. being losses are therefore the Credit: Tomislav Georgiev/World Bank equivalent loss of consumption for a given population: if the analysis loss than are wealthier individuals. of factors that contribute to finds that a disaster causes US$1 Socioeconomic resilience measures higher socioeconomic resilience million in well-being losses for a the ability of a given economy’s can provide actionable insights given population, it means that ability to absorb the impact of well- and encourage decision-makers the impact of the disaster on being losses as a consequences of to incorporate socioeconomic well-being is equivalent to a US$1 asset losses (and is calculated as resilience-informed considerations million decrease in their level the ration of expected asset losses as part of broader DRM of consumption (and therefore to well-being losses.) An overview interventions both spatially (which the share of consumption in the of the Unbreakable framework sub-national areas should be country’s GDP), perfectly shared (compared to the traditional risk prioritized?) and at a sector level among the population (Hallegatte assessment framework) is shown in (which intervention makes more et al. 2017). Well-being losses figure 4. sense?). Compared to previous therefore incorporate people’s disaster risk analytics in the ECA socioeconomic resilience, including The analysis presented in the region,8 this study aims to provide (a) their ability to maintain their remaining chapters of this report more in-depth insights focusing consumption for the duration of seek to examine the level of on the second-order effects of their recovery, (b) their ability to socioeconomic resilience in select disasters (asset losses being save or borrow to rebuild their asset countries in the ECA region in the first-order consequences) in stock, and (c) the decreasing returns the face of disasters (specifically the following countries: Albania, in consumption—that is, the fact earthquakes and floods). A more that people who live on US$2 per comprehensive understanding day are more affected by a US$1 8 See World Bank (2016) for a detailed example. Debris from the 2015 Tbilisi flood in Georgia. Credit: HeavyDPJ Armenia, Bulgaria, Croatia, Georgia, Monitoring Database (GMD) 9 to is consistent with the granularity of Greece, Romania, and Turkey (in model the distributional impacts the reported administrative level alphabetical order). These eight of disasters across households for the households in each country countries were selected because and various income groups.10 The available in the GMD datasets. This they are all characterized by (a) level of granularity in the analysis distributional analysis is intended relatively high levels of income to enhance existing studies, by inequality, (b) relatively high levels of providing more spatially detailed 9 The GMD is an ex post harmonization exposure to floods and earthquakes, effort developed by the World Bank based risk assessments that can inform the and (c) significant existing on available multitopic household surveys, development of DRM interventions including household budget surveys and engagements in DRM with the a Living Standards Measurement Survey. and related resilience-building World Bank and other development The main purpose of the GMD is to create measures that are more socially a globally comparable harmonized micro- partners. For this study, this analysis database across countries, regions, and survey inclusive and better targeted. combines data generated from years for the purpose of creating, monitoring, and tracking international poverty and shared exceedance probability curves prosperity indicators as well as data sortable (as described in the previous by groups such as gender, age, employment status, and urban/rural. chapter) with household data 10 The full methodology and detailed model from the World Bank’s Global description are provided in the technical appendix from a previous study focusing on the Philippines (Walsh and Hallegatte 2019). CHAPTER 4: Estimated asset losses due to disasters show significant disparity among regions In Europe and Central Asia, both rural and urban communities are exposed to regular and potentially destructive floods and seismic activity—which poses unique challenges to their economic development. Progress on poverty reduction and shared prosperity—the World Bank’s twin goals—has also slowed down since the 2008 global financial crisis: 60 million people in the region are still poor and live on US$5.50 or less per day, and nearly 70 percent of the population in lower-middle-income countries are vulnerable and live on US$5.50–US$11 per day. Given this context and with the increase of climate risks, effective DRM strategies are essential for building resilience and protect development gains from such shocks. To reduce the adverse effects of disasters, the three essential elements of risk, that is, hazard (earthquake and flood event in this study), exposure (the value of natural and built assets that might face a destructive event), and vulnerability (the expected consequences to exposed assets when a destructive event occurs) need to be identified, measured, and managed. Combined, these In Europe and Central three dimensions help understand the potential extent of disaster risk11 and estimate the average annual asset losses in a given geographical location. Asia, both rural and With access to comprehensive and reliable risk information, policy makers urban communities are and at-risk communities can start to better understand the potential impacts exposed to regular and of hazards and carry out risk-informed planning and investment to anticipate future disasters. potentially destructive floods and seismic The terms ‘loss’ and ‘damage’ are often used interchangeably in reference to the adverse impacts of disasters on society, economy, and the environment activity—which pose (GFDRR 2014). Direct economic asset losses—defined as the monetary unique challenges value of physical damage12 to assets located in the affected area—primarily to their economic emerge during the disaster event or within the first few hours after the event and are often assessed soon after the event to estimate recovery cost and development. determine insurance payouts (UNDRR 2018). These are tangible, relatively easy to measure, and frequently used to estimate the severity of disasters and quantify disaster risks over time (for example by comparing the impacts of disasters across different time periods and/or geographical areas). As a result, direct economic asset losses have become the main financial indicator 11 The potential loss of life, injury, or destroyed or damaged assets which could occur to a system, society or a community in a specific period of time, determined probabilistically as a function of hazard, exposure, vulnerability and capacity (UNDRR 2018). 12 Direct economic asset loss is nearly equivalent to physical damage. Damage is a generic term without quantitative characteristics, which does not mean that damage cannot be measured and expressed as a loss. The damage to a roof, for instance, can be translated into monetary terms (the cost of repairs), which in turn can be incorporated as part of loss inventories overtime (GFDRR 2014). 24 | Estimated asset losses due to disasters show significant disparity among regions Figure 5: Multicountry comparison of risk to assets from earthquake and floods (in GDP percentage). to monitor and track disaster risk To quantify the total asset losses country, for various frequencies or reduction achievements in the in a given geographical area,14 return periods.16 For the sake of context of the 2015–2030 Sendai an important input to the model simplicity, the model was assumed Framework for Disaster Risk adopted in this study is obtained that a disaster affects only one Reduction (United Nations 2015).13 from exceedance probability region at a time. curves,15 which provide the probable In anticipating such hazards, it maximum (asset) loss (PML) for Figure 5 displays a summary of is essential to account for direct earthquakes and floods, in each average annual asset losses per and indirect losses on public administrative unit in a select year due to earthquakes and floods infrastructures and private assets for eight countries in the ECA and incorporate their impacts at region (which will be the focus of household level. For a given disaster 14 For this study, the NUTS 2 administrative the chapters below)17. These values unit was used for Albania, Armenia, Bulgaria, event with direct economic asset Georgia, Greece, Romania, and Turkey. represent the expected long- losses, while a poor household Due to lack of NUTS 2 level household term average replacement cost of information in our datasets, NUTS 1 might be affected by long-lasting administrative unit for Croatia was used in household’s private and income- impacts and resort to coping this study. The Nomenclature of territorial generating assets damaged or units for statistics, abbreviated NUTS, is a strategies, a wealthier household geographical nomenclature subdividing the destroyed by floods and earthquake might be in a better position to economic territory of the EU into regions at three different levels (NUTS 1, 2, and 3, mobilize resources to weather respectively, moving from larger to smaller this shock and promptly recover. territorial units). Above NUTS 1, there is the 16 For each region, the input data detail the ‘national’ level of the member states. total value of assets lost due to hazards as 15 This model uses exceedance probability well as the frequency of each type of disaster curves based on historical data on floods and over a range of magnitudes. Magnitudes are 13 For example, loss information can be earthquakes from, respectively, D. Guha-Sapir, expressed in terms of total asset losses. For harnessed for, and integrated into, risk R Below, and Ph. Hoyois, EM-DAT: International example, the curves specify “An earthquake assessments as part of efforts to promote Disaster Database (Universite Catholique that causes at least US$X million in damages community resilience. Loss and hazard de Louvain, Brussels, Belgium), www.emdat. in Y region is, on average, expected to occur profiles can inform land-use planning, be, and Daniell and Schaefer 2014. Damage once every Z years.” The following are return zoning, and development decisions; local estimates for all historical events have been periods in this study: 1 year, 2 years, 5 years, ordinances on building codes and housing inflated to 2015 US$. These exceedance 10 years, 20 years, 50 years, 100 years, 200 density; taxation and budget decisions; and probability curves provide the probable years, 250 years, 500 years, 1,000 years, policy setting at local to national levels. A maximum (asset) loss (PML) for earthquakes 2,000 years, 5,000 years, and 10,000 years. sound understanding of the drivers and and fluvial and pluvial flooding), each 17 Due to the significant impact of earthquakes causes of losses, as well as their societal, administrative unit in the country, and various and floods on the annual average of affected environmental, and economic implications, frequencies or return periods from 1 year to GDP in these eight ECA countries (according enables communities to manage hazards 10,000 years. For more detailed information to their country disaster risk profiles), the and disasters proactively rather than about the datasets used, the assumptions and focus of this study concentrated on these reactively (GFDRR 2014). the methodology, see Annex 2. two types of hazards. Estimated asset losses due to disasters show significant disparity among regions | 25 Destroyed homes in Krupanj, Serbia following severe mudflows in 2014. Credit: Zoran Dobrin each year in a given geographical an incomplete picture of the total area. These losses consist of the economic cost of a given disaster Direct economic cost to repair or replace damaged event. This is in part due to the fact asset losses primarily assets including homes, vehicles, that direct and indirect economic roads and bridges, factories, losses vary from the immediate to emerge during the and so on. As shown, long-term delayed losses following a disaster. disaster event or within average annual losses in GDP Indirect economic losses are the percentage due to earthquakes for subsequent or second-order effects the first few hours the countries in this study range of the initial destruction such as after the event and from 0.05 percent in Bulgaria to lost wages, business interruption are often assessed 0.48 percent in Georgia. For floods, losses due to disruption of supply the range is wider, between 0.04 chains, interruption of basic soon after the event percent (Greece) and 0.97 percent services (health, education), and to estimate recovery (Croatia). The total annual expected disruption caused by temporary or asset losses due to floods and permanent relocation, among other cost and determine earthquakes in GDP percentage are things. Indirect economic losses insurance payouts. more than 1.39 percent in Croatia can occur inside or outside the and 1.14 percent in Georgia. hazard area and often have a time lag, and, as a result, they become However, as illustrated in intangible or difficult to measure subsequent chapters, direct in a consistent manner (UNDRR economic asset losses provide 2018). Moreover, it is worth noting 26 | Estimated asset losses due to disasters show significant disparity among regions concentration of valuable assets productive assets in the capital city, As illustrated in is and therefore where potential in Armavir, high asset losses due to subsequent chapters, losses are the largest. For example, earthquakes lead to outsized well- as shown in figure 6 (left side), the being losses. Looking at anticipated direct economic asset expected average annual losses post-disaster recovery time at a losses provide an due to earthquake and flood risk in sub-national level can also provide Armenia are concentrated around useful insights for decision-makers, incomplete picture of the capital city, Yerevan. Due to for example, by predicting the most the total economic cost its elevated hazard and exposure, disadvantaged regions during the of a given disaster. more than 30 percent of annual aftermath of a major earthquake losses are expected to occur in and better inform the allocation the capital region, Yerevan. In an of relief efforts to support post- that another disadvantage with the average year in Armenia, flooding disaster reconstruction efforts. For use of direct economic losses as a causes 59 percent less damage than instance, it would take Shirak and metric is that it cannot be used to earthquakes, with losses valued at Aragatsotn almost three times as measure many of the resilience- US$22 million per year. Flood risk long as Yerevan to recover 75 percent building benefits associated with is slightly more distributed than of damaged/destroyed assets in the the three specific Sendai Priorities earthquake risk and is particularly aftermath of a 200-year earthquake. for Action: strengthening disaster concentrated in the Gegharkunik DRM strategies seeking to minimize governance, investing in resilience, region (29 percent) and Yerevan time to recover after disasters should and enhancing preparedness for (25 percent). therefore focus on providing timely effective response (Markhvida et post-disaster support in areas where al. 2020). Well-being losses and post-disaster recovery is expected to be slower. recovery times are also not uniformly These different maps highlight the Natural hazard risk is also not distributed at sub-national level. importance of not using averages equally distributed across the A slightly different pattern emerges and demonstrate that while a region or across individual when one looks at the distribution of sole focus on asset losses already countries. Significant differences well-being losses or how long each provides a robust set of priorities, across regions are explained by region would take to recover in the it still does not capture how severe variations in the hazard, exposure, aftermath of a major earthquake socioeconomic consequences and vulnerability of each part of the in figure 6 (right side). In Armenia, can be at a local level or help us country. In most cases, asset risks while Yerevan suffers large losses identify which region is more likely are near major economic centers due to the density of exposed to struggle in the aftermath of a because that is where the high infrastructure, facilities, and other disaster. Figure 6: Multihazard (earthquake and precipitation flooding) risk to assets, by region in Armenia, and compared with maps revealing which regions are least likely to cope (and which regions are likely to face challenges with post-disaster reconstruction and recovery). SHIRAK ARAGATSOTN YEREVAN 1 12 23 2 8 53 20% 31% 43% 4 6 9 Socioeconomic resilience for Average number of years needed to recover Expected average annual losses Regional wellbeing losses in millions (US$) floods and earthquakes 75% of assests (200 year earthquake event) (floods and earthquakes) in millions (US$) Regional average = US$21 million Results are expressed in US$, millions. CHAPTER 5: Disasters increase existing poverty levels As seen in recent events in the region, when a major disaster event occurs, a significant number of residents in these countries are expected to face transient consumption poverty—which increases the local poverty rate. Depending on available coping mechanisms (or overall level of socioeconomic resilience of a given household or community) such disaster impacts tend to widen income inequalities and create additional sources of vulnerability. Disaggregating asset losses at household level enables this analysis to move from expected asset losses to some of the second-order effects of a given disaster, focusing specifically on income and consumption losses, which are to be expected when households are affected by a given shock. From this perspective, disaster losses are assessed not just in terms of economic costs but also as an obstacle to poverty reduction efforts and other sustainable development aspirations. If a disaster hits a region, it can potentially have significant second-order As seen in recent events effects on a large part of the population—particularly on immediate consumption and well-being levels. They may be because they have lost their in the region, when a job or another source of income, because they have lost their home and need major disaster event to pay for repairs and temporary housing or because they need to replace occurs, a significant possessions that have been lost. This sudden reduction in consumption levels makes some people fall under the poverty or subsistence line, and many number of residents of these people are already poor or near-poor. By definition, the poorest in these countries income groups do not have much assets to lose but minor losses can have disproportional impacts on their well-being—especially relative to wealthier are expected to face income groups, who have higher precautionary savings or better access to transient consumption financial resources to smooth the impacts of shocks. At a community level, poverty – which disasters also affect livelihoods, disrupt local markets and income flows, and make it hard for affected populations to rebuild their homes and businesses. increases the local poverty rate. Net income losses therefore incorporate these socioeconomic characteristics, which influence households’ recovery pathways. Accordingly, income losses describe better than asset losses the real impacts of disasters on well- being and recovery prospects and provide a complementary perspective to asset losses, by generating new insights into the costs of disasters in these countries. However, as a metric of disaster impacts, income losses may underestimate the link between poverty and disasters. First, they do not account for the reconstruction costs that households incur in rebuilding their assets after a disaster. Second, while wealthy households may have access to considerable savings, credit, remittances, and insurance to finance 28 | Disasters increase existing poverty levels Table 2: Most exposed regions in selected countries used to calculate their recovery, these resources consumption loss distributions for 200-year earthquake scenario and are rarely available, if at all, to 200‑year flood scenario. poor households (Morduch 1995; Townsend 1995). This is important, Country Earthquakes region Flooding region because access to these tools can mean the difference between a Albania Tirana Shkoder speedy, smooth recovery and a Armenia Yerevan Gegharkunik long-term poverty trap (Carter and Barrett 2006; Carter et al. 2007). Bulgaria South-Western South-Western These costs and resources all affect and South Central and South Central households’ ability to maintain a certain level of consumption after Croatia Countrywide Countrywide a shock, or else their consumption Georgia Tbilisi Municipality Tbilisi Municipality level drops. Greece Attica North Greece Romania Bucharest-Ilfov North-East In the eight countries Turkey Istanbul Mediterranean that were analyzed, (listed in table 2).18 It is also worth point to consider when deciding earthquakes and floods noting that disaster events with where to intervene and invest to could have the most return periods of at least 200 years manage existing disaster risks: risk significant impacts on are not as rare as they may seem: analyses that solely focus on asset statistically speaking, there is about losses could generate different pre-COVID poverty levels 5 percent probability (4.8 percent priorities than risk analyses that in Yerevan (Armenia), to be accurate) of observing at least look at poverty incidence or other one such catastrophic earthquake socioeconomic impacts. Figure 7 Tbilisi (Georgia), and in a given region in a 10-year period. summarizes the post-disaster Bucharest (Romania). In addition, there is about 5 percent increases in consumption poverty probability of observing at least one in each of these eight countries such catastrophic flood in a given due to a 200-year earthquake and region in a 10-year period.19 The due to a 200-year flood in the sub- The key findings generated in this goal of this analysis is to illustrate the national areas that had the highest analysis are presented below and impact of a plausible major disaster concentrations of asset risks due to aim to link regionwide earthquake event in a foreseeable future to earthquakes and floods. and flood risk assessment with more inform the development of resilience- granular insights on the household building measures that are risk In the event of a 200-year economy. The analysis particularly informed and socially inclusive. earthquake scenario, Yerevan focuses on how individual (Armenia) Tbilisi (Georgia) and households’ socioeconomic In the eight countries that were Bucharest (Romania) could characteristics can mitigate or analyzed, earthquakes and floods experience the highest poverty magnify the impact of disasters. could have the most significant increases (compared to 2016 To illustrate the post-disaster impacts on pre-COVID poverty poverty levels), with, respectively, an consumption loss distributions levels in Yerevan (Armenia), additional 18.7 percent of the local across selected countries, two Tbilisi (Georgia), and Bucharest population (167,800 individuals), major disaster scenarios were (Romania). This is an important 15.4 percent of the local population simulated—a 200-year earthquake (175,900 individuals), and 14.3 event and a 200-year flood event— percent (328,800 individuals) of 18 Note that Croatia was analyzed at country in the sub-national areas with level because the geographical locations of the local population who could major concentration of flood and households included in the GMD datasets immediately fall into poverty in the were not available. earthquake disaster risks for each aftermath of such an event. 19 A disaster event with X-year return period has of these select eight countries a 1/X chance of occurrence in every given year. Disasters increase existing poverty levels | 29 Figure 7: Multicountry comparison of post-disaster impact on 2016 poverty rates for 200-year earthquake scenario and for a 200-year flood scenario Istanbul region (345,800 individuals) Turkey Mediterranean region (94,300 individuals) Bucharest-Ilfov region (328,800 individuals) Romania North-East region (108,900 individuals) Attica region (14,700 individuals) Greece North Greece region (20,800 individuals) Tbilisi Municipality (175,900 individuals) Georgia Tbilisi Municipality (59,200 individuals) Countrywide (50,100 individuals) Croatia Countrywide (37,900 individuals) South-Western & South-Central region (45,000 individuals) Bulgaria South-Western & South-Central region (41,200 individuals) Yerevan region (167,800 individuals) Armenia Gegharkunik region (10,500 individuals) Tirana region (24,000 individuals) Albania Shkoder region (9,200 individuals) 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% Percentage of Regional Population Earthquake scenario Flood scenario Select regions listed in table 2 are considered for each country. Community members raise the walls banks with sandbags during the worst flooding ofthe Sava River on record across the Balkans in May, 2014. Credit: Nebojsa Markovic 30 | Disasters increase existing poverty levels Figure 8: Per capita consumption (US$) in capital region of Yerevan (Armenia), In the event of a 200-year flood before and immediately after a 200-year earthquake event (scenario projection). scenario, Shkoder (Albania), Tbilisi (Georgia), and Gegharkunik (Armenia) could experience the highest poverty increases (compared to 2016 poverty levels), with, respectively, an additional 7 percent of the local population (9,200 individuals), 5.2 percent of the local population (59,200 individuals), and 5.1 percent of the local population (10,500 individuals) who could immediately fall into poverty in the aftermath of such an event. To further illustrate post-disaster poverty increases, the impacts of a 200-year earthquake event in Armenia’s capital, Yerevan, and a 200-year flood event in Georgia’s capital, Tbilisi, are represented in figure 8 and in figure 9, respectively. These histograms illustrate the Note: PPP = Purchasing power parity. distributional impact of damages Source: World Bank calculations using GMD data (2016). Pre-disaster consumption uses Food Insecurity Experience Scale (FIES) data. on consumption level and the proportion of people living below the international poverty line before and after a disaster.20 This poverty Figure 9: Per capita consumption (US$) in Tbilisi Municipality (Georgia), before might be transitory, but rather and immediately after a 200-year flood event (scenario projection). than making assumptions on the duration of the consumption loss, the analysis focused on the increase in poverty as a direct impact of a given disaster scenario (200-year earthquake or flood event). In figure 8 and in figure 9, the poverty line and middle-class threshold are indicated by the dotted lines which correspond to US$5.50 and US$15.00 per day per person. In Yerevan, a closer look at the consumption distribution show that the large majority of individuals in the region consume between US$1,000 and US$3,000 per year per person (black outline), with approximately 25 percent of Yerevan’s population Note: PPP = Purchasing power parity. 20 Based on the World Bank’s International Source: World Bank calculations using GMD data (2016). Pre-disaster consumption uses FIES data. Poverty Line for upper-middle-income countries (US$5.50 a day/person in US$ 2011 PPP). Disasters increase existing poverty levels | 31 Romania Urban Search And Rescue (USAR) crews at work after a large earthquake in the Durrës region (Albania) on November 26, 2019. Credit: Italy Civil Protection. Box 2: The November 2019 Earthquake in Albania worsened local poverty levels. A magnitude 6.4 earthquake hit Table 3: Subjective poverty levels before and after the 2019 earthquake in Albania on November 26, 2019, causing affected municipalities in Albania. 51 fatalities, leaving 17,000 people displaced, impacting more than After Before 1 percent of its GDP, and ultimately Municipalities After-Before earthquake earthquake affecting more than 200,000 people across 11 municipalities. Throughout Percentage Percentage p.p. Percentage Albania, a total of 11,490 homes were damaged or destroyed, and another Durres 5.0% 8.0% 2.9 59% 83,000 needed repairs. According to Kamza 20.3% 21.6% 1.3 6% the PDNA (Government of Albania et al. 2020), an additional 26,000 people Kruja 2.2% 5.2% 3.0 135% have been pushed into poverty in the affected districts as a result of the Kurbin 4.4% 6.8% 2.4 53% earthquake and its impact on economic Shijak 6.0% 11.9% 5.9 99% activity, representing a 2.3 percentage point increase over the pre-earthquake Tirana 14.0% 15.4% 1.4 10% situation (table 3).a To put these figures in context, across Albania, 14.3 percent Vora 13.4% 29.3% 15.9 118% of population (414,000 people) lived below the national poverty line (Dávalos Total 11.9% 14.2% 2.3 19% et al. 2016).b Source: Government of Albania et al. 2020. poor municipalities and an expansion of middle-income Class Poverty Line (US$5.5 per a. Subjective poverty based on self-assessment poverty in others. person per day, 2011 PPP). The latest official of the household in seven municipalities was poverty figures for Albania date to 2012, when estimated at 11.9% before the earthquake, b. The World Bank poverty projections in Albania the poverty headcount was 39.1 percent and it is raised to 14.2% after the earthquake. indicate there has been a slow decline in (measured as US$5.5 per person per day, Similarly, in Serbia and in Bosnia and poverty levels in recent years, down to about 2011 PPP), so it is not possible to accurately Herzegovina, the floods of 2014 and 2015 34.5 percent (approximately 990,000 people) measure changes in poverty since then. resulted in a deepening of poverty in already in 2019 (World Bank, 2019). Based on upper- 32 | Disasters increase existing poverty levels Figure 10: Per capita consumption (US$) in the capital city of Tirana (Albania), representing a 11.1 percent drop, before and immediately after a 200-year earthquake event (scenario projection). compared to pre-disaster levels. For benchmarking purposes, results of an analysis conducted for the Albanian capital, Tirana (figure 10), were compared to the poverty results from the post-disaster needs assessment (PDNA) as a result of the recent November 2019 Albania earthquake (Government of Albania et al. 2020). While the post-disaster poverty levels were calculated using a different methodology (subjective poverty rate) and the November 2019 earthquake in Albania was not a 200-year event, the results for the actual event were relatively consistent with those from the simulated 200-year Note: PPP = Purchasing power parity. earthquake event. Consumption Source: World Bank calculations using GMD data (2016). Pre-disaster consumption uses FIES data. losses resulting from a simulated earthquake with 200-year return period are expected to push some also displace some 34,200 people 24,800 individuals into consumption Consumption losses individuals away from middle-class poverty in Tirana (4.7 percent of the consumption level, representing a regional population) and displace resulting from a 34 percent drop of the of Yerevan’s some 4,400 individuals away from simulated earthquake middle-class population,21 compared middle-class consumption level (corresponding to a 13 percent with 200-year return to pre‑disaster levels. decrease compared to pre-disaster period are expected Similarly, as shown in figure 9, levels) (figure 10). The PDNA for to push some 24,800 consumption losses in the Tbilisi the actual event that took place Municipality (Georgia) resulting in November 2019 indicates individuals into from a 200-year flood event are that following the earthquake, consumption poverty expected to push some 59,000 subjective poverty rates in the individuals (corresponding to 5.2 affected districts increased by in Tirana. 2.3 percentage points, equal to percent of its local population). In a municipality which had 11 26,000 people (this would be the currently living below the poverty line percent of its population already equivalent of a 19 percent increase (Fuchs Tarlovsky et al. 2019). Figure 8 living below the poverty line in 2016 in poverty rate compared to pre- shows in red the same distribution, (approximately 121,000 individuals), disaster levels) (box 2). but after a 200-year earthquake this simulated post-disaster effects event hits the region. In Yerevan, the shows a significant increase in the consumption losses caused by an population of individuals below earthquake of this magnitude have poverty line. In addition, the 200- approximately a 5 percent chance of year flood event would also push out occurring in the next 10 years, and some 15,000 individuals away from is expected to push close to 168,000 middle-class consumption level, individuals into consumption poverty (corresponding to 18.7 percent of its region’s population). Moreover, 21 US$15.00 a day in 2011 PPP dollar a day a 200-year earthquake would per person was used for delineating middle‑class line. CHAPTER 6: Disasters accentuate existing socioeconomic inequalities When disasters damage or destroy the assets on which individuals rely for their livelihood, affected households face income losses. Households’ ability to anticipate or mitigate these losses will inherently depend on their socioeconomic capacities, such as their pre-disaster income and their access to other types of financial resources. 6.1 Disasters tend to increase income inequality and reveal uneven access to mitigation measures In the event of a disaster, income losses are intrinsically linked to destruction of, or damage to, an individual’s own assets (local business or agriculture field), but they are also due to the disruption of factories, supply chains, and other types of infrastructure owned and maintained by others. Some individuals who have lost assets may receive public assistance or additional When disasters damage remittances, which supplement their regular income as they rebuild, while others will be left to their own devices. As seen with the Roma populations in or destroy the assets Bulgaria (box 3) or in Serbia in the aftermath of the 2014 floods (box 5),22 poor on which individuals households can also be further constrained if there are already underlying factors of vulnerability (for example, absence of recorded land tenure) and rely for their livelihood, uneven access to protective infrastructure and risk information (for example, affected households neighborhoods with inadequate drainage or outdated hazard maps) before face income losses. a disaster. Spatial patterns of socioeconomic inequalities between and within countries23—whether it’s related to income distribution or access to essential services (health, education, transport connectivity, and so on)—constrain poverty reduction efforts but can also have important implications in terms of potential post-disaster recovery (UNDP 2016). Asset losses and consumption poverty analysis describe individual households’ status in the instant right after a given disaster event takes place, and this information can provide important insights into how governments, local decision-makers and other stakeholders involved in the recovery should sequence early recovery efforts accordingly. But another important concern is whether disaster-affected households become mired in chronic poverty or are able to achieve a speedy 22 Box 5 in chapter 7 describes a similar post-flood experience of the Roma community in Serbia, looking specifically at the extent to which preexisting marginalization delayed the community’s recovery. 23 These patterns of spatial socioeconomic inequalities are often referred as ‘lagging regions’ in the European Union context (Farole, Goga, and Ionescu-Heroiu 2018). recovery24 as poverty impacts only highlight the situation of people living near the poverty line (and disregards what happens to other income groups in a given affected area). In contrast to asset losses per se, income and consumption losses can also support policy makers to better manage the severity and the duration of a disaster-related shock. 6.2 Anticipated post‑disaster recovery Natural hazards in Bulgaria can expose the decades-old, unsafe housing trajectories (national conditions of Roma communities. Credit: Nikolai Totukhov and sub-national level) Box 3: In Bulgaria, unequal access to preventive Although a disaster is a shock measures and risk information reveal vulnerability from which recovery takes years, of Roma populations. households in some regions will take longer to recover than in others. In Major damages due to disasters (flooding, earthquakes, landslides) the aftermath of a disaster, affected tend to have a disproportionate impact on low-income communities households lose productive assets, living in marginalized areas. Such occurrences are also more common in which directly reduce their income these areas because infrastructure is less developed (for example, flood but household-level consumption protection barriers) and emergency response planning (for example, losses do not end there, however, access to life-saving facilities) more limited. In June 2014, torrential rains because destroyed assets do not caused severe flash flooding across northeastern Bulgaria,a a submerging rebuild themselves. Rather, affected large parts of several cities in the region. The city of Asparuhovo—home households will have to increase to many minority groups, including nearly 1,000 Roma—was hardest their savings rate—that is, avoid hit by the unprecedented amount of rainfall. The municipality noted consuming some fraction of their that 122 families had to evacuate their houses and that 60 percent of post-disaster income—to recover destroyed houses were owned by members of the Roma community. these assets. Assuming each household pursues an exponential Because many of these homes were part of informal settlements and asset reconstruction pathway, had been built without complying with existing building codes or a reconstruction rate for each zoning restrictions, they were not stable enough to survive torrential household to optimize its well- waters. This event highlighted the inadequate and unsustainable being over the 10 years following housing conditions of the Roma community as well as the lack of a disaster while avoiding bringing preventive measures put in place by the municipality. As a result of the consumption below the subsistence flood, 14 people—most of them Roma—died. The Roma community level (if possible) is calculated. To in Asparuhovo is only one of many Roma communities living in flood- model each household’s recovery, prone zones across Central and Eastern Europe. There are no clear data indicating how many Roma live in flood-prone regions across the Balkans, but in Bulgaria, where EU estimates place the Roma population at about 24 The recovery phase is usually structured by 800,000, government data suggest that about half live in informal three main stages: (a) humanitarian relief, settlements. Many of these are poorly constructed, lack basic facilities, including search and rescue, and medical care; (b) restoration of basic services, and are isolated and located outside residential areas. including the supply of clean water, food, and sanitation, basic energy, mobility, and health Sources: Ionova 2014; Minority Rights Group International 2018; Naydenova 2014. care needs; and (c) the reconstruction phase, including infrastructure reconstruction, a. Bulgaria is highly exposed to disasters. From 2010 to 2016, disasters caused damages repair or replacement of building and up to US$1 billion and prompted the government to spend US$600 million on recovery production equipment, and asset recovery and over US$100 million on rescue and emergency. Flood and extreme temperatures by households—typically the longest and were the most frequent disasters, with flood causing the greatest damage and largest most costly phase of recovery (Hallegatte, numbers of affected population (Government of Bulgaria 2018). Rentschler, and Walsh 2018). Disasters accentuate existing socioeconomic inequalities | 35 it was assumed that disaster- Figure 11: Multicountry comparisons of time to reconstruct after a affected households rebuild their 200-year earthquake. lost assets exponentially over some number of years after a given shock. Time required to recover (number of years) 10 In addition, another parameter was 9 that households must maintain their consumption above a certain level 8 to meet their essential needs.25 7 6 5 4 Although a disaster 3 is a shock from which 2 recovery can take years, 1 households in some 0 Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey regions will take longer 25% post-disaster recovery rate 50% post-disaster recovery rate 75% post-disaster recovery rate to recover than in others. point during recovery but rather time: to reach a reconstruction When a given disaster takes place, covers the cost of public asset completion rate of 75 percent, the it is assumed that the government reconstruction for the duration of expected recovery milestones are borrows externally to finance the reconstruction and collects taxes to shortest in Turkey and Bulgaria cost of public asset reconstruction, fund this process many years later, (approximately 2 years), and the to hasten recovery and minimize after full recovery. longest in Albania and Georgia the financial burden on affected (8–9 years). As shown in figure 11, households. Eventually, the Looking at anticipated post-disaster a full recovery requires more than government recovers these costs recovery durations can provide 10 years in some countries, while in through taxation but only when useful insights for decision-makers other countries, the recovery time recovery is complete. Through and better inform post-disaster is significantly shorter. this mechanism, all households recovery and reconstruction efforts throughout the country share the (for example, by predicting which The speed of reconstruction can cost of public asset reconstruction affected areas are more likely to vary significantly from region to in the affected area. Because this face reconstruction bottlenecks region within the same country. is conceived as a one-time tax and recovery challenges than For each country, figure 12 shows to fund reconstruction, public other areas). Figure 11 shows how the fastest and slowest times to asset reconstruction costs are long it takes on average in each of recover 75 percent of assets. As not spread across the duration of these eight countries to recover shown, the recovery rate gap (that the reconstruction (in contrast to 25 percent, 50 percent, and 75 is, the difference between the income lost due to the destruction percent of assets destroyed in the fastest and slowest reconstruction of public assets, which does last aftermath of a 200-year earthquake rates) in Armenia and Georgia is for years after the disaster). It is event. While most countries would very high and corresponds to more assumed that the government does be able to recover 25 percent of than five years, while this same not collect the special tax at any their assets in less than a year (with gap is much shorter in Bulgaria, the exception of Albania26), the Greece, and Turkey, corresponding 25 If the households cannot avoid having completion rate of reconstruction to less than a year. Box 4 looks at consumption below the subsistence line rates diverges significantly over this example in more detail and (for instance, because consumption is below provides reasons why urban and the subsistence level even without repairing and replacing lost assets), then it is assumed 26 In Albania, for example, national average rural marginalized communities in that reconstruction takes place at the pace Romania’s northeastern regions time to recover for 25 percent, 50 percent, possible with a saving rate equal to the and 75 percent of assets is 1.21 years (442 average saving rate of people living at or are disproportionately affected by days), 3.26 years (39 months), and more below subsistence level in each region of the disasters and have access to limited than 8.75 years (8 years and 9 months), eight selected countries. respectively (Figure 11). 36 | Disasters accentuate existing socioeconomic inequalities Figure 12: In-country gap between fastest and slowest sub-national post-disaster recovery for 200-year earthquake (time required for recovering 75 percent of assets). Tbilisi Guria, 4.68 Kvemo Number of years for post-disaster recovery Kartli and 12 Berat Kukes Mtskhe- 7.2 10 ta-Mtianeti Yerevan Shirak 10 3.68 9.22 10 Bucharest North 8 -Ilfov -East 5.22 6.3 Countrywide 6 South- Northern 4.95 Attica Northern Western and 2.75 and West Central and South- South- Anatolia, East South- Eastern Eastern West Anatolia 4 Central 2.46 2.84 Marmara 2.23 2.29 2.13 2 0 Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey Fastest region Slowest region capital (at national and sub-national as well: households reconstruct Romania’s northeastern level),27 the extent of damage, the their own assets using their savings level of social protection (and post- and allocating their consumption regions are disaster assistance) provided by budget to reconstruction, unless disproportionately governments, and feasible level of they carry private insurance; the capital reallocation from household national or regional taxpayers affected by disasters consumption expenditures to reconstruct public assets; and other and have access to very rebuild the lost assets—with other private assets are rebuilt by private limited resources to conditions remaining the same. business owners or corporations. Again, this modeling result is largely cope with and recover The productivity and vulnerability independent of initial asset losses, from such shocks. of these assets to various hazards expressing instead the expected can vary from region to region— capacity of each region in each of that will result in variations in the the eight countries (except Croatia)28 resources to cope with and recover reconstruction time. In addition, it to cope with losses in the aftermath from such shocks (both at the is important to note that the liability of a 200-year earthquake event. household and community levels). for reconstruction and well-being In terms of sub-national recovery losses that follow a disaster vary within the same country, DRM strategies seeking to minimize time to recover after disasters should 27 The average capital productivity is the measure of how well physical capital is used therefore focus on providing timely in providing goods and services. Productive post-disaster support in areas use of physical capital and labor are the two most important sources of a nation’s that are likely to face bottlenecks material standard of living. In addition, how and have a much slower expected well a nation uses its physical capital affects the return people get on money they save. recovery process. The higher the returns, the less they need to save for the future and they can consume 28 Due to the lack of NUTS 2 geospatial data As stated earlier, one of the reasons today (Agrawal et al. 1996). Therefore, for households in the Croatia dataset, the the higher the productivity of capital in a analysis for this country is limited to the for different recovery trajectories is region is, the lower the capital required to country level and is not able to measure the primarily the average productivity of restore production, thus the lower need for post-disaster recovery gap between regions reconstruction resources. within Croatia. Disasters accentuate existing socioeconomic inequalities | 37 Box 4: Enhance socioeconomic resilience across Romania: How to best target interventions? The risks posed by disasters are Marginalized Communities in Romania (Anton et al. 2014) and Atlas Marginalized Rural disproportionately higher for the poor Areas and Local Human Development in Romania (Sandu et al. 2016).a These tools and other vulnerable groups, whose demonstrate that in Romania both the urban and the rural marginalized areas are welfare and long-term prospects have spread across all counties and regions, with a particular concentration in the North- always been acutely vulnerable to East region. Most rural marginalization areas are small (under 500 inhabitants) exogenous shocks. In Romania, from and have an ethnic dimension, as Roma people are statistically overrepresented a poverty alleviation perspective, the and concentrated in segregated communities. The main limitation of the existing concept of marginalized areas – in both targeting tools is the lack of municipal geographical references, which sometimes urban and rural areas – are defined and prevents the data from shaping more granular interventions in target areas. empirically identified as census sectors that simultaneously experience severe levels of deprivation in three distinct Figure 13: Distribution of rural marginalized areas across Romania, 2011. areas: human capital, employment, and housing conditions. The populations living in these areas are characterized by a deficit of human capital, tend to generate revenue from the informal sector (combined with social transfers and agriculture in rural areas), and often live in precarious dwellings even by the usual low standard for rural housing. Such areas are therefore territorial concentrations of multidimensional poverty and, as a result, are particularly vulnerable to exogenous shocks such as disasters (floods, earthquake, landslides) or climate risks (heatwaves, droughts, and so on). There are significant differences between marginalized areas that are urban (ghettos, slums, mahalas, social housing concentrations, and historical areas) and those that are rural (as shown in figure 13). While urban marginalized areas are often localized centrally (as shown in figure 14), rural marginalized County limit Village with marginalized communities in which areas are prone to geographic isolation less than 20% of the inhabitants are Roma and are usually located at the outskirts Country capital Village with marginalized communities in which of well-connected villages. In terms of European road more than 20% of the inhabitants are Roma percentage of population, 6.2 percent of Romania’s rural population and 3.2 percent of its urban population live in marginalized areas (estimates Note: World Bank estimations were based on data from the 2011 Population and Housing Census. based on the 2011 Population and The analysis was carried out at the census sector level for all rural administrative units. Sectors with fewer than 50 inhabitants were not included in the analysis. Dwelling Census). Source: Sandu et al. 2016 (World Bank). The main targeting tools used for a In 2012, the Government of Romania and the World Bank partnered to facilitate the preparation and implementation of projects funded by the European Union. This agreement included these two atlas the analysis of marginalized areas initiatives, which were framed within a wider project of designing strategies for the integration of in Romania are The Atlas of Urban poor areas and disadvantaged communities across Romania. Box 4: Continued. For both rural and urban areas, the empirical evidence generated in both atlases confirmed that marginalized communities face disproportionately high exposure to hazards and have access to limited resources to cope with and recover from such shocks (both at the household and community levels). Poverty reduction efforts in Romania could thus generate significant disaster risk reduction dividends if interventions prioritize the targeting of marginalized areas in both urban and rural zones. Figure 14: City maps with marginalized communities reported by the local authorities: Medgidia town example. Region: South‐East    County: Constanța    City: Medgidia    Marginalized communities declared by local authorities    Legend      City limit         Types of marginalized urban areas      (number) Estimated   Ghetto‐type areas with blocks of flats   number of  Ghetto‐type areas in former industrial colonies inhabitants    Slum‐type areas with houses  in the area      Slum‐type areas with improvised shelters    Cartography: ESRI,  Areas with modernized social housing  ArcGIS 10.1      Historical (central) neighborhoods with social housing and/or buildings abusively occupied      Mixed areas    Next  to  the  marginalized  communities,  the  local  name and  the  estimated  number of  inhabitants are  shown,  only  if and  as declared by the local authorities.      Source:   Anton et al. 2014 (World Bank).     Sources: Anton et al. 2014; Sandu et al. 2016 ; Walsh and Hallegatte 2020.   104 CHAPTER 7: When hit by disasters, poor people tend to incur more well-being losses than wealthier groups 7.1 Regional benchmarking Given their limited livelihood choices, poor people are often more vulnerable than others to direct and indirect disaster losses and more likely to rely on unsustainable coping mechanisms. Even if disaster-affected households suffer identical asset losses, their consumption losses and recovery time will vary according to their socioeconomic status and the resources available to them (in part because insurance, savings, remittances, and public support help some households to smooth the consumption shocks). In particular, wealthier households will be able to spend down their savings or reduce luxury consumption, while poorer households will often have to cut spending on basic needs and essential consumption (for example, reducing food and health expenditures, withdrawing children from school activities, sending children to work, and taking on additional debt) thus potentially damaging their long- term prospects (UNDP 2016). Moreover, since most assets in low-income Given their limited households and communities are uninsured, poor people often shoulder the greatest share of disaster losses. The example of the Roma community the livelihood choices, poor 2014 floods in Serbia is given in box 5 to illustrate how the same disaster can people are often more affect several income groups in different ways: US$1 in consumption losses vulnerable than others can have widely different consequences for individual households, depending on their income level. to direct and indirect disaster losses and To maximize development co-benefits, DRM strategies and budgets should account for these differences, and here the concept of well-being losses can more likely to rely on be useful. The full impact of disasters on the population’s well-being depends unsustainable coping not only on direct impacts but also on the duration of the recovery and mechanisms. reconstruction period and the tools that affected populations have to cope with and recover from such shocks. Ideally, these include social transfers, formal and informal post-disaster support, savings, insurance, and access to credit. The effects of disasters on individuals’ welfare therefore depends on their ability to cope with and recover from a shock. Even if the asset losses of the poor account for just a small fraction of the total in any disaster, US$1 in asset losses matters more to a poor household than to a wealthy one. Well‑being losses are defined so that US$1 in well-being losses indicates the same impact on poor people as on the wealthy even when their asset losses differ. Therefore, well-being losses measure the effects of disasters on all individuals’ welfare more fully than asset losses. FACING PAGE Dried sunflowers on the side of the road. Credit: Boris Rumenov Balabanov Roma settlement in Belgrade (Serbia) in the aftermaths of the May 2013 floods. Credit: Baloncici/Shutterstock.com Box 5: Socioeconomic marginalization constrained the recovery of the Roma population after the 2014 floods in Serbia. In the context of disasters, minorities therefore more likely than less products from flood‑affected areas are often at risk due to a combination vulnerable groups to have their homes was also temporarily banned due to of different types of vulnerabilities. and livelihood prospects damaged or possible contamination risks. Children These include their pre-disaster level destroyed by floods. At the time of the in poor and vulnerable Roma families of economic poverty, likelihood of floods, 93 percent of Roma households were also at elevated risk of not living and working in hazard-prone owned their own dwellings but had no enrolling in school in September 2014 areas, likelihood of living and working insurance coverage (UNDP 2016). because floods had lowered incomes in substandard buildings that are and had worsened living conditions of poor quality and/or insufficiently Along with the lack of insurance, the (Government of Serbia et al. 2014). maintained, lack of access to optimal lack of tenure security added further On average, Roma represented social services, and exclusion from constraints to the reconstruction 2.1 percent of the population in information channels and decision- process. A major obstacle to flood‑affected municipalities of Serbia, making and consultation structures. housing reconstruction during the but their spatial distribution suggests In several cases, discrimination and recovery phase was the absence of that as a population group they were marginalization are the root causes documentation for informal Roma disproportionately affected by the May of these layers of vulnerabilities settlements. Because most Roma 2014 Floods in Serbia.a (UNDP 2016). housing units did not meet the requirements for registration as Sources: Government of Serbia et al. 2014; An example demonstrating how these UNDP 2016. property during the pre-disaster phase, layers of vulnerability interplay with households were unable to provide a. According to the 2011 Population Census, 147,607 Roma live in Serbia, representing risk drivers is the experience of the documentation for their lost homes. 2 percent of the total population. However, Roma population in Serbia during the Lacking official identification, Roma unofficial data place their numbers at May 2014 floods. Roma, who are among 450,000 to 500,000 (6.0–6.5 percent of families became effectively invisible the total population). The Roma National Serbia’s most vulnerable communities, and were not able to access help. Council found that the 2014 floods affected tend to live in informal settlements Floods also affected Roma livelihoods. 6,032 Roma in 714 households from 2,255 municipalities. Obrenovac, with in high-risk and underserviced areas The large portion of the population 2,064 individuals distributed across 296 such as floodplains and in housing engaged in subsistence farming households, was the municipality most built without construction permits affected by the floods (Government of Serbia suffered the destruction of farms et al. 2014). or property registration. They are and yards. The sale of agricultural When hit by disasters, poor people tend to incur more well-being losses than wealthier groups | 41 Table 4: Comparison of annual average risk to assets, annual average risk to well-being, and socioeconomic resilience for select countries in the ECA region. Annual average Annual average Socioeconomic asset risk well-being risk resilience (%) (US$, millions in 2019 Well-being risk Well-being risk Well-being risk Well-being risk Type of hazard Type of hazard (US$, millions) (US$, millions) (US$, millions) (US$, millions) Earthquakes Earthquakes AND floods Asset risk Asset risk Asset risk Asset risk OR floods Country (GDP %) (GDP %) (GDP %) (GDP %) values) GDP EQ 48.2 0.32% EQ 101.5 0.66% 47% Albania 15,279 124.6 0.82% 267.5 1.75% 47% F 76.4 0.50% F 166 1.09% 46% EQ 54.4 0.40% EQ 167.1 1.22% 32% Armenia 13,672 76.5 0.56% 226.6 1.66% 34% F 22.1 0.16% F 59.5 0.44% 37% EQ 31.7 0.05% EQ 90.4 0.13% 35% Bulgaria 68,558 135.9 0.20% 788.2 1.15% 17% F 104.2 0.15% F 697.8 1.02% 14% EQ 253.6 0.42% EQ 849.1 1.40% 29% Croatia 60,752 844.3 1.39% 3,460.9 5.70% 24% F 590.7 0.97% F 2,611.80 4.30% 22% EQ 83.1 0.48% EQ 285.5 1.63% 32% Georgia 17,477 199.8 1.14% 579.3 3.31% 34% F 116.7 0.67% F 293.8 1.68% 39% EQ 89.8 0.04% EQ 238.4 0.11% 37% Greece 209,852 166.5 0.08% 698.1 0.33% 24% F 76.7 0.04% F 459.7 0.22% 16% EQ 139.2 0.06% EQ 332.6 0.13% 41% Romania 250,077 575.4 0.23% 1,429.8 0.57% 40% F 436.2 0.17% F 1,097.20 0.44% 39% EQ 711 0.09% EQ 1,598.60 0.21% 44% Turkey 761,425 1,554.2 0.20% 3,800.0 0.50% 41% F 843.2 0.11% F 2,201.40 0.29% 38% EQ = Earthquake, F = Flood *GDP in US$, billions in 2019 values. 42 | When hit by disasters, poor people tend to incur more well-being losses than wealthier groups Figure 15: Comparison of socioeconomic resilience for entire population of Figure 15 displays the average select countries. socioeconomic resilience of all eight countries with respect to both floods and earthquakes. As Countrywide Socioeconomic Resilience shown, socioeconomic resilience to 50% earthquakes is highest for Albania 45% (47 percent) and lowest for Croatia 40% (29 percent). Albania also has the 35% highest socioeconomic resilience to 30% floods (46 percent), while Bulgaria 25% has the lowest socioeconomic 20% resilience to floods (14 percent).30 15% 10% 5% 0% Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey Socioeconomic Flood Earthquake resilience integrates the real impacts of disasters Socioeconomic resilience integrates respect to floods and earthquakes on households’ the real impacts of disasters for the entire population of each well‑being by measuring on households’ well-being by of the eight countries of interest. measuring an economy’s ability to As shown, Georgia has the highest an economy’s ability to minimize the impact of asset losses expected annual asset losses due minimize the impact on well-being levels, for example, by to earthquakes in GDP percentage of asset losses on evaluating the ability of households (0.48 percent), and Greece has the (or communities) to cope with lowest (0.04 percent). Given its higher well‑being levels. and recover from disasters.29 exposure of assets to disaster risks These variations from a modelling (in GDP percentage), Georgia has perspective are due to the level of the highest expected annual well- The difference in socioeconomic asset exposures (and subsequent being losses due to earthquakes in resilience of each region is due to damages), susceptibility of GDP percentage (1.63 percent), and variations in estimated asset losses household income and consumption Greece has the lowest (0.11 percent). and well-being losses. In other to asset losses, and the variety of Moreover, Croatia has the highest words, while countries (or sub- reconstruction dynamics in each expected annual asset losses due national areas) that enjoy higher geographic area. From an economic to floods in GDP percentage (0.97 resilience levels are better able to point of view, these variations are percent), while Greece has the lowest absorb the disaster-related shocks, due to the level of investments that (0.04 percent). Similarly, Croatia has countries (or sub-national areas) with are allocated to housing, private, the highest expected annual well- a lower socioeconomic resilience and public infrastructure, and social being losses due to floods in GDP level would benefit from resilience- transfers, as well as structural percentage (4.30 percent), while characteristics of local and regional Greece has the lowest (0.02 percent). 30 Socioeconomic resilience to each type of economies and their level of While the expected annual asset natural disaster in a region is obtained susceptibility to natural hazards. losses due to both earthquakes and by dividing the estimated asset losses by floods combined in GDP percentage estimated well-being losses. For example, socioeconomic resilience of Albania to Table 4 summarizes the average for Croatia are estimated at 1.39 earthquakes is obtained as 48.2/101.5 = 47%. annual risk to assets, average percent, at the other end of the range annual risk to well-being, and is Greece at 0.08 percent. Similarly, socioeconomic resilience with Croatia has the highest expected FACING PAGE annual well-being losses due to both A family stand in front of their house, earthquakes and floods combined in following flash floods in Athens (Greece) 29 Socioeconomic resilience can be defined GDP percentage (5.70 percent) while in November 2017. Credit: Alexandros as the ratio of expected asset losses to well being losses. Greece has the lowest (0.14 percent). Michailidis/Shutterstock.com 44 | When hit by disasters, poor people tend to incur more well-being losses than wealthier groups Figure 16: Comparison of socioeconomic resilience levels in different loss of labor income, availability of socioeconomic groups for select countries. savings, the need to relocate during repair works, and the household’s pre-disaster socioeconomic Countrywide Socioeconomic Resilience status (Markhvida et al. 2020). 50% By comparing annual asset losses 45% and annual well-being losses due 40% to floods and earthquakes across 35% different socioeconomic groups in 30% Greece, 31 figure 18 illustrates the 25% disproportionate consequences 20% 15% of disasters on the poorest 10% households, looking at overall totals 5% (left) and per capita levels (right). 0% Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey Poor National Average This significant gap in socioeconomic building investments—including figures, low-income groups with resilience can be financial inclusion, public and private a low socioeconomic resilience partly explained by insurance, and other optimized post- level would benefit from targeted disaster support mechanisms—to solutions and measures that the difference in complement the traditional DRM specifically strengthen their pre‑disaster income toolbox of measures, policies, and ability to weather disaster-related investments. shocks—including adaptive social levels and access to safety nets, increase in income financial resources. diversification, or improved post- 7.2 Quantifying disaster disaster recovery policies—in impact on the poor and addition to traditional DRM toolbox As shown, while each household of measures and investments (a) to in the poorest income category other socioeconomic reduce their exposure to hazards suffers only US$4 in asset losses groups (for example neighborhood every year, they experience US$226 upgrading through enhanced in well-being losses every year, far Since this study was conducted drainage or risk-informed land exceeding the value of their asset based on household data in each use planning) and (b) to reduce losses. On the other hand, the country, it is also possible to the vulnerability of their assets wealthiest income group suffers understand patterns of asset losses (providing reliable land tenure to 83 percent of the overall asset and well-being losses not just at encourage housing investments losses while experiencing only the regional and country level but or small-scale protective 46 percent of the overall well-being also across different socioeconomic groups. Figure 16 compares the infrastructure). level of socioeconomic resilience At a household level, asset losses of each country’s poor population 31 Socioeconomic groups have been tend to increase with pre-disaster with the average national defined as follows: poor (households income because wealthier with consumption level of less than PPP socioeconomic resilience in each US$5.50 per household member per day), households own higher valued of the eight countries. As shown, vulnerable (households with consumption assets (figure 17). This type level of equal or more than PPP US$5.50 the socioeconomic resilience gap of asset-centered loss metric, but less than PPP US$10.00 per household between the poor and the national member per day), secure (households with however, does not indicate how consumption level of equal or more than average in Albania is extremely the overall consumption and well- PPP US$10.00 but less than PPP US$15.00 high, while socioeconomic resilience per household member per day), and being of the household is disrupted middle class (households with consumption gap is relatively lower in Georgia since it does not take into account level of more than PPP US$15.00 per and Armenia. Similar to previous household member per day). When hit by disasters, poor people tend to incur more well-being losses than wealthier groups | 45 losses. 32 On the other hand, while Figure 17: Average total annual losses for various socioeconomic groups each household in the middle class in Greece. category on average loses US$34 in assets every year (that is, about 450 7.0 mil. eight times greater than each poor household’s asset losses), their 400 Average annual losses (million PPP$ per year) average well-being loss is about 350 US$61 per year (that is, about a fourth of each poor household’s 300 well-being losses). 250 In addition, as shown in figure 18, the level of resilience of households 200 1.3 mil. 1.9 mil. of poor, vulnerable, secure, and Population = 0.7 mil. 150 middle-class socioeconomic groups in Greece is 2 percent, 9 percent, 100 18 percent, and 56 percent, respectively. This significant gap can 50 be partly explained by the difference 0 in pre-disaster income levels and Assets Well-being Assets Well-being Assets Well-being Assets Well-being access to financial resources, which Poor Vulnerable Secure Secure affects the households’ ability to i < $5.50/day $5.50 < i < $10 $10 < i < $15 i > $15 recover in a timely and efficient Earthquakes Flooding manner (Markhvida et al. 2020). Figure 18: Average per capita annual losses for various socioeconomic groups in Greece. Average impact (PPP per capita) 3.0 asset loss = $4 well-being loss = $226 Average annual losses (% of total consumption) resilience = 2% 2.5 2.0 asset loss = $11 well-being loss = $127 resilience = 9% 1.5 asset loss = $17 well-being loss = $95 resilience = 18% asset loss = $34 1.0 well-being loss = $61 resilience = 56% 0.5 0.0 Assets Well-being Assets Well-being Assets Well-being Assets Well-being Poor Vulnerable Secure Middle class 32 It is also worth noting at this stage that the i < $5.50/day $5.50 < i < $10 $10 < i < $15 i > $15 urban poor are likely to be underrepresented within country-level averages. Members Earthquakes Flooding of this group face higher costs of living in search of economic opportunity and may receive little support from their home communities while still being expected to send remittances home. Indeed, while the analysis is limited to only the poorest group in each country, the range of socioeconomic resilience values narrow significantly. A woman fills a pail of water in Azerbaijan. Credit: Allison Kwese CHAPTER 8: Policy actions that can minimize disaster impacts at a household level Policy actions need to reflect communities’ ability (or inability) to cope with and recover from shocks, such as disasters. Determining well-being losses and the resulting level of socioeconomic resilience can inform resilience- building and risk reduction interventions for both pre-disaster and post- disaster contexts, and also examine their added value across different income groups (or between different geographical areas such as different regions within the same country country). This section examines and compares three targeted policy options at country level: (a) reducing the asset vulnerability of the poor population by 30 percent, (b) increasing the income of the poor population by 30 percent, and (c) reducing the post-disaster reconstruction time by 30 percent. The objective is to benchmark the impacts of each of those policies with respect to the status quo in each of these countries and create opportunities to review optimal policy actions for each of these countries. The World Bank’s Unbreakable report (Hallegatte et al. 2017) and subsequent Policy actions need to studies not only developed the concept of well-being loss and socioeconomic resilience but also introduced a set of policy actions that could either focus reflect communities’ on reducing asset losses (table 5) or on increasing resilience against disasters ability (or inability) to and climate shocks (table 6). While the former policy actions focus on reducing cope with and recover exposure or asset vulnerability (enforcing risk-informed land-use planning, rolling out nature-based solutions to mitigate flood hazards, enforcing robust from shocks, such as building codes, and so on), the latter policy actions are applicable when asset disasters. losses cannot be entirely prevented and when there is a need to improve the ability of people to cope with shocks that cannot be avoided (financial inclusion, revenue diversification, access to catastrophe insurance, and so on). For example, social protection—through scalable or adaptive social safe nets—is increasingly perceived as an effective approach for building community resilience against climate change and disasters. By ensuring basic levels of consumption, encouraging investments in resilient productive livelihood assets (drought-resistant crops, purchasing solar panels, and so on), or accelerating the reconstruction and recovery process (housing repairs and so on), social protection facilitates access to basic services and increases the capacity of at-risk populations to withstand, manage, and recover from shocks. The rationale behind adaptive social services is that these systems should be available to all and adjusted to the specific needs of the most vulnerable and marginalized population, considering cultural characteristics, social fabric, and structural inequities (UNDP 2016). However, when a specific shock overwhelms existing coping mechanisms, more robust instruments are required. Wealthier 48 | Policy actions that can minimize disaster impacts at a household level Table 5: Policy actions focusing on asset loss reduction. Policy actions focusing on Policy examples asset loss reduction Reduce the exposure of the poor Upgrade neighborhoods with improved drainage; initiate preventive resettlement programs away from at-risk areas; undertake ecosystem conservation; and so on. Reduce the exposure of the non-poor Adopt risk-informed land use and urbanization plans; influence future urban development. Reduce the vulnerability of the poor Record land tenure to enhance investments in housing; improve infrastructure that people’s assets serves the poor. Reduce the vulnerability of non-poor Change construction and building norms; improve general infrastructure. people’s assets Provide universal access to early Invest in hydrometeorological observation systems’ and weather forecasting capacity; warning systems ensure capacity to issue and communicate early warning and for people to react. Source: Hallegatte et al. 2017. Table 6: Policy actions focusing on increasing resilience. Policy actions focusing on Policy examples increasing resilience Favor savings in financial forms Develop banking sector and favor mobile banking; support development of savings instrument for the poor. Accelerate reconstruction Develop access to borrowing and insurance for people, firms, and local authorities to facilitate recovery and reconstruction; ensure the government has the liquidity to fund reconstruction; increase openness for workers, materials, and equipment to facilitate reconstruction; streamline administrative processes (for example, building permits, scale-up of post-earthquake inspection checks, debris clearing, and so on). Increase income diversification Create new cash transfers; ensure that contributory social protection schemes are (social protection and remittances) available to poor people; reduce the cost of remittances. Make social safety nets more scalable Create social registries that can add beneficiaries; implement a budgetary process to increase social expenditures after a disaster; create the right delivery mechanisms; develop indicators and procedures for the automatic scale-up of social safety nets. Develop contingent finance and Create a reserve funds with utilization rules; prepare access to contingency credit lines reserve funds (such as CAT-DDOs); create regional risk pools (such as CCRIF); transfer part of the risk to global reinsurance or global capital markets (such as FONDEN bonds). Improve capacity to deliver Combine the two previous sets of actions. post‑disaster support Improve access to insurance for Create insurance markets and ensure their sustainability. firms and households Note: Cat-DDO = Catastrophe Deferred Drawdown Option; CCRIF = Caribbean Catastrophe Risk Insurance Facility; FONDEN = Natural Disasters Fund (Mexico). Source: Hallegatte et al. 2017. Policy actions that can minimize disaster impacts at a household level | 49 income groups might have access to It is worth noting that while this 2. Increasing the income of savings or market insurance, but the method of analysis can help examine the poor population by 30 poorest households might face high the advantages and disadvantages percent. In the short term, transaction costs and might resort of promising solutions, a this hypothetical policy can be to detrimental coping strategies comprehensive set of DRM solutions achieved through various social as a result. In all cases, a variety of needs to be systematically sought safety nets and government- social protection programs exist after. One of the main reasons is that funded social transfers and even that build resilience in the long term, one standalone, specific instrument potentially by reducing the cost and their benefit can be measured will be insufficient because of remittances. In the long term, in avoided well-being losses, even if interventions need to be adapted this can be achieved through they do not necessarily help lower to the scale of different disasters inclusive growth, for example, the reconstruction cost or cost of and reflect the socioeconomic by improving the productivity of insurance premiums. needs of poor populations (and the poor through better health other vulnerable people). In other care, investing in education, and Well-being losses and socioeconomic words, determining a definitive creating more opportunities resilience are therefore useful decision on any of these potential for stable and well-paying jobs. not just as diagnostics of disaster interventions would necessitate However, for this study, only the impacts, but they can also be used more comprehensive policy analysis income of the poor was increased to measure the expected or actual than what is presented here. Instead, by 30 percent, ceteris paribus. benefits of DRM investments and what is provided is a starting point related resilience-building measures for stakeholders and decision- 3. Reducing the post-disaster (for both pre- and post-disaster makers to initiate discussions on reconstruction time by 30 interventions) across different policy options in different sectors percent. This hypothetical policy income groups. The proposed that could reduce the impact of can be achieved through the disaster impact metrics—poverty disasters and increase the level combination of engineering and headcount, poverty gap, and of socioeconomic resilience at macroeconomic solutions that well-being losses—can be used to community level. Theoretically, there would lead to improving the better understand the effectiveness are virtually infinite combinations average productivity of capital of resilience-building measures of policy options to be examined. and enabling better access outside the traditional DRM toolbox However, for comparative purposes, to financial products (such as (Walsh and Hallegatte 2020). Even the impacts of the following three accessing reconstruction loans; if asset losses will not necessarily policy actions33 on asset losses, well- streamlining administrative be reduced, these interventions being losses, and socioeconomic procedures such as building can still enhance communities’ resilience of these eight countries in permits; providing better access level of socioeconomic resilience— ECA region are examined. to equipment, workers, and that is their capacity to cope with materials for reconstruction; and and recover from asset losses 1. Reducing the asset so on). when they occur and reduce the vulnerability of the poor well-being impact of disasters population by 30 percent. While these interventions have (Hallegatte et al. 2020). In addition, This hypothetical policy could intrinsic characteristics and require a while DRM interventions that are be achieved through various range of policy instruments tailored asset-informed primarily focus on strategies including the structural for different hazard contexts or types protective infrastructure (flood reinforcement of buildings where of affected populations, this type of barriers for example) and location households from the poorest benchmarking is not always possible and maintenance of key assets income groups reside (that is, with traditional direct asset loss (for instance, risk-informed land those living with less than US$5.50 assessments. The implementation use planning or building codes), per person per day). strategy of these policy actions is not interventions that are well-being in the scope of this study. However, informed can justify a wider set the assessment of impacts of each of interventions such as financial of these three hypothetical policies is inclusion, adaptive social safety 33 Each of these policy options, based on the focus of this report—regardless findings provided in the previous chapters, nets, emergency preparedness, are also compared with a counterfactual of the means to achieve those ends. or contingent planning. scenario (with no policy action in place) to Figure 19 illustrates the annual better measure their added value. 50 | Policy actions that can minimize disaster impacts at a household level Figure 19: Comparison of expected annual Avoided losses with different cases, it can even result in higher policies for select countries. annual average losses in Albania, Bulgaria, and Romania. This is due to the acceleration of households’ Annual Avoided Losses (million US$) 160 spending rates to rebuild the 140 damaged assets in a shorter period. 120 Therefore, without increasing 100 the post-disaster support, the 80 acceleration of reconstruction can 60 potentially create more harm and 40 push the most vulnerable population 20 deeper into consumption poverty. 0 Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey Similarly, figure 20 illustrates the gains in socioeconomic resilience Reduced vulnerability Increased incomes Reduced times that can be achieved through each of these policies. As shown, reducing the asset vulnerability of Figure 20: Comparison of socioeconomic resilience with different policies the poor population by 30 percent for select countries. seems to be the policy option that 55% yields the highest benefits (among policy options listed), given that it 50% Socioeconomic resilience would increase the average level of 45% socioeconomic resilience in most 40% countries, closely followed by 30 35% percent income increase for the 30% poor population. Albania, closely 25% followed by Georgia and Romania, 20% are the countries where reducing 15% the asset vulnerability of the poor Albania Armenia Bulgaria Croatia Georgia Greece Romania Turkey population can realize the largest increases in socioeconomic resilience Reduced vulnerability Increased incomes Reduced times Status quo benefits (slightly over 50 percent). It is worth noting that results from this study indicate that Albania, Georgia, average avoided losses34 due to prone areas), for example, by and Romania are also the same implementing each of these three thoroughly investigating the seismic countries that are projected to have policies compared to the status quo safety of apartment buildings.35 the slowest pace of post-disaster in each of these eight countries. On the other hand, the reduction recovery and reconstruction in the As shown, reducing the asset of reconstruction time has limited aftermath of a 200-year earthquake. vulnerability of the poor population effectiveness and, in some rare It is worth noting that reducing by 30 percent leads to higher annual the reconstruction time without average avoided losses in all eight increasing the government support countries. This is promising given 35 According to the World Bank, some 20 for the poor either has no significant that several governments in the countries across the ECA region have a 10–20 impact on socioeconomic resilience region are increasingly recognizing percent chance of being affected by a major earthquake in the next 50 years. Of particular or, in some cases, can result in the need to improve the housing concern in these countries are pre-1990s decreasing household resilience. That stock for the most at-risk populations mass-produced, prefabricated, multifamily residential buildings which constitute a is due to the higher allocation rate (including those living in disaster- significant percentage of the housing stock. of households’ consumption budget The poor quality of these buildings was tragically illustrated in the 1988 Armenia to reconstruction that is needed in a 34 Annual average avoided losses is defined (Spitak) earthquake, when many of them shorter period. as the sum of annual well-being losses per collapsed—significantly contributing to the year that can be prevented to accrue by nearly 50,000 fatalities and 130,000 injuries implementing each policy compared to the that resulted from this disaster (Mathema and status quo. Simpson 2018). CHAPTER 9: Conclusions and considerations for further analysis As illustrated in this report, the risks posed by disasters and climate risks are rising most acutely for the poor and other vulnerable population groups, whose welfare and long-term economic prospects have been acutely vulnerable to exogenous shocks such as disasters and climate risks. Today, with access to a variety of analytical tools, commensurable datasets, global best practices, and other forms of innovative methods to better grasp, quantify, and explain the effects of disaster impacts, decision-makers can rethink the ways such shocks are quantified so that they can better anticipate or identify which population groups (or geographical areas) are likely to be the most impacted, and why. Additional areas of analysis are also highlighted for consideration in similar future research in the ECA region. This report has presented the results of a multicountry risk assessment in the ECA region based on an expanded framework which highlights the ability of households to cope with and recovery from disaster asset losses. The risks posed by The concept of well-being losses was used as the main metric to measure disasters and climate the severity of disaster impacts. The proposed framework adds to the three traditional components of disaster risk assessment—hazard, exposure, and risks are rising most vulnerability—and a fourth component, socioeconomic resilience. Similar to acutely for the poor traditional risk assessments, socioeconomic resilience can be quantified using and other vulnerable a variety of spatial resolutions, ranging from household to national averages. This risk assessment methodology has also been applied in several countries population groups. to shape DRM strategies, to better incorporate socioeconomic resilience considerations—including the capacities of affected communities, economies, and other networks to recover from such shocks. In particular, from a socioeconomic perspective, it generates empirical evidence to identify priority income groups as well as at-risk sub-national areas that are significantly vulnerable to disasters. Informed by a more socially inclusive accounting of disaster costs, this methodology can provide new justifications to invest in disaster risk reduction and guide new policy tools to do so. Following this logic, increasing resilience can lead to greater equity while reducing the costs of disasters by prioritizing interventions in a way that mainstream DRM practices into wider sustainable development agendas. This model provides more in-depth insights into the consequences of disasters than a traditional risk assessment. In particular, it shows how regions (or specific areas) that are prioritized for DRM interventions can differ depending on the objective of decision-makers and on which risk metric is used since each metric informs a different set of policy objectives. While a simple cost- 52 | Conclusions and considerations for further analysis benefit analysis based on asset This study could scale up the investing in cost-effective risk losses would drive risk reduction analysis to additional countries reduction, developing financial investments toward asset-heavy with a similar disaster and protection solutions, and so regions and areas, a focus on factors socioeconomic profile, provided on), it would be worth exploring that increase poverty levels (such that household survey data (GMD if this methodology could be as well-being losses) broadens the dataset) and hazard data were applied to rapidly quantify well- analysis and can provide a different available and reliable. being losses in the immediate set of intervention priorities—for aftermath of a disaster—similar example, by anticipating which • Integrating more granular to the Global RApid post‑disaster income groups (or geographical data related to housing Damage Estimation (GRADE) areas) are likely to face larger conditions. The breadth and methodology.36 If it is feasible, socioeconomic obstacles in terms quality of household survey quantifying well-being losses in of post-disaster recovery and data was inconsistent across the immediate aftermath of a reconstruction. Finally, quantifying the selected countries, limiting disaster could be potentially used disaster impacts by using poverty the comparative perspective to better target post-disaster and well-being considerations can at times. For instance, future financing mechanisms as well also help quantify the benefits research could also include data as inform reconstruction efforts of interventions that may not on housing conditions to better and other related early recovery necessarily reduce asset losses examine how building damages interventions. but instead could reduce the well- can mitigate or accelerate well- being consequences of disasters being losses, depending on the by increasing the level of resilience level of socioeconomic resilience. of a given at-risk population. Such interventions could include, for • Including additional hazards example, financial inclusion, social to better understand protection, or more generally the distributional impacts in provision of timely and well-targeted rural areas. Depending on data post-disaster support to affected availability, further research households. could expand the scope to other hazards beyond earthquake and floods, such as extreme weather Follow-up work events (hail, frost, droughts) or landslides. Given that a Future research could further significant number of economies expand these initial findings from in the region rely significantly the eight countries that were on agriculture and that climate analyzed and confirm the pervasive shocks are expected to affect socioeconomic impact of disasters sectors that are particularly and climate risks in the ECA region. sensitive to weather patterns For instance, additional research on (irrigation, forestry, energy, and this issue could include some of the so on), expanding this analysis following items: to additional hazards will provide a more comprehensive • Expanding the analysis to 36 The GRADE approach is a remote, desk-based, understanding of the rapid damage assessment method deployed other similar countries. soon after a disaster. The approach adopts distributional impact of disasters The eight countries that were evolving and innovative natural hazard risk in the region. modeling technology to rapidly fulfill post- examined were selected given event damage assessment requirements. that they all have (a) a high level • Testing potential applications The GRADE assessment provides decision- makers with a first order of the economic of disaster exposure to floods for rapid post-disaster needs impact to gauge the magnitude of the and earthquakes, (b) a relatively assessments. While the insights event’s consequences, identify reconstruction priorities, provide information on geographic high level of income inequality, and the results generated by impacts, and inform on relative public versus and (c) an existing engagement this study are about improving private sector damages. For more information: https://www.gfdrr.org/en/publication/ in DRM with the World Bank and ex ante measures (such as methodology-note-global-rapid-post-disaster- with other development partners. damage-estimation-grade-approach-2018. ANNEX 1: Poverty trends in the ECA region Poverty remains a central issue in Figure 21: ECA population by income and consumption level as share of Europe and Central Asia. Poverty the whole, 1993–2015. In the ECA region, the middle class has significantly levels have significantly declined expanded in the last 20 years, but a large proportion of the population is still over the past 30 years, but those at risk of falling back in poverty. development gains are still fragile. 100% Almost all countries in the ECA region underwent a transition at the end of 80% the 20th century from various types of closed, planned economies to above 10$ a day more open, free market economies. 60% This transition, with the dissolution of trade networks and production shifts, was accompanied by a steep 40% increase in poverty and inequality within the region (World Bank 5.50 - 10$ a day 20% 2014a). Owing to rapid economic 3.20 - 5.50$ a day growth during the 2000s, the region 1.9 - 3.20$ a day experienced a rapid and dramatic 0% below 1.90$ a day 1993 1995 2000 2005 2010 2015 decline in poverty. Note: Estimates rely on a combination of income and consumption data. This was followed by a phase of Source: World Bank, PovcalNet 2019.a slower poverty reduction (figure 21), a. PovcalNet is an interactive computational tool that allows users to replicate the calculations made further constrained by repercussions by the World Bank’s researchers in estimating the extent of absolute poverty in the world. PovcalNet from the global financial and food also allows researchers to calculate the poverty measures under different assumptions and assemble crises that affected the world in the estimates using alternative economy groupings or for any set of individual economies of the user’s choosing. PovcalNet is self-contained; it has reliable built-in software that quickly does the 2008 and 2009. The extreme poverty relevant calculations for researchers from the built-in database. For more information: http:// rate, measured as US$1.90 a day per iresearch.worldbank.org/PovcalNet/home.aspx. person (2011 PPP), was 7.9 percent (equal to 37 million people) in 1999. the proportion of population living In 2018, the share of people living Almost half of the ECA population in households with consumption on US$5.50 or less per day in the was living in moderate poverty in or income per person below the region was 12.1 percent—equivalent 1999 (36.9 percent, equivalent to 172 poverty line (moderate poverty) to a population of 59.73 million million people), that is, on US$5.50 remained high in many countries (Bussolo et al. 2018). Moreover, in a day (2011 PPP), the majority in in the ECA region, ranging from line with differences across living Central Asia. By 2015, both moderate 2 percent (Russia) to 50 percent standards, sub-national poverty and extreme poverty rates had (Armenia).37 rates can vary significantly within declined, to 12.5 percent (61 million countries, particularly in ‘lagging people) and 1.5 percent (7.1 million 37 Poverty data are provided by World Bank regions’ (Farole, Goga, and Ionescu- people), respectively. As of 2018, PovcalNet: http://iresearch.worldbank.org/ Heroiu 2018). In Armenia, for PovcalNet/data.aspx. 54 | Poverty trends in the ECA region example, the difference in poverty Figure 22: In the ECA region, due to harsh winter conditions, the poorest and rates between the less well-off poor population groups tend to spend more on food (and energy) compared and the more well-off regions rose to other regions. from 25 percentage points to 38 80 percentage points between 2005 70 and 2014 (Bussolo et al. 2018). Percent of total expenditure 60 The middle class expanded 50 significantly across the ECA 40 region between 2000 and 2015. 30 Across many countries in the 20 region, a significant proportion of 10 households moved out of poverty into the middle class (especially 0 Sub-Saharan Eastern Europe, South Asia Middle East, East Asia, Latin America, during the period preceding Africa Central Asia North Africa Paci c Caribbean the 2008 financial crisis); more Poorest Poor Middle Wealthier importantly, once out of poverty, they tended not to fall back into Note: Calculated based on total consumption value in 2010 (US$ PPP values) in developing countries. it. As seen in figure 21, this trend Consumption groups defined based on global income distribution data: poorest = US$2.97 Per capita a day; poor = between US$2.97 And US$8.44 Per capita a day; middle = between US$8.44 And US$23.03 was especially evident before the Per capita a day; wealthier = above US$23.03 Per capita a day. 2008 financial crisis (Bussolo et Source: Hallegatte et al., 2018. al. 2018). Upward mobility in the majority of countries during the 2000s translated into a sizable At a household level, the ‘poorest’ only to stay warm but also to ensure expansion of the middle class in and ‘poor’ in the ECA region tend that they get the minimum required the region, measured as the share to spend more of their budget on number of calories to survive of people living on over US$10 a food and energy, making them winters (box 6). Paying for heating day. In 2015, almost half of the more vulnerable to spikes in food and food, which are essential to region’s population could be prices and to unanticipated weather survive the region’s winters, can considered middle class, a rapid extremes. 39 drive almost every decision made increase from the 17 percent in by poorest and poor households 2000 (Dávalos et al. 2016). Income Winters in ECA can be characterized in the region throughout the year inequality, measured by the Gini with temperatures as low as -30°C (World Bank 2014b). However, the coefficient, also declined in many (-22°F), sometimes as low as -45°C significant proportion of expenses countries. However, according to (-49°F) in the coldest parts of the also left the most vulnerable the World Bank, a large proportion region. Given these temperature households with little to no money of the overall population in the patterns, the average household given that the poorest and poor ECA region is still vulnerable and at in ECA spends over 7 percent of people in the region also devoted, risk of falling back into poverty— their income to pay for energy and 55 to 60 percent of their household corresponding to 70 percent of the food, compared to 4.7 percent in budget on food, on average (figure population in lower-middle-income the East Asia and Pacific region and 22). This means that other vital countries and 40 percent of the 4.6 percent in the Latin America expenditures, such as medical costs population in upper-middle-income and the Caribbean region. The or school fees often have to be put countries.38 severe cold in the region means off or eliminated altogether when that households need to spend a households face shocks, such as significant amount of money not a sudden spike in food prices or 38 Calculated using 2011 PPP, ECATSD (Europe higher energy consumption needs and Central Asia Team for Statistical Development) calculations using ECA (Dávalos et al. 2016). Poverty data (ECA POV) and EU-SILC 39 The World Bank defines consumption groups data; see http://ecadataportal. Note. The based on global income distribution data. European Union Statistics on Income and This means that while the ‘poorest’ group Living Conditions (EU-SILC) aims at collecting (also referred as ‘extreme poor’) earns timely and comparable cross-sectional and US$2.97 per capita a day, the ‘poor’ group longitudinal multidimensional microdata on (also referred to as ‘moderately poor’) earns income, poverty, social exclusion, and living between US$2.97 and US$8.44 per capita a conditions. day (Hallegatte et al. 2016). Poverty trends in the ECA region | 55 A man waits by the bus station in Sofia, Bulgaria. 2016. Credit: Ivelina Taushanova/World Bank Box 6: Energy poverty affects tens of millions in the EU and is particularly prevalent in Bulgaria. There are an estimated 50 million households living in energy poverty in Europe.a Energy poverty rates are highest in Southern and Central-Eastern European countries, but across the entire EU, those in the bottom 20 percent of the income distribution have considerably higher rates of energy poverty than national averages.b This difference is due to the limited disposable incomes of households in the bottom 20 percent, who spend a large share of their disposable income on heating their poor-quality, energy-inefficient homes. In other words, countries with low-quality housing tend to see higher levels of energy poverty, as well as higher at-risk-of-poverty rates, further food poverty (that is, the inability to afford basic food staples), and sometimes, higher rates of self-reported health issues. Unexpected and unpredictable energy costs in the winter can push more and more households below the poverty line during the winter season, and only some of them are able to receive a heating allowance. Bulgaria has the largest percentage of energy-poor people in the EU: 34 percent of its population of 8 million struggled to keep their homes sufficiently warm during winter months in 2018. A significant part of the problem is related to the quality of buildings. According to the National Program for Housing Renewal in Bulgaria adopted in 2005, over 20 percent of buildings are panel buildings, most of them needing renewal. In Bulgaria, energy poverty itself has over time become a barrier to energy efficiency programs: since poor households often cannot afford credits, they are unable to invest in energy efficiency. To prevent low-income households from becoming too exposed to shifts in energy prices and stagnating wages, governments also need to ensure that buildings occupied by low- income families are able to generate ‘positive energy’—produce more energy than they consume—to minimize low- income families’ energy bills. Sources : CEB 2019; Joyce 2017; Peneva 2019; Thomson and Bouzarovski 2018; World Bank 2020b. a. The term ‘energy poverty’ has no universal definition, but it typically means that households spend an unreasonably high proportion of their income on energy or that households are unable to meet basic energy needs in both winter and summer. The causes of energy poverty can be multidimensional and usually include low-income, low-quality homes, and energy-inefficient appliances. b. Nearly 85 percent of people in energy poverty can be found in 10 of 32 European states: Bulgaria, France, Germany, Greece, Italy, Poland, Portugal, Romania, Spain, and the United Kingdom. Drainage improvements in Azerbaijan. Credit: Allison Kwesell ANNEX 2: Methodological Summary Data description METH ANNEX 2.Github DOLOGIare Oresearchers and 40 CAinvited L SUMMThis ARY is interpreted as the value to contact the authors to discuss of assets necessary to generate Asset loss exceedance curves are methodological details and potential each household’s earned income, Data description adapted from “Europe and Central adaptations of the simulations here including assets not owned by Asia – Asset exceedance loss Risk Country curves are adapted Profiles for from “Europe and Central Asia Country presented. theRisk Profiles for household (like Floods roads and and Floods Earthquakes” (World (World and Earthquakes” Bank 2017). These assessments were generated for purpose, and technical factories). documentation keff generally includesis Bank provided 2017). Thesein the Asset losses technical annex of this report. assessments Risk data are available upon request. privately owned assets, including were Socioeconomic purpose, and generated for amicrodata A set of disaster events is housing and equipment used in including household income and expenditures are from the World Bank’s Global technical documentation defined, each parameterized family business or livestock; public Monitoring provided Databaseis(2019), a globally harmonized database of household income and expenditure surveys. in theSpatial technical annex is resolution this report.by the in ofdetermined four dimensions: of representativeness country, each country’s most recent assets, such as survey. road These and the power microdata Risk data cannotarebeavailable upon request. made available hazard directly, as they subjectearthquake], are:[flood, to use agreements with the countries’ grid (and possibly respective the environment statistical agencies. location:[admin1|admin2], and and natural capital); and some Socioeconomic microdata, return period (RP):[5,10,20,50,100, assets privately owned by others, Methodology including household and Modeland income Description 500,1000,2000] years). PMLs such as factories. ν represents a expenditures, are from the World expected from each of The Unbreakable methodology combines conventional disaster risk assessments with these rudimentary vulnerability social welfare curve, theory. This Bank’s Global Monitoring Database events were reported by GFDRR approach is summarized here, and described in full in Walsh and Hallegatte 2019 and in Walsh and Hallegatte equal to 70 percent, 40 percent, (2019), a globally 2020. Python scripts are provided oninGithub harmonized 2017 (World 44 Bank 2017).are and researchers These and 10 the invited to contact authors percent to discuss for households methodological details and potential adaptations of the simulations among database of household income PML values are distributed here presented.with fragile, moderate, and robust and expenditure surveys. Spatial representative households domiciles, respectively. In this way, Asset losses. A set of disaster events is defined, each parameterized in 4 dimensions: country, hazard:[flood, resolution is determined by according to the following the construction and condition earthquake], location:[admin1|admin2], and return period (RP):[5,10,20,50,100,500,1000,2000] years). Probable the representativeness of each expression, where the sum is over of domiciles is used to proxy the maximum asset losses (PML) expected from each of these events were reported by GFDRR in 2017 (World Bank country’s most recent survey. These all households exposed to the vulnerability of all assets used 2017). These PML values are distributed among representative households according to the following expression, microdata cannot be made available disaster event: where the sum is over all households exposed to the disaster event: by each household to generate directly, as they are subject to use income. This implies, for example, agreements with the countries’ "#$(&'&()) = -!"!#$ / ∙1 that the roads used by people . !%% respective statistical agencies. who live in makeshift dwellings are In this expression, - is the exposure (fraction of households affected), /!%% is the effective equally capital stock, vulnerable 1 the as to flooding vulnerability (fraction of assets lost, given household In this expression, ⊆ affected). /!%% is defined are α is the exposure their household for each homes. This asassumption earned Methodology income divided by the national average(fraction productivity of capital (Penn of households World Tables). avoids affected), This is interpreted as the value significant increases in data andof model description assets necessary to generate each household’s income, earned capital keff is the effective including stock, not owned by the household ν assetsrequirements—indeed, global data (like roads and factories). /!%% generally theincludes privately-owned vulnerability (fraction of assets, assets includingonhousing and equipment infrastructure used is not vulnerability The Unbreakable methodology in family business or livestock; public assets, such as road and lost, given household affected). the power grid (and possibly available—andthe environment is necessary and to avoid natural combines capital); and conventional some assets privately disaster owned by others, keff is defined for each household such as factories. 1 represents a rudimentary overtly complex representations vulnerability with risk assessments curve, equal to 70%, 40%, and 10% for households with fragile, moderate, social of economic and robust domiciles, interactions between as earned income divided by the welfarerespectively. theory. This In this is construction way, the approach and condition of domiciles is used to proxy the vulnerability of all assets national average productivity each household and assets held each and used by here summarized household to generate described in income. This implies, for example, that the roads used by people who live in of capital (Penn World Tables). in common. full inmakeshift Walsh and dwellings Hallegatte are equally vulnerable to flooding as is their home. This assumption avoids significant (2019) and in increases Walsh and data requirements—indeed, in Hallegatte (2020). global data on infrastructure vulnerability is not available—and is necessary to avoid Python scripts are provided on overtly complex representations economic offollowing 40 Available at the link: interactions https://github. between each household and assets held in common. com/walshb1/hh_resilience_model. Vulnerability modifiers: early warning systems and access to finance. When available, data on the presence of early warning systems in affected regions is incorporated. This reflects the assumption that early warning systems allow exposed households to move, reinforce, or otherwise protect their most fragile or valuable assets, thus 58 | Methodological Summary Vulnerability modifiers: early Figure 23: Each household’s effective asset stock includes private, public, and warning systems and access to other-owned assets. A fraction of these are damaged or destroyed in the event finance of a shock, and the household rebuilds its private stock exponentially, When available, data on the presence at a rate which maximizes utility of consumption during recovery. of early warning systems in affected regions is incorporated. This reflects the assumption that early warning systems allow exposed households to move, reinforce, or otherwise protect their most fragile or valuable assets, thus reducing their vulnerability to disaster. Using the same assumption as in the Unbreakable report (Hallegatte et al. 2017), it is assumed that households who receive a warning are able to reduce their vulnerability by 20 percent, relative to identical households without access to early warning systems, by moving valuable items (from important papers to car or motorbikes) and implementing other mitigating measures (for example, boarding windows, sandbagging doors). Exposure FIGURE 23. EACH HOUSEHOLD'S EFFECTIVE ASSET STOCK INCLUDES PRIVATE, PUBLIC, AND OTHER-OWNED ASSETS. SOME FRAC On these assumptions, the expression OF THESE ARE DAMAGED OR DESTROYED IN THE EVENT OF A SHOCK, AND THE HOUSEHOLD REBUILDS ITS PRIVATE S above can be solved for αevent. In this Pre-disaster equilibrium This model assumes a closed national EXPONENTIALLY, AT A RATE WHICH MAXIMIZES UTILITY OF CONSUMPTION DURING RECOVERY. approach, household exposure is Households begin the simulation economy, meaning that 100 percent Precautionary interpreted as the probability for in equilibrium: savings. without Precautionary capitalplay a key role savings of household income in managing butfrom is derived disasters, savings microdata any given household to be unavailable. decay and between affected The difference consumption income equal set and assets consumption, located inside frequently usedtheas country a proxy,and is highly variable by an event, when it occurs. This to income. The survey is assumed that post-disaster reconstruction negative for many households (aggregate consumption greater than reported income), making it an unce indicator probability is assumed uniform to at for of savings capture the household level. In this case, the costs each household’s cangap average be distributed (income less to consumption) non- by region decile all households, due to lack annual income. of anis calculated. It is assumed that each household maintains one year’s surplus as 41 Earned income affected taxpayers throughout the precautionary sav separate from their established empirical relationship includes labor, productive capital, assets, andrealavailable country estate, to be spent but not outside on recovery its borders or consumption smoothing. between income and exposure. and in-kind sources and excludes (figure 23). found transfers, If low‑income groups areSocial to taxation, and remittances. public transfers (conditional, Reported incomes are assumed net of the income tax that fina general be more likely to be affected, it is spending of the government unconditional, and in-kind) andan additional and of tax that finances social progra flat income savings Precautionary Extraordinary possible to introduce a ‘poverty bias’ or adaptive social transfers private remittances. Note that the are represented as closed Precautionary(revenue savingsneutral) systems in the mod play a key that their in the form of a higher probability of costs are distributed to all households value of housing services provided (debited role in managing disasters, butvia flat tax.48 from nominal income) nationally being affected for households with by owner-occupied dwelling savings microdata are unavailable. is amount received, The household surveys provide estimates of the but it is impossible to represent the bila lower income. included in income data, so that the The difference between income flows of resources among households. For this reason, remittances are modeled like an additional s loss oftransfers housing services is recorded and consumption, frequently used protection scheme: the received42 from friends and family are added to social transfers, and it is assu Two corollaries follow from the above: as an impact ona income. as a proxy, is highly variable and that these transfers comes from single fund, in which all households contribute proportionally to their inc first, larger disasters affect greater negative for many households (like a flat tax). Under these assumptions, remittances can be aggregated with social protection and redistribu numbers of people, with constant (aggregate consumption greater systems. This is of course a simplification, especially in that it does not account for international remitta severity among the affected. Second, 41 In some countries, it may be necessary to than reported income), making it which have been shown role in to play aincome infer household expenditures, contexts (Yang et al. 2007). post-disaster from this definition of household income an uncertain indicator of savings whether because incomes are not reported, and assets implies that the total capital because consumption is more stable over at the household level. In this time, or because official poverty statistics case, the average gap (income less stock of any country is given by total are calculated from consumption rather effective capital of all households, than income. consumption) by region and decile 47 Although a flat tax is assumed, the model is capable of handling more complicated tax regimes, including progressive taxation. although gross national income (GNI) costs are not included in the assessment of the cost of is 48 Administrative 42 Similarly, services provided by other assets thecalculated. It is assumed programs. When household that each data do not include the tran (for example, air conditioners, refrigerators) is generally lower than GDP.then transfers from social programs could be added be canas anmodeled additionalon the basis of the actual disbursement rules that qualify households for partici income household maintains one year’s in each program (e.g. PMT score, that household can be number threatened of dependents or senior citizens, employment status, etc.). by disasters. Methodological Summary | 59 surplus as precautionary savings: Figure 24: Consumption losses are defined as income loss, net of reconstruction separate from their productive costs. Disaster-affected households must decide which fraction of their post- assets and available to be spent on disaster income to divert from consumption into reconstruction, in anticipation recovery or consumption smoothing. of recovery to their pre-disaster state. Consumption losses are offset by expenditure of savings and any post-disaster support a household may receive. Social transfers, taxation, and remittances Reported incomes are assumed net of the income tax that finances general spending of the government and of an additional flat income tax that finances social programs.43 Extraordinary or adaptive social transfers are represented as closed (revenue neutral) systems in the model, in that their costs are distributed to all households (debited from nominal income) nationally via flat tax.44 The household surveys provide estimates of the amount received, but it is impossible to represent the bilateral flows of resources among households. For this reason, remittances are modeled like an additional social protection FIGURE 24. CONSUMPTION LOSSES ARE DEFINED AS INCOME LOSS, NET OF RECONSTRUCTION COSTS. DISASTER-AFFECTED scheme: the transfers received from HOUSEHOLDS MUST DECIDE WHICH FRACTION OF THEIR POST-DISASTER INCOME TO DIVERT FROM CONSUMPTION INTO Income losses Consumption losses friends and family are added to RECONSTRUCTION, IN ANTICIPATION OF RECOVERY TO THEIR PRE-DISASTER STATE. CONSUMPTION LOSSES ARE OFFSET BY social transfers, and it is assumed EXPENDITURE When OF households SAVINGS AND ANY lose POST - assets, DISASTER they SUPPORT A The reconstruction HOUSEHOLD MAY RECEIVE. of private and that these transfers comes from Savings a lose the income these assets had public assets is not free: households and post-disaster support. Households’ precautionary savings, increases by post-disaster support, single fund, in which all households generated unless and until the and governments have potentially including cash transfers to affected households, increases in social protection transfers, to invest in help through contribute proportionally to their assets are restored to productivity. the reconstruction, informal mechanisms at the community level, potential increases in remittances, and other further reducing exceptional cash income (like a flat tax). Under these transfers to This includes households the value following households a disaster. When available, personal consumption these resources and requiring help households to smooth their consumption over time, derive from and decrease their domicile; their losses. authorities to reallocate or generate consumption assumptions, remittances can be aggregated with social protection Private assetappliances, reconstruction. vehicles, and livestock; To recover additional their pre-disaster incomes, funds. affected The model household pursue exponential asset and redistribution systems. This is and reconstruction the infrastructure pathways. Each diverts they use income, plus savings represents and post-disasterthese processes support, into to reconstruction at a rate of course a simplification, especially that maximizes to its expected commute to well-being work or over access the 10 years following estimatethe disaster. their In this longitudinal way, households impacts use their income, savings, markets.and post-disaster Income support losses are to smooth consumption defined on and maximize consumption, utility, subject well-being, andto the constraint in that it does not account for of avoiding subsistence if possible. If households cannot avoid having their consumption level below the international remittances, which subsistencefor each household as asset losses poverty. It is important to note that line (for instance because consumption is below the subsistence level even without repairing and have been shown to play a role in times national average productivity replacing lost assets), then it is assumed that reconstruction the model takes place assumes at the pacethat households possible with a saving rate post-disaster contexts (Yang and equal to the of capital. average From saving this rate of initial and governments people living at or below subsistence aim level in each at returning country. Choi 2007). condition, asset and income losses to the pre-disaster equilibrium Public asset reconstruction. When disasters occur, governments borrow externally to finance the cost of public become time dependent. Initial described above. asset reconstruction and recovers these costs through a general flat tax when recovery is complete. Through this losses decrease throughout the mechanism, all households throughout the country share the cost of public asset reconstruction in the affected 43 Although a flat tax is assumed, the model is reconstruction area. It is assumed that that and recovery process the government Total does not collect thereconstruction special tax at any costs point are equal during recovery, but capable of handling more complicated tax regimes, including progressive taxation. rather coversas houses, the cost ofinfrastructure public asset (that is, reconstruction for the to the reduction duration of in consumption reconstruction and collects taxes to fund the process 44 Administrative costs are not included in this many years roads later, after or electric fulland lines), recovery. natural needed to rebuild households’ assessment of the cost of the programs. assets are repaired and replaced. asset stock, plus the increase in When household data do not include the transfers, then transfers from social programs Total asset losses are inclusive of taxes needed for the government 52 can be modeled on the basis of the actual to rebuild public assets such as disbursement rules that qualify households all asset classes, irrespective of for participation in each program (for ownership (private, public, and roads and water infrastructure. example, proxy means test score, household’s The contribution of reconstruction number of dependents or senior citizens, other). These assets are rebuilt employment status, and so on). independently and at different rates. costs to consumption losses at Well-being losses. Finally, cons (CRRA) welfare function ( = 1 60 | Methodological Summary affects poor households more se Well-being losses. Finally, consumption losses are translated into utility (U) using a c (CRRA) welfare function (3 = 1.5). This concave function accounts for the fact that e each moment depends on the affects poor households ownership of the damaged assets more severely than wealthy post-disaster support to smooth consumption and maximize Well-being losses (W) are defined ones: as the future-discounted time and on reconstruction rate (figure utility, subject to the constraint of integral of utility over 10 years 24). These two dimensions will avoiding subsistence if possible. Well-being 8()) losses (W) are defin &'( after a disaster. Expressed as 7()) = 1 be discussed next. Consumption losses due to reconstruction costs If households cannot avoid having their consumption level :as− 3 currency, where Expressed consumption, currency,is national mean where ̅ vary by asset type (that is, private, below the subsistence line (for Well-being losses (W) are defined as the future-discounted10 time integral of utility o public, or other). ∫0 − instance because consumption is = Expressed as currency, where 8 ̅ is national mean consumption: below the subsistence level even ⁄ a. Affected households pay directly without repairing and replacing ̅ replacement of and entirely the $% lost assets), then it is assumed *! !"# +$ the lost assets ∫ < = % they owned. that ,* Following from this constructio that reconstruction takes place Following from this construction, at the pace possible with a saving each household reconstructs at a /,- ̅ b. All households pay indirectly and proportionally to their optimal value of satisfies th rate equal to the average saving rate which maximizes its well-being. rate of people living at or below The optimal value of τh satisfies incomeFollowing fromofthis construction, each household reconstructs at a rate which m for the replacement subsistence level in each country. the following and can be found lost public assets through an optimal value of =0 satisfies extraordinary tax. the following, and can be found numerically: Public asset reconstruction numerically: When disasters occur, governments c. Households do not pay for the borrow externally to finance the >< replacement of the assets they cost of public asset reconstruction Replenishment =0 of household sa use to generate an income but >=0 do not own (such as the factory and recover these costs through a general flat tax when recovery is are assumed widely distributed where they work). Replenishment of household savings and the taxes Replenishment complete. Through this mechanism, that fund public complete. asset of Therefore, household reconstruct the well-b savings all households throughout the Savingsare assumed widely distributed and far in the future, the so they reduce consumption and the taxes that fund public asset support and post-disaster country share the cost of public marginal utility of reconstruction and post-disaster consumpt complete. Households’ Therefore, precautionary savings the well-being impact of savings asset reconstruction in the affected of expenditures supportthe areloss assumedalong and widelythePDS-related reconstruct area. It is assumed that that the distributed and far in the future, the marginal utility of consumption of each household. increase with post-disaster support, government does not collect the Total well-being losses of a ho so they reduce consumption only potentially including cash transfers ofhouseholds, to affected along the reconstruction path and the long-term the loss increases special tax at any point during Robustness losses. after reconstruction assessment. is complete. This a recovery but rather covers the cost in social protection transfers, help of public asset reconstruction for inter-comparisons among cou Therefore, the well-being impact of savings expenditures and taxes Robustness through informal mechanisms assessment. at the community level, potential This analysis relies on many the duration of reconstruction and simplifying subnational related scales. to post-disaster assumptions. supportAnalytics can Wh pre increasesinter-comparisons in remittances, and among countries, they limit the collects taxes to fund this process scope be estimated to provide many years later, after full recovery. usingand precision the marginal a starting of its point for st other exceptional cash transfers utility of consumption of each subnational scales. to households following a disaster. Analytics presented here are intended to illustrate household. Total well-being losses relevant dyna Well-being losses to provide When available, a starting point these resources for Finally, strategic consumption and losses aretechnical Many decision significant of a household are equal to the sum processes. uncertainties n of the loss along the reconstruction help households smooth their translated into utility (U) using path and the long-term losses. consumption over time and decrease consumption losses. • public/private Many significant uncertainties need to be incorporated for robust asset decision making, a constant relative risk aversion in (CRRA) welfare function (η=1.5). This Robustness assessment Well-being Private losses. asset reconstruction concave function • poverty bias of expo accounts for the This analysis relies on many • public/private asset allocation; Finally, consumption losses are translated into utility (U) using fact that each unit of consumption a constant relative risk aversion simplifying assumptions. While To recover their pre-disaster (CRRA) welfare function (3 = 1.5). This concave function • asset-vulnerability c accounts for more the fact that each unit of consumption loss • poverty bias of exposure; incomes, affected households loss affects poor households these facilitate suggestive inter- affects poor households pursue exponentialmore severely than wealthy asset ones: severely than • survey recency, acc wealthy ones: comparisons among countries, • asset-vulnerability catalogue; reconstruction pathways. Each they limit the scope and precision 8()) &'( • precautionary savin of its conclusions, particularly • survey recency, accuracy & representativeness; diverts income, plus savings and post-disaster support, into 7()) = : 1 − 3 at sub-national scales. Analytics • precautionary savings & informal coping mechanisms; reconstruction at a rate that • income substitution presented here are intended to Well-being losses (W ) are defined as the future-discounted time integral of utility over 10 years afteranda disaster. maximizes its expected well-being • income substitution; • non-ergodicity of st illustrate relevant considerations and dynamics to provide a as currency, Expressedover the 10 years where 8 ̅ is following national mean consumption: the disaster. In this way, households • non-ergodicity of stochastic idiosyncratic/compound • macroeconomic shocks; dyn starting point for strategic and $% use *! !"# +$ income, savings, and their technical decision processes. < = ∫% ,* • macroeconomic dynamics. /,- ̅ Potential applications. The prim Following from this construction, Potential each household applications. The primary ofathe use at reconstructs data which for methodology maximizes rateUnbreakable disaster response isThe its well-being. to targe synth Methodological Summary | 61 Many significant uncertainties need to be incorporated for robust decision making, including the following: • Public/private asset allocation; • Poverty bias of exposure; • Asset-vulnerability catalogue; • Survey recency, accuracy, and representativeness • Precautionary savings and informal coping mechanisms • Income substitution; • Non-ergodicity of stochastic idiosyncratic/compound shocks; • Macroeconomic dynamics. Potential applications The primary use of the Unbreakable methodology is to synthesize risk and socioeconomic data for disaster response targeting. Many significant indicators of individual frailty and systems failure are visible ex ante, before a catastrophe occurs, and the benefits of contingency planning and risk management far exceed their costs. Social registries, employment programs, beneficiary lists, and budget allocations can all be set in advance, triggering as soon as possible to help communities and nations cope and recover efficiently and equitably. Search and Rescue team search for the wounded in Van (Turkey) after the October 2011 earthquake. Source: hasanucarphotography/Shutterstock.com 62 | References References • Agrawal, Raj, Stephen Findley, Sean Greene, • CEB (Council of Europe Development • ESA (European Space Agency). 2019. Kathryn Huang, Aly Jeddy, William W. Lewis, Bank). 2019. Energy Poverty in Europe: “Extreme Heatwave.” July 26, 2019. and Markus Petry. 1996. “Why The US Leads How Energy Efficiency and Renewables Can https://www.esa.int/ESA_Multimedia/ and Why It Matters”. The McKinsey Quarterly Help. Paris: Council of Europe Development Images/2019/07/Extreme_heatwave. Summer 199, Number 3. http://www2. Bank. https://coebank.org/en/news-and- lawrence.edu/fast/finklerm/McKinsey_cap_ publications/news/ceb-publishes-study- • Euronews. 2019. “Faced with Increasing prod_9603.pdf. energy-poverty-europe/. Heat Waves Frequency, European Cities Need to Adapt.” November 22, 2019. https:// • Anton, Simona, Koo Bryan, Man Titus- • Chakraborty, Tirthankar, Angel Hsu, www.euronews.com/2019/09/23/faced-with- Cristian, Sandu Ciprian Moldova, Manuela Diego C. Manya, and Glenn Sheriff. increasing-heat-waves-frequency-european- Sofia Stanculescu, and Robertus A. Swinkels. 2019. “Disproportionately Higher cities-need-to-adapt. 2014. Elaboration of Integration Strategies Exposure to Urban Heat in Lower- for Urban Marginalized Communities: The Income Neighborhoods: A Multi-City • Farole, Thomas, Soraya Goga, and Marcel Atlas of Urban Marginalized Communities Perspective.” Environmental Research Ionescu-Heroiu. 2018. “Rethinking Lagging in Romania. Washington, DC: World Letters 14 (10). https://iopscience. Regions: Using Cohesion Policy to Deliver Bank. http://documents.worldbank. iop.org/article/10.1088/1748-9326/ on the Potential of Europe’s Regions: org/curated/en/668531468104952916/ ab3b99#erlab3b99app1. Overview.” World Bank, Washington, DC. Elaboration-of-integration-strategies-for- http://documents.worldbank.org/curated/ urban-marginalized-communities-the-atlas- • Daniell, L., and A. Schaefer. 2014. Eastern en/636851525400071277/Overview. of-urban-marginalized-communities-in- Europe and Central Asia Region Earthquake Romania. Risk Assessment Country and Province • Fuchs Tarlovsky, Alan, Cesar A. Cancho, Profiling. Final report to GFDRR. Ana Maria Munoz Boudet, Sailesh Tiwari, • Bogaerts, Vica Rosario. 2020. Project Natsuko Kiso Nozaki, Moritz Meyer, Xinxin Information Document – Earthquake Recovery • Dávalos, María E., and Thomo, Ledia. 2016. Lyu, Maria Fernanda Gonzalez Icaza, and Public Health Preparedness Project “Portraits of Poverty and Inequality in Osman Kaan Inan, Akhmad Rizal Shidiq, – P173998 (English). Washington, D.C. : Albania (English).” Washington, DC: World Matias Ignacio Morales Cerda, Vikram World Bank. http://documents.worldbank. Bank. http://documents.worldbank.org/ Srinivas, and Isaura Espinosa De Los org/curated/en/208431590094404051/ curated/en/659991480409319302/Portraits- Monteros. 2019. South Caucasus in Motion Project-Information-Document-Earthquake- of-poverty-and-inequality-in-Albania. (English). Washington, DC: World Bank. Recovery-and-Public-Health-Preparedness- http://documents.worldbank.org/curated/ Project-P173998. • Dávalos, María E., Giorgia Demarchi, Indhira en/614351556553124178/South-Caucasus- Vanessa Santos, Barbara Viony Kits, and Isil in-Motion. • Bussolo, Maurizio, Maria Eugenia Dávalos, Oral Savonitto. 2016. Voices of Europe and Vito Peragine, and Rayma Sundaram. Central Asia : New Insights on Shared Prosperity • Guha-Sapir, D., R. Below, and Ph. Hoyois 2018. Toward a New Social Contract: Taking and Jobs. Washington, DC: World Bank. – EM-DAT: The CRED/OFDA International on Distributional Tensions in Europe and http://documents.worldbank.org/curated/ Disaster Database – www.emdat.be – Central Asia. Washington, DC: World Bank. en/372521478680637391/Main-report. Université Catholique de Louvain – Brussels http://documents.worldbank.org/curated/ – Belgium. en/569321538150509492/Toward-a-New- • EEA (European Environment Agency). 2019. Social-Contract-Taking-on-Distributional- “Economic Losses from Climate-Related • GFDRR (Global Facility for Disaster Reduction Tensions-in-Europe-and-Central-Asia. Extremes in Europe.” EEA, Copenhagen. and Recovery). 2014. Understanding Risk in https://www.eea.europa.eu/data-and-maps/ an Evolving World. Global Facility for Disaster • Carter, Michael R., and Christopher B. indicators/direct-losses-from-weather- Reduction and Recovery, World Bank, Barrett. 2006. “The Economics of Poverty disasters-3/assessment-2. Washington, DC. https://www.gfdrr.org/ Traps and Persistent Poverty: An Asset- sites/gfdrr/files/publication/Understanding_ Based Approach.” Journal of Development • Erdik, M., Y. Fahjan, O. Ozel, H. Alcik, A. Mert, Risk-Web_Version-rev_1.8.0.pdf. Studies 42 (2). and M. Gul. 2003. “Istanbul Earthquake Rapid Response and the Early Warning • Government of Albania, World Bank, United • Carter, Michael R., Peter D. Little, Tewodaj System.” Bulletin of Earthquake Engineering Nations Development Programme, and Mogues, and Workneh Negatu. 2007. “Poverty 1: 157–63. https://link.springer.com/ European Commission. 2020. Albania: Traps and Natural Disasters in Ethiopia and article/10.1023%2FA%3A1024813612271. Post-Disaster Needs Assessment. Volume A Honduras.” World Development 35 (5). Report. https://eeas.europa.eu/delegations/ albania/74065/release-post-disaster-needs- assessment-albania-recovery_en. References | 63 • Government of Bulgaria. 2018. “Advisory • Kahn, Matthew .2005. “The Death Toll from • Restrepo Cadavid, Paula, Grace Cineas, Services on a Climate Change Adaptation Natural Disasters: The Role of Income, Luis E. Quintero, and Sofia Zhukova. Strategy and Action Plan— Assessment Geography, and Institutions.” Rev Econ Stat 2017. “Cities in Europe and Central Asia: of the Disaster Risk Management Sector 87 (2): 271–284. A Shifting Story of Urban Growth and (Draft Version).” Sofia, Bulgaria. https://www. Decline.” World Bank, Washington, DC. moew.government.bg/static/media/ups/ • Markhvida, Maryia, Brian Walsh, Stephane http://documents.worldbank.org/curated/ articles/attachments/DRM%20-%20Full%20 Hallegatte, and Jack Baker. 2020. en/319131510892209158/Cities-in-Europe- Report%20-%20First%20Draft%20(2018-04- “Quantification of Disaster Impacts through and-Central-Asia-a-shifting-story-of-urban- 27)%20-%20EN%20-%20for%20printing%20 Households Well-being Losses”. Nature growth-and-decline. v2675c5b0db190d4aa5dae48daf147c909.pdf. Sustainability, March 30, 2020. https://www. nature.com/articles/s41893-020-0508-7 • Saheb, Yamina. 2018. “Keeping Cool in a • Government of Serbia, World Bank, United Warming World Is Not a ‘Luxury’ for the Nations Development Programme, and • Mathema, Ashna, and Alanna Simpson. Rich”. Euractiv, August 6, 2018. https:// European Commission. 2014. Serbia Floods: 2018. “On Shaky Ground: Housing in Europe www.euractiv.com/section/energy/opinion/ Recovery Needs Assessment. July. https://www. and Central Asia.” Eurasian Perspectives keeping-cool-in-a-warming-world-is-not-a- recoveryplatform.org/assets/publication/ (blog), August 23, 2018. https://blogs. luxury-for-the-rich/. Serbiafloods2014/serbia-rna-report.pdf. worldbank.org/europeandcentralasia/shaky- ground-housing-europe-and-central-asia. • Saheb, Yamina. 2019. “Heatwaves Hurt the • Guerreiro, Selma B., Richard J. Dawson, Poor Most: Time for An EU Plan”. Euractiv, Chris Kilsby, Elizabeth Lewis, and • McGregor, Glenn, Mark Pelling, Tanja July 17, 2019. https://www.euractiv.com/ Alistair Ford. 2018. “Future Heat-Waves, Wolf, and Simon Gosling. 2007. “The section/energy/opinion/heatwaves-hurt-the- Droughts and Floods in 571 European Social Impacts of Heat Waves.” Science poor-most-time-for-an-eu-plan/. cities.” Environmental Research Letters 13 Report–SC20061/SR6. Environment Agency, (3): 034009. https://iopscience.iop.org/ Almondsbury, Bristol, UK. https://www. • Sandu, Dumitru D., Bogdan Corad, article/10.1088/1748-9326/aaaad3. researchgate.net/publication/45513548_ Cătălina Iamandi-Cioinaru, Titus-Cristian The_Social_Impacts_Of_Heat_Waves. Man, Monica Marin, Ciprian Moldovan, • Hallegatte, Stephane; Bangalore, Mook; Georgiana Neculau, Emil Daniel Tesliuc, Bonzanigo, Laura; Fay, Marianne; Kane, • Michelon, T., P. Magne, and F. Simon- Vlad Alexandru Grigoras, and Manuela Sofia Tamaro; Narloch, Ulf; Rozenberg, Julie; Delavelle. 2005. “Lessons of the 2003 Stanculescu. 2016. “Atlas Marginalized Rural Treguer, David; Vogt-Schilb, Adrien. 2016. Heat-Wave in France and Action Taken to Areas and Local Human Development in Shock Waves : Managing the Impacts of Limit the Effects of Future Heat-Waves.“ Romania (English).” Washington, DC: World Climate Change on Poverty. Climate Change In Extreme Weather Events and Public Bank. http://documents.worldbank.org/ and Development;. Washington, DC: World Health Responses, edited by W. Kirch, R. curated/en/847151467202306122/Atlas- Bank. https://openknowledge.worldbank. Bertollini, and B. Menne. Berlin, Heidelberg: marginalized-rural-areas-and-local-human- org/handle/10986/22787 Springer. https://link.springer.com/ development-in-Romania. chapter/10.1007/3-540-28862-7_13. • Hallegatte, Stephane, Adrien Vogt-Schilb, • Stadtherr, Lisa, Dim Coumou, Vladimir Mook Bangalore, and Julie Rozenberg. 2017. • Minority Rights Group International. Petoukhov, Stefan Petri, and Stefan Unbreakable: Building the Resilience of the 2018. “World Directory of Minorities and Rahmstorf. 2016. “Record Balkan Floods of Poor in the Face of Natural Disasters. Climate Indigenous Peoples—Bulgaria: Roma 2014 Linked to Planetary Wave Resonance.” Change and Development, Washington, Populations.” July. https://www.refworld.org/ Sciences Advances 2 (4). https://advances. DC: World Bank. https://openknowledge. docid/49749d463c.html. sciencemag.org/content/2/4/e1501428. worldbank.org/handle/10986/25335. • Morduch, Jonathan. 1995. “Income • Schumacher, I., and E. Strobl. 2011. • Hallegatte, Stéphane, Jun Rentschler, and Smoothing and Consumption Smoothing”. “Economic Development and Losses Due Brian Walsh. 2018. Building Back Better Journal of Economic Perspectives 9 (3): to Natural Disasters: The Role of Hazard : Achieving Resilience through Stronger, 103–14. Exposure. Ecol Econ 72: 97–105. Faster, and More Inclusive Post-Disaster Reconstruction. World Bank, Washington, • Naydenova, Violeta. 2014. “A Flood Lays Bare • Townsend, Robert M. 1995. “Consumption DC. https://openknowledge.worldbank.org/ Inequality in Bulgaria.” Voices (blog). July 14, Insurance: An Evaluation of Risk-Bearing handle/10986/29867. 2014. https://www.opensocietyfoundations. Systems in Low-Income Economies.” Journal org/voices/flood-lays-bare-inequality- of Economic Perspectives 9 (3): 83–102. • Ionova, Mariana. 2014. “Lost in the Floods.” bulgaria. Transitions Online, July 24, 2014. https://www. • Thomson, Harriet, and Stefan Bouzarovski. tol.org/client/article/24397-lost-in-the-flood. • Peneva, Teodora. 2019. “Energy Poverty in 2018. Addressing Energy Poverty in the html?print. Bulgaria.” Annual of University of Sofia 17 European Union: State of Play and Action. (11): 187–210. http://www.feba.uni-sofia.bg/ EU Energy Poverty Observatory. August. • Joyce, Adrian. 2017. “Energy Poverty: A Blight sko/yrbook/Yearbook17-11.pdf. https://www.energypoverty.eu/sites/ on the EU.” Euractiv, April 6, 2017. https:// default/files/downloads/publications/18-08/ www.euractiv.com/section/energy/opinion/ • Pollner, John Daniel, Jolanta Kryspin-Watson, paneureport2018_final_v3.pdf. energy-poverty-a-blight-on-the-eu/. and Sonja Ingrid Nieuwejaar. 2008. Climate Change Adaptation in Europe and Central • United Nations. 2015. Sendai Framework for • Joyce, Adrian. 2019. “Time to Address Asia (ECA) : Disaster Risk Management Disaster Risk Reduction 2015-2030. Available Europe’s Hot and Cold Homes Crisis.” (English). Washington, DC: World Bank. via UN Office for Disaster Risk Reduction: Euractiv, February 20, 2019. https://www. http://documents.worldbank.org/curated/ https://www.undrr.org/implementing- euractiv.com/section/energy/opinion/time-to- en/699781484888380512/Climate-change- sendai-framework/what-sf. address-europes-hot-and-cold-homes-crisis/. adaptation-in-Europe-and-Central-Asia-ECA- disaster-risk-management. 64 | References • UNDP (United Nations Development • World Bank. 2017. “Europe and Central Programme). 2016. Risk-Proofing the Western Asia – Country Risk Profiles for Floods and Balkans: Empowering People to Prevent Earthquakes (English).” Washington, DC: Disasters. Human Development Report. World Bank. http://documents.worldbank. http://hdr.undp.org/sites/default/files/risk_ org/curated/en/958801481798204368/ proofing_the_western_balkans.pdf. Europe-and-Central-Asia-Country-risk- profiles-for-floods-and-earthquakes. • UNDRR. 2018. Technical Guidance for Monitoring and Reporting on Progress in • World Bank. 2019. “Poverty and Equity Brief– Achieving the Global Targets of the Sendai Albania.” World Bank, Washington, DC. https:// Framework for Disaster Risk Reduction. databank.worldbank.org/data/download/ https://www.undrr.org/publication/technical- poverty/33EF03BB-9722-4AE2-ABC7- guidance-monitoring-and-reporting- AA2972D68AFE/Global_POVEQ_ALB.pdf. progress-achieving-global-targets-sendai. • World Bank. 2020a. Analysis of Heat Waves • UNECE (United Nations Economic and Urban Heat Island Effects in Central Commission for Europe). 2016. Housing European Cities and Implications for Urban and Urban Development in the Economic Planning. Washington, DC: World Bank. Commission for Europe Region: Towards a https://openknowledge.worldbank.org/ City-Focused, People-Centered and Integrated handle/10986/34335 Approach to the New Urban Agenda. Habitat III Regional Report. https://www.unece.org/ • World Bank. 2020b. Earthquake Risk in fileadmin/DAM/hlm/projects/HIII_Regional_ Multifamily Residential Buildings in the Europe Report/HabitatIII-Regional-Report-Europe- and Central Asia Regions. Washington, Region.pdf. DC: World Bank. https://openknowledge. worldbank.org/handle/10986/34439 • UNEP (United Nations Environment Programme). 2004. “Impacts of Summer • World Bank. 2020c. Croatia March 2020 2003 Heat Wave in Europe.” Environment - Earthquake Rapid Damage and Needs Alert Bulletin 2. http://www.grid.unep.ch/ Assessment 2020 (English). Washington, D.C. : product/publication/download/ew_heat_ World Bank Group. wave.en.pdf. • World Bank. 2021. Croatia December 2020 • Walsh, Brian, and Stephane Hallegatte. Earthquake Rapid Damage And Needs 2019. “Measuring Natural Risks in the Assessment (English). Washington, D.C. : Philippines: Socioeconomic Resilience and World Bank Group. Well-being Losses.” Policy Research Working Paper 8723, World Bank, Washington, DC. • Yang, Dean, and HwaJung Choi. 2007. “Are https://openknowledge.worldbank.org/ Remittances Insurance? Evidence from handle/10986/31227. Rainfall Shocks in the Philippines.” The World Bank Economic Review 21 (2): 219–248. • Walsh, Brian, and Stephane Hallegatte. https://www.jstor.org/stable/40282243 2020. “Measuring Natural Risks in the Philippines: Socioeconomic Resilience and Wellbeing Losses.” Economics of Disasters and Climate Change (2020): 249–293. https:// link.springer.com/article/10.1007/s41885- 019-00047-x. • World Bank. 2012. Turn Down the Heat: Why a 4°C Warmer World Must Be Avoided. Washington, DC: World Bank. https://openknowledge.worldbank.org/ handle/10986/11860. • World Bank. 2014a. Turn Down the Heat: Confronting the New Climate Normal. Vol. 2: Main Report. Washington, DC: World Bank. http://documents.worldbank.org/curated/ en/317301468242098870/Main-report. • World Bank. 2014b. “The Face of Poverty in Europe and Central Asia.” February 10, 2014. https://www.worldbank.org/en/news/ FACING PAGE feature/2014/02/10/face-of-poverty-in- Rescue workers searching through europe-and-central-asia. assorted debris after floods in Tbilisi, Georgia in 2015. Credit: Dmytro Vietrov ABOUT THIS REPORT ABOUT GFDRR This report presents a disaster risk assessment model The Global Facility for Disaster Reduction and Recovery for economic impacts of disasters in select countries in (GFDRR) is a global partnership that helps developing the Europe and Central Asia region—Albania, Armenia, countries better understand and reduce their Bulgaria, Croatia, Georgia, Greece, Romania, and Turkey. vulnerabilities to natural hazards and adapt to climate This disaster risk assessment approach adds a new change. Working with over 400 local, national, regional, dimension—that is, socioeconomic resilience—to the and international partners, GFDRR provides grant conventional risk assessment framework—which consists financing, technical assistance, training, and knowledge of hazard, exposure, and vulnerability. The socioeconomic sharing activities to mainstream disaster and climate resilience represents the ability of affected households to risk management in policies and strategies. Managed by cope with and recover from disasters. the World Bank, GFDRR is supported by 34 countries and 9 international organizations. The results indicate that the economic well-being of citizens in all these countries is affected far more than the estimated cost of physical damages to buildings and public infrastructure. This study also shows that the recovery and reconstruction process depends not only on the extent of damages due to disasters but also on the economic structure of each country and the level of socioeconomic resilience of its citizens. www.gfdrr.org