Economic

Losses,
Poverty &
DISASTERS
1998-2017
 Economic
Losses,
Poverty &
DISASTERS
1998-2017
Foreword
If development and economic growth are not risk informed, they are not sustainable and
can undermine efforts to build resilience. The economic losses which often ensue from
the creation of new risk or exacerbation of existing levels of risk can have a significant
human cost.

That human cost is there for all of us to see in the alarming numbers of people who are now internally
displaced every year by disasters, often losing their homes and their livelihoods, in extreme weather
events and earthquakes.

We live in a world where the bar for resilience is constantly being raised by human actions. The
most egregious failure in this regard is the lack of political will and commitment to make serious
progress on reducing greenhouse gas emissions, thus allowing climate change to play an increasingly
important role in driving up disaster losses around the world for the foreseeable future.

As the UN Secretary-General has recently warned: “If we do not change course by 2020, we risk
missing the point where we can avoid runaway climate change, with disastrous consequences for
people and all the natural systems that sustain us.”

Alongside the global push by the UN and other stakeholders for reductions in greenhouse gas
emissions, there is also widespread recognition that we need to accelerate implementation of the
Sendai Framework for Disaster Risk Reduction 2015-2030, the global plan to reduce disaster losses
by reducing existing levels of risk, avoiding the creation of new risk and managing risks that cannot
be eliminated.

These words must translate into actions which lead to robust institutions tasked with disaster risk
management, enforcement of land use regulations, implementation of building codes, preservation
of protective eco-systems, risk-informed urban development and special attention to the housing
needs of the poor and vulnerable in society.

The focus of this report is on Sendai Framework target (c) which seeks to “reduce direct disaster
economic loss in relation to global gross domestic product (GDP) by 2030.” This is also the theme of
this year’s International Day for Disaster Reduction, on October 13.

In particular, the report highlights the ever widening “protection gap” that exists between rich and
poor across the planet. It is often said that those who suffer the most from climate change are those
who contribute least to it. We are acutely reminded that disasters are a combination of hazard,
exposure and vulnerability.

It is also clear that the economic losses suffered by low and lower-middle income countries have
crippling consequences for their future development and undermine our efforts to achieve the 17
Sustainable Development Goals, in particular the eradication of poverty.

There is a deeper understanding today than ever before of the underlying factors which drive up the
likelihood of a future disaster event. More and more countries are moving to put in place national
and local strategies for disaster risk reduction by 2020, in line with target (e) of the Sendai Framework.

It is our hope that this report will encourage those efforts with further evidence that reducing disaster
risk and building resilience is essential to sustainable development.

Mami Mizutori Debarati Guha-Sapir

Special Representative of the Secretary-General Professor, Centre for Research on the Epidemiology
for Disaster Risk Reduction, Head of the UN Office of Disasters, Institute of Health and Society,
for Disaster Risk Reduction Université catholique de Louvain, Belgium.

Economic Losses, Poverty and Disasters 1998-2017 | 01

Who we are

CRED
The Centre for Research on the Epidemiology of Disasters (CRED) is the world’s foremost agency for the study of public
health during mass emergencies, including the epidemiology of diseases, plus the structural and socio-economic
impacts of natural and technological disasters and human conflicts. Based since 1973 at the School of Public Health of the
Université Catholique de Louvain, Belgium, CRED became in 1980 a World Health Organization (WHO) collaboration centre.
Since then, CRED has worked closely with United Nations agencies, inter-governmental and governmental institutions,
non-governmental organizations (NGOs), research institutes and other universities. Disasters preparedness, mitigation
and prevention for vulnerable populations have also gained a higher profile within CRED’s activities in recent years.

www.cred.be

EM-DAT
CRED’s Emergency Events Database (EM-DAT) contains the world’s most comprehensive data on the occurrence and effects
of more than 23,000 technological and natural disasters from 1900 to the present day. Created with the support of the WHO
and the Belgian government, the main objective of EM-DAT is to inform humanitarian action at the national and international
levels in order to improve decision-making in disaster preparedness, provide objective data for assessing communities’
vulnerability to disasters and to help policy-makers set priorities. In 1999, a collaboration between the United States Agency
for International Development’s Office Foreign Disaster Assistance (USAID/OFDA) and CRED was initiated. Since 2014, EM-DAT
also georeferences natural disasters, adding geographical values to numeric data which is essential for deeper analysis.

Details of EM-DAT’s methodology and partner organizations can be found on our website www.emdat.be

UNISDR
The UN Office for Disaster Risk Reduction was established in 1999 and serves as the focal point in the United Nations System
for the coordination of disaster risk reduction. It supports the implementation of the Sendai Framework for Disaster Risk
Reduction 2015-2030 which maps out a broad people-centered approach towards achieving a substantial reduction in disaster
losses from man-made and natural hazards and a shift in emphasis from disaster management to disaster risk management.
UNISDR and partners produce the biennial Global Assessment Report on Disaster Risk Reduction which provides evidence for
the integration of disaster risk reduction into private investment decision-making and public policy in urban, environmental,
social and economic sectors. UNISDR also coordinates the Making Cities Resilient Campaign and Worldwide Initiative for Safe
Schools and engages with governments in developing national disaster loss databases.

www.unisdr.org

02 | Economic Losses, Poverty and Disasters 1998-2017

Executive
Summary
Between 1998 and 2017 climate-related and geophysical disasters killed 1.3 million
people and left a further 4.4 billion injured, homeless, displaced or in need of
emergency assistance. While the majority of fatalities were due to geophysical
events, mostly earthquakes and tsunamis, 91% of all disasters were caused by
floods, storms, droughts, heatwaves and other extreme weather events.

In 1998-2017 disaster-hit countries also reported direct CRED, meanwhile, is employing an analytical technique
economic losses valued at US$ 2,908 billion1, of which known as georeferencing to drill down into EM-DAT data
climate-related disasters caused US$ 2,245 billion or 77% to reveal the relative vulnerabilities of rich and poor,
of the total. This is up from 68% (US$ 895 billion) of losses and quantify how the human cost of disasters increases
(US$ 1,313 billion) reported between 1978 and 1997. Overall, relentlessly in cases where national income levels decline.
reported losses from extreme weather events rose by 151%
between these two 20-year periods. For disasters since 2000, georeferencing has found that in
low income countries, an average of 130 people died per
In absolute monetary terms, over the last 20-year, the USA million living in disaster-affected areas, compared to just
recorded the biggest losses (US$ 945 billion), reflecting high 18 in high income countries. That means people exposed to
asset values as well as frequent events (Figure 1). China, natural hazards in the poorest nations were more than seven
by comparison, suffered a significantly higher number of times more likely to die than equivalent populations in the
disasters than the USA (577 against 482), but lower total richest nations.
losses (US$ 492 billion).
A similar pattern of deep inequality is revealed by
Such losses are only part of the story, since the majority of georeferenced ratios of people affected (but not killed) by
disaster reports to EM-DAT (63%) contains no economic data. disasters. While the largest absolute numbers of people
The World Bank has calculated that the real cost to the global affected by disasters lived in upper-middle income countries,
economy is a staggering US$ 520 billion per annum, with by far the highest number per 100 inhabitants lived in low
disasters pushing 26 million people into poverty every year2. income countries. Again the contrast is sharpest between
low income countries (7.8%) and high income countries
Absolute losses also mask the relatively greater burden of (1.3%), meaning that people in the poorest countries were on
disasters on the poor. When economic costs are expressed as average six times more likely than people in rich nations to
an average percentage of Gross Domestic Product (GDP), this be injured, to lose their home, be displaced or evacuated, or
becomes clearer. Figure 1 shows that only one high income require emergency assistance.
territory ranked among the top 10 in terms of percentage of
GDP losses over the past 20 years (Puerto Rico). Apart from Such data demonstrate that while absolute economic losses
upper-middle income Cuba, the other worst-hit nations were might be concentrated in high income countries, the human
all lower income countries, led by Haiti. cost of disasters falls overwhelmingly on low and lower-
middle income countries: vulnerability to risk, and degrees of
Again, inequality is even greater than available losses suffering, are determined by levels of economic development,
data suggest because of systematic under-reporting by low rather than simple exposure to natural hazards per se.
income countries. While high income countries reported
losses from 53% of disasters between 1998 and 2017, low At a time when climate change is increasing the frequency and
income countries only reported them from 13% of disasters. severity of extreme weather events, disasters will continue
No losses data are therefore available for nearly 87% of to be major impediments to sustainable development so
disasters in low income countries. long as the economic incentives to develop in hazard-
prone locations such as flood plains, vulnerable coasts
A similar divergence in record-keeping is evident geogra- and earthquake zones continue to outweigh the perceived
phically. Oceania recorded losses for 51% of climate-related disaster risks.
disasters in 1998-2017; in Africa, the figure is just 14%. Thus
the economic statistics in this report are the tip of the iceberg Integrating disaster risk reduction into investment decisions
as far as low income countries are concerned. is the most cost-effective way to reduce these risks; investing
in disaster risk reduction is therefore a precondition for
UNISDR and its partners are currently working with governments developing sustainably in a changing climate.
to establish robust national disaster loss databases as part of
the Sendai Framework for Disaster Risk Reduction 2015-2030.
Better record-keeping and standardized loss indicators will
1 
help planners improve how they manage “the risk of small- All economic losses and GDP are adjusted at 2017 US$ value
2 
scale and large-scale, frequent and infrequent, sudden and The World Bank - Results Brief - Climate Insurance (2017) Available at :
slow-onset disasters caused by natural or man-made hazards.” https://www.worldbank.org/en/results/2017/12/01/climate-insurance

Economic Losses, Poverty and Disasters 1998-2017 | 03

 Figure 1

Top 10 countries/territories for cumulative losses compared to

top 10 countries/territories for losses relative to GDP 1998-20173

Top 10 countries/territories in terms of absolute losses

(billion US$) 1998-2017

n High income
n Upper-middle income
n Lower-middle income
n Low income

Germany
57.9
France Italy Japan
USA 43.3 56.6 376.3
China
944.8
492.2

Puerto
Rico
71.7

India
79.5
Mexico
46.5
Thailand
52.4

Disaster type responsible for the majority of losses

Extreme
Earthquake
temperature

Earthquake &
Tsunami Flood

Storm Drought
3 
Excluding small states. See Annex for the list of small states.
See Annex for the methodology on the calculation of
economic losses related to GDP

04 | Economic Losses, Poverty and Disasters 1998-2017

Top 10 countries/territories in terms of average
annual percentage losses relative to GDP

n High income
n Upper-middle income
n Lower-middle income
n Low income

Mongolia
2.8%
Georgia
Honduras Haiti 3.5%
7.0% 17.5%

Puerto
Rico
12.2%

Korea
D.P.R
7.4%

El Salvador
4.2% Tajikistan
2.7%

Nicaragua
Cuba
3.6% Disaster type responsible for the majority of losses
4.6%

Extreme
Earthquake
temperature

Earthquake &
Tsunami Flood

Storm Drought

Economic Losses, Poverty and Disasters 1998-2017 | 05

Chapter 1

Disaster
trends
1998-2017
Information about the occurrence and severity of disasters has
greatly improved over recent decades, with an upswing in data
reported to CRED in the past five years, encouraged perhaps by
increasing international cooperation on disaster risk reduction,
a growing number of national disaster loss databases, and efforts
to accelerate implementation of the Sendai Framework.

When interpreting disaster data, one must be aware of the complex

interactions between natural hazards and human vulnerabilities:
the most violent storm over an uninhabited region will not be a
disaster if no people are harmed, while even a small tsunami
hitting a populous city with no early warning system can quickly
become a major disaster. Continued variability in reporting and
recording must also be taken into account, particularly when
looking for trends and patterns within the numbers.

06 | Economic Losses, Poverty and Disasters 1998-2017

In terms of occurrences, climate-related disasters dominate the
picture over the past 20 years, accounting for 91% of all 7,255
recorded events between 1998 and 2017 (Figure 2). Within this
total, floods were the most frequent type of disaster, 43% of all
recorded events (Figure 3).

Figure 2

Number of disasters by major category per

year 1998-2017

450

400

350

300

250

200

150
n Climate-related
n Geophysical
100

50

0
1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

Figure 3

Numbers of disasters per type 1998-2017

n Flood
n Storm
n Earthquake
n Extreme temperature

3,148
n Landslide
n Drought
n Wildfire
n Volcanic activity

2,049 n Mass movement (dry)

563
405 378 347 254 12
99
43.4% 28.2% 7.8% 5.6% 5.2% 4.8% 3.5% 1.4% 0.2%

Economic Losses, Poverty and Disasters 1998-2017 | 07

BOX 1
Sendai Framework Floods also affected the largest number of people, at more
than two billion, followed by drought, which affected a
for Disaster Risk further 1.5 billion people in 1998-2017 (Figure 4).

Reduction 2015-2030 Having become increasingly frequent in the late 1990s, the
sustained high level of climate-related events pushed the
average number of disasters per year to 329 in the latest
International efforts to reduce or mitigate the impacts 20-year period. This is double the average of 165 events per
of disasters have in the past 20 years become increa- annum in 1978-1997, although better reporting of disaster data
singly focused on human vulnerabilities. In 2000, the in the latest decades partly accounts for the apparent increase.
Millennium Declaration recognized the specific risk to
development arising from disasters and called on the
global community to “intensify our collective efforts to
reduce the number and effects of natural hazards and
man-made disasters” (United Nations 2000).
Figure 4
The Hyogo Framework for Action (UNISDR 2005) deve- Number of people affected
loped systems of indicators of disaster risk and vulne-
rability at national and subnational scales to enable
per disaster type 1998-20173
decision-makers to assess the impact of disasters on
social, economic, and environmental conditions and
then to warn others – both officials and the people at
risk – of the dangers.

Advances were made in terms of strengthening

disaster preparedness, response and early warning
systems, but progress was limited in most countries in
terms of managing underlying risks such as poverty,
environmental degradation, rapid urbanization and
population growth in hazard-prone areas.

The Sendai Framework was adopted at the Third UN

World Conference on Disaster Risk Reduction in Japan
in March 2015. It has seven strategic targets and 38
45%
indicators for measuring progress on reducing disaster 2.0 billion
losses. These indicators align implementation of the 3%
Sendai Framework with the UN’s global Sustainable 125
Development Goals (SDGs) and the Paris Agreement million
on climate change.

16%
726 million

33%
1.5 billion
2%
ACTION POINT
97
million
In 2016, the UN Secretary-General launched
“The Sendai Seven Campaign” to promote
each of the seven targets over seven years.
These targets start with 0.1%
(a) “aiming to lower average per 100,000 4.8
global mortality between 2020 and 2030 million
0.1%
compared to 2005-2015”, and 6.2
(b) “substantially reduce the number n Flood million
of affected people globally by 2030”.
In 2018, the focus is on target n Drought
(c) “Reduce direct disaster economic loss in n Storm
relation to global gross domestic product n Earthquake
(GDP) by 2030.”
n Extreme temperature
n Landslide
n Wildfire, Volcanic activity, Mass movement (dry)

08 | Economic Losses, Poverty and Disasters 1998-2017

 BOX 2
In 1998-2017, storms, including tropical cyclones and
hurricanes, were second only to earthquakes in terms of
Classification of
fatalities, killing 233,000 people over the 20-year period Natural Hazards4
(Figure 5).

In general, EM-DAT classifies disasters according to the

type of hazard that triggers them. This report focuses
Figure 5 on hydrological, meteorological and climatological
events – which collectively are termed weather- or
Number of deaths climate-related – plus geophysical disasters. CRED
per disaster type 1998-2017 defines a disaster as “a situation or event which
overwhelms local capacity, necessitating a request at
national or international level for external assistance;
an unforeseen and often sudden event that causes
great damage, destruction and human suffering”.
Disasters occur when natural hazards – storms,
earthquakes etc. – impact on vulnerable people.
Vulnerabilities arise (and increase) for many reasons,
including population growth, urban development in
risk-prone locations, land use changes, environmental
degradation, weak governance, poverty and inequality,
and climate change.

56%
747,234 deaths
Hydrological Meteorological Climatological

Flood Storm Drought

Extreme
11% Landslide
temperature
Glacial lake
outburst
17% 142,088
deaths Wave
action Fog Wildfire
232,680
deaths
13%
166,346
deaths
2%
21,563
deaths
1%
18,414
deaths

Geophysical

Earthquake
0.2%
2,398 Mass
deaths Movement
(dry)

Volcanic
activity
n Earthquake
n Storm
n Extreme temperature
n Flood
n Drought
n Landslide 4
See Annex for the reference to the full classification,
n Wildfire, Volcanic activity, Mass movement (dry) definitions and sources of EM-DAT

Economic Losses, Poverty and Disasters 1998-2017 | 09

Storms were also by far the costliest type of disaster, with By contrast, a similar scale earthquake to Haiti’s in New
reported storm losses amounting to US$ 1,300 billion over Zealand in 2010 affected 300,000 people but killed no
the past 20 years, twice the reported losses for either flooding one due in large measure to New Zealand’s strict building
or earthquakes (Figure 6). codes and high levels of preparedness. In Iran, investment
in earthquake risk reduction, particularly in safe school
Globally, geophysical disasters - primarily earthquakes - buildings, has reduced the country’s earthquake mortality
killed more people than any other type of natural hazard in substantially despite it being in an active seismic region.
the past 20 years, with the cumulative toll of 747,000 fatalities
exacerbated by the vulnerability of poor and badly prepared
populations exposed to two major events: the 2004 Indian
Ocean tsunami and the Haiti earthquake of 2010 (Figure 7).

Figure 6

Breakdown of recorded economic losses (US$)

per disaster type 1998-2017

23%
US$ 661
billion
23%
US$ 656
billion

46% 4%
US$ 124 billion$
US$ 1,330
billion

2%
US$ 68 billion

2%
US$ 61 billion
n Storm
n Earthquake
US$ 8 billion
n Flood
n Drought
n Wildfire
n Extreme temperature
n Others : Landslide, Volcanic activity, Mass movement

10 | Economic Losses, Poverty and Disasters 1998-2017

 In terms of risk reduction, earthquakes pose a particular set 5 
 cott, A., Misiani, H., Okoth, J., Jordan, A., Gohlke, J., Ouma, G., Arrighi, J.,
S
of problems for countries in seismically active areas. ‘Return Zaitchik, B., Jjemba, E., Verjee, S. and Waugh, D. (2017). Temperature and
times’ are long and unpredictable, so low mortality in the heat in informal settlements in Nairobi. PLOS ONE, 12(11), p.e0187300.
recent past is not an indicator of future risk. Available at : https://journals.plos.org/plosone/article?id=10.1371/journal.
pone.0187300
Fatalities from climate-related disasters also reflect vul-
6
nerabilities rather than being a crude function of the rising Howell, J. and Elliott, J. (2018). Damages Done: The Longitudinal Impacts
number of occurrences. Unlike earthquakes, populations vul- of Natural Hazards on Wealth Inequality in the United States. Social
Problems. Available at : https://academic.oup.com/socpro/advance-
nerable to extreme weather events are more evenly spread
article/doi/10.1093/socpro/spy016/5074453
around the globe. The spike year of 2003, for example,
includes 72,000 killed in heatwaves in Europe that year. The
2008 peak was caused by the 138,000 deaths from Cyclone
Nargis in Myanmar (Figure 7). In 2010, some 56,000 Russians
died due to a heatwave and 20,000 Somalis due to drought.

Slum dwellers are more affected by heatwaves. A recent John

Hopkins University5 study found that average temperatures in
Nairobi’s informal settlement were higher than in other parts ACTION POINT
of the city, because of lack of trees and vegetation to mitigate
extreme temperatures. The same trend is also found in devel- In all at-risk countries, preparedness
oped countries where the presence of green urban spaces is responses need to include public awareness
strongly correlated with the wealth of a neighborhood. campaigns, strategic risk assessment, and
enforced building codes focused on schools,
Another study in the USA found that wealth inequality be- health facilities, housing and work places in
tween races increased in the aftermath of a disaster6. order to reduce vulnerability.

Figure 7

Annual disaster deaths

Climate-related
by major disaster category 1998-2017 Geophysical

250,000

200,000

150,000

100,000

50,000

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

In terms of people affected (but not killed) by disasters, While droughts represent only 4% of total economic losses
Figure 8 highlights an abiding facet of such statistics: that (Figure 6), research by CRED in 20117 into 2,481 disasters
catastrophic events in the most densely populated countries where information is available found that, when considering
in Asia can alter the apparent profile of disasters over an the impact of disasters according to their type, droughts often
extended period of time. The high starting point for upper- inflict significantly greater losses on national economies
middle income countries in 1998 reflects flooding in China, than other types of disasters, with almost 40% of droughts
which affected 239 million people that year, with the 2002 studied provoking damage equal to or greater than 0.5%
peak increased by a sandstorm in China, which affected 100 of the GDP of the country where they occurred. This level
million Chinese. of 0.5% of GDP losses is the International Monetary Fund’s
threshold for a major economic disaster.
For lower-middle income countries, the 2002 peak represents
a severe drought in India, which affected 300 million people. 7
Guha-Sapir, D., D’Aoust, O., Vos, F. and Hoyois, P. (2013) The frequency
The 2015 spike is also the result of drought, this time not just and impact of natural disasters, in: The Economic Impact of Natural
in India but also the D.P.R Korea, Ethiopia and Malawi. Taken Disasters (Edited by D. Guha-Sapir and I. Santos); Oxford University Press:
together, drought affected 365 million people in these four Oxford: pp.1-27
countries in that one year.

Economic Losses, Poverty and Disasters 1998-2017 | 11

 Droughts are typically slow-onset events with a long duration In a review undertaken in 53 developing countries over the
and high spatial coverage, which have a high impact on period 2006–2016, the FAO found that, taken together, crops
human welfare and food security in countries dependent and livestock, farming fisheries, aquaculture and forestry
on agriculture. According to the UN’s Food and Agriculture absorbed 26% of all damage and loss caused by floods,
Organization (FAO)8, agriculture is the first sector affected drought and tropical storms. Almost two-thirds of all damage
when drought hits and also the most affected sector, and loss to crops was caused by floods in that decade, but
absorbing around 80% of all direct impacts, with multiple drought was by far the most harmful disaster for livestock.
effects on production, food security and rural livelihoods.
Droughts can lead to famine and migration, the loss of natural
resources, and seriously undermine economic performance
causing widespread hardship for rural communities and
price rises in basic commodities for the urban poor.

The FAO has predicted the rising incidence of weather extremes

will have increasingly negative impacts on agriculture,
including long-lasting and multi-pronged consequences such ACTION POINT
as loss of harvests and livestock, outbreaks of disease, and the
To monitor progress on disaster risk reduction
destruction of rural infrastructure and irrigation systems.
effectively, both the geographic and GDP
contexts of impacts must be taken into
account, as well as the variable impacts of the
8
FAO (2018) 2017 - The impact of disasters and crises on agriculture and different types of disasters that pose specific
food security. Available at : http://www.fao.org/3/I8656EN/i8656en.pdf threats to a region or nation.

Figure 8
n High income
Annual affected populations n Upper-middle income
by national income bracket9 1998-2017
n Lower-middle income
n Low income

400

350

300

250
millions

200

150

100

50

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

The erratic impact of major disasters is also apparent from The other two peak years in terms of reported cost (2005 and
data on their reported economic costs over the period 1998- 2017) are explained by the multiple storms that battered the
2017. In Figure 9, the peak year of 2011 reflects the immense Americas. The 2017 hurricane season was remarkable in terms
damage caused by the Great East Japan Earthquake and of the number of Category 5 tropical cyclones that swept in
Tsunami, with the consequent shut down of the Fukushima quick succession across numerous, vulnerable Caribbean
nuclear energy plant, with losses totalling US$ 228 billion. In island states and made landfall in Central and North America
2008, the earthquake in Sichuan, China, cost US$ 96 billion as well.
and affected 46 million people.
9
See Annex for the list of countries/territories per income group

12 | Economic Losses, Poverty and Disasters 1998-2017

 Together, storms cost more than 60% of all reported climate-
related disaster losses in seven out of the past 20 years,
and accounted for 59% on average of total reported climate-
related losses over that period (Figure 10).

Figure 9

Total reported economic losses per year,

with major events highlighted 1998-2017

Economic Hurricanes
Damages Harvey, Irma & Maria
(billion US$) (US$ 245 billion)

400
Earthquake/Tsunami
Tohoku, Japan
Hurricanes Earthquake
350 (US$ 228 billion)
Katrina, Rita & Wilma Sichuan, China
USA, Central Am. & Caribbean (US$ 96 billion)
300 (US$ 201 billion)

250

200

150

100

50

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

n Total Damages
n Selected disasters with large economic impact

Figure 10

Share of losses due to storms as a percentage

of annual climate-related disaster losses n Storm
1998-2017 n Other climate-related disaster

100%

90%

80%

70%

60%

50%
86 89 88
40%
74
61 64 62
30% 58 58
50 47 44 43
20% 37 38 39
35 33
28 29
10%

0%

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Economic Losses, Poverty and Disasters 1998-2017 | 13

Chapter 2

Economic
patterns
in disasters
data
The direct economic costs of disasters have been systematically
under-reported worldwide for decades, both in wealthier
countries and, most especially, in poorer ones. Throughout the
period 1998-2017, economic losses data only exist for 37% of
disasters; the direct cost of the majority of disasters (63%) is
unknown or not well documented.

14 | Economic Losses, Poverty and Disasters 1998-2017

Table 1 shows how uneven reporting patterns are within global since the adoption of the Sendai Framework in 2015 of the
income groups, with high income countries reporting damages need for better data collection and UN Member States are
for 61% of geophysical disasters, but records are available for signing up to use the Sendai Framework Monitor, launched in
just 13% of climate-related disasters in low income countries. March 2018, to report on disaster losses, including economic
losses. There is, however, still a long way to go, particularly in
Under-reporting is most acute in Africa, where economic cost lower-middle and low income countries.
calculations are available for less than 14% of all disasters.
The problem is not confined to one continent, however. In absolute terms, aggregated losses in lower income countries
Europe, Asia and the Americas reported losses for less than will inevitably remain below those in higher income countries,
half of all of their climate-related disasters. Even in Oceania, due to lower asset values. The costs in human and financial
which records losses for 51% of climate-related events, terms can, however, be enormous at the household and
the cost of nearly half of all storms, droughts, floods etc. community levels, especially when damaged or destroyed
remains unknown (Table 2). With the adoption of the Sendai property is uninsured. Lost crops and damaged agricultural
Framework, UNISDR is actively supporting 100 countries, land also impact hardest on the poorest, with chronic long-
including 30 in Africa, in the development and maintenance term consequences. Malnutrition and stunted growth are both
of national disaster loss databases. high in areas of repeated flooding in India, for instance10.

Economic losses reporting patterns have also been skewed According to the FAO, climate shocks were one of the leading
in the past 20 years in favour of certain types of disaster, with causes of acute food crisis and malnutrition in 2017, affecting
storm damage recorded for 55% of occurrences, compared 59 million people in 24 countries in Africa alone11.
to just 11% for extreme temperature events (Table 3). The
least well reported disaster type is dry mass movements or
landslides (Table 4). This under-reporting of economic losses
10 
means that the US$ 2,908 billion of losses recorded within Rodriguez-Llanes, J.M., Ranjan-Dash, S., Degomme, O., Mukhopadhyay, A.,
EM-DAT over the period 1998-2017 is a fraction of the real Guha-Sapir, D. (2011). Child malnutrition and recurrent flooding in rural
eastern India: a community-based survey. BMJ Open 2001;1: e000109.
total, especially for low income countries.
Available at : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3208901/
The good news is that there has been an upswing in the 11 
FAO, IFAD, UNICEF, WFP and WHO. (2018). The State of Food Security and
percentage of reports containing economic losses data for all Nutrition in the World 2018. Building climate resilience for food security
income groups, especially in the last five years, reversing an and nutrition. Rome, FAO. Available at : http://www.fao.org/3/I9553EN/
earlier declining trend. There has been a growing awareness i9553en.pdf

Table 1

Reporting of economic losses

per income group (%)

ALL Climate-related Geophysical

High income 53 52 61

Upper-middle income 40 40 37

Lower-middle income 31 30 31

Low income 13 13 20

Table 2

Reporting of economic losses

per continent (%)
ALL Climate-related Geophysical

Oceania 48 51 23

Americas 42 43 32

Asia 42 42 40

Europe 38 37 54

Africa 14 14 24

Economic Losses, Poverty and Disasters 1998-2017 | 15

 Table 3 Table 4

Reporting of economic losses Reporting of economic losses

per disaster type (climate-related) per disaster type (geophysical)

% reported % reported

Storm 55 Earthquake 43

Wildfire 41 Volcanic activity 11

Flood 32 Mass movement (dry) 8

Drought 29

Landslide 13

Extreme temperature 11

Despite these caveats about the partiality of losses data, it

is still possible to use the available records to demonstrate
how the economic burden of disasters weighs more on lower
income countries than higher income nations, with the ACTION POINT
percentage losses increasing step by step as a proportion of
GDP as national income levels decline. Gaps in losses data for many disasters, as
well as variations in their availability across
While high income countries reported US$ 1,432 billion in regions and disaster types, make analysis
climate-related disaster losses, or 65% of the global total, challenging and may result in pronounced
that only represented 0.41% of their GDP. The US$ 21 billion underestimations of damage for some regions
in climate-related disaster losses recorded by low income of the world. When implementing risk
countries amounted to an average of 1.8% of the GDP, well reduction efforts, decision-makers need to be
above the IMF’s threshold for a major economic disaster of aware of cases where bias is likely to occur.
0.5% (Figure 11).

Figure 11

Recorded climate-related disaster losses per income group

compared to GDP losses 1998-201712

n Absolute value (billion US$) n %GDP

1,600 2.0
1,432 1.77
1,400 1.8

1.4
1,200

1.2
1,000 1.14
1.0
800
0.8
567
600
0.60 0.6
400 0.41
0.4
194
200 0.2
21
0 0.0
High Upper-middle Lower-middle Low
12 
income income income income See Annex for methodology

16 | Economic Losses, Poverty and Disasters 1998-2017

Available economic records can also be added to other – High income countries experienced 27% of global disasters,
more comprehensive – EM-DAT data (occurrences, deaths and 9% of deaths and 3% of the total affected population, yet
affected totals) to indicate the relative scale of human and recorded 65% of all economic losses. While this percentage
economic burdens on countries grouped by national income. is inflated by under-reporting from other income groups, it
nevertheless illustrates how concentrated reported losses
Figure 12 shows that low and lower-middle income countries have been in the past 20 years.
carried a disproportionate burden in terms of disaster deaths,
experiencing 43% of all major recorded disasters in the past Historically, the large, populous continent of Asia has borne
20 years but the greatest proportion (68%) of fatalities. the brunt of global disasters of all types; this remained
true in the past 20 years. For geophysical disasters, Asia
Upper-middle income countries experienced lower mortality accounts for the majority of all recorded impacts, including
proportionately (22% of global deaths against 30% of an extraordinary 85% of all affected people, and 78% of
disaster events) but accounted for the largest proportion reported economic damage, as well as 62% of all occurrences
of people affected, 54% of the total. This high percentage and 69% of deaths (Figure 13).
reflects populous China being in the upper-middle income
bracket, and the fact that China suffered the highest number
of disasters (577) of any country in 1998-2017.

Figure 12

Climate-related and Geophysical Disasters

1998-2017

100%
6% 8%
14%
90% 22% n High income
80%
26% n Upper-middle income
37%
70% 29% n Lower-middle income

60%
n Low income
46%
50%

40% 30%

54% 65%
30%

20% 22%
27%
10%
9% 3%
0%
Occurrence Deaths Affected Economic
losses

Figure 13

Relative human and economic costs of

geophysical disasters on continents 1998-2017

100%
5% 5%
5%
n Africa
90% 8%
n Americas
80%
n Asia
70% n Europe
69%
60% n Oceania
62% 85%
50% 78%

40%

30%

20%
19% 30%
10%
13% 10%
0%
6%

Occurrence Deaths Affected Economic

losses

Economic Losses, Poverty and Disasters 1998-2017 | 17

For climate-related disasters, the continental pattern is
more mixed (Figure 14). Affected populations once again
overwhelmingly lived in flood- and storm-prone Asia (86%).
But there is a wider global distribution of occurrences,
including extreme temperatures in Europe, drought in
Africa, and storms in Oceania and the Americas. Mortality is
also more geographically spread, compared to geophysical
disasters, including a relative high proportion of deaths
(25%) in Europe, reflecting the toll of recent heatwaves. The
high level of reported economic losses in the Americas (53%)
is very largely due to storm damage.

It is important to note that the small totals for Oceania in

Figures 13 & 14 disguise the catastrophic damage inflicted by
tropical cyclones on the economies of small island states of
Vanuatu, American Samoa and Tonga.

Figure 14

Relative human and economic costs of

climate-related disasters on continents 1998-2017

100%
4% 2%
n Africa
90% 15% 12%
25% n Americas
80%
n Asia
32% n Europe
70%

60%
39% 86% n Oceania

50%
53%
40%

30% 25%
53%
20%
15%
10% 6%
17%
7% 7%
0%
Occurrence Deaths Affected Economic
losses

18 | Economic Losses, Poverty and Disasters 1998-2017

Chapter 3

Human
cost of
disasters
1998-2017
In EM-DAT, the human cost of disasters is measured by two main
parameters: the number of people killed, missing or presumed
dead, and the number of people affected by the events, that is,
in need of immediate assistance for basic survival needs (food,
water, shelter, sanitation, medical assistance). The affected
totals also include people injured, made homeless, displaced or
evacuated during the emergency phase of a disaster.

Raw numbers like these can, however, mask the reality of the
relative burden of disasters on different populations, and make
valid comparisons difficult. CRED, therefore, also examines EM-
DAT data by standardized measurements (per million inhabitants
and percentages, for example) to demonstrate the proportion of
vulnerable populations killed or affected by disasters worldwide13.

This report adds a further layer of analysis: georeferencing, a

technique that increases the precision of data by identifying the
population potentially exposed (PPE) to actual disasters using
the FAO’s Global Administrative Units Layers (GAUL).

13 
See Annex for methodology

Economic Losses, Poverty and Disasters 1998-2017 | 19

 GAUL was devised by the FAO to have two levels of precision: income groups (high income, upper-middle, lower-middle and
a 1st level Administrative Unit and a 2nd more localized level. low) with the proportions of people killed and affected within
In the USA, for example, a 1st level Administrative Unit would the PPEs.
be the State, the 2nd level is the County. In Sierra Leone and
Cambodia, the 1st level is the province, and the 2nd is the district. For deaths, this is done per million. For affected, the data are
CRED used this database in order to identify the administrative expressed as percentages of the PPE15.
units directly affected by disaster events. The lower the level,
The georeferenced data used for these two figures include
the more accurate the footprint of the disaster14.
6,636 disasters of all types: that is 91.5% of all records for the
The PPE (Population Potentially Exposed) is the total 20-year period covered by this report. Figures 17 & 18 repeat
population living in the administrative units known to these comparisons for climate-related disasters alone. The
have been affected by a disaster. Thus, rather than taking bars coloured in dark brown (the georeferenced data) are the
a country’s entire population as a basis for comparative most accurate comparative data available on the human cost
analysis, georeferenced data focus on people known to live of disasters triggered by natural hazards.
within standardized disaster-affected areas.

In Figures 15 & 16, the graphs compare the absolute numbers 14 

See Annex for more information on the precision of EM-DAT georeferencing
of people killed and affected by disasters within four national 15 
See Annex

Figure 15

Disaster deaths in absolute numbers per million population

potentially exposed (PPE) 2000-2017

n Deaths n Deaths/million

600,000 569,042 130 140

500,000 120

92 100
400,000
75 80
285,136
300,000
236,078 60
200,000
120,339 40

100,000 18
20

0 0
High Upper-middle Lower-middle Low
income income income income

Figure 16

Disaster affected totals in absolute numbers and percentage

of population potentially exposed (PPE) 2000-2017

n Affected (million) n % Affected

2,500 9.0
7.8
8.0
1,945
2,000 7.0

1,527 5.5 6.0

1,500 4.9
5.0

4.0
1,000
3.0

500 2.0
1.3
270
129 1.0

0 0
High Upper-middle Lower-middle Low
income income income income

20 | Economic Losses, Poverty and Disasters 1998-2017

Georeferenced disasters
2000-2017
In absolute terms, lower-middle income countries suffered Figure 16 also shows the stepped increase in affected numbers
the highest number of deaths due to disasters over the past per PPE as income levels decline, from 1.3% (high income),
18 years, with almost 570,000 people killed, the majority 4.9% (upper-middle income), 5.5% (lower-middle income) to
in geophysical events. By contrast, analysis of deaths per 7.8% (low income).
million PPE illustrates a clear global pattern, with death rates
tending to increase as income levels decline (Figure 15). Broadly similar patterns are evident when analysing climate-
related disasters alone.
In low income countries, an average of 130 people died per
million PPE (i.e. those living in the immediate disaster zone) Figure 17 shows lower-middle income countries again suffering
compared with just 18 in high income countries. That means the highest rate of mortality in absolute terms (with 243,000
people exposed to disasters in the poorest nations were recorded deaths between 1998 and 2017) but also in this case
more than seven times more likely to die than people in the the greatest number of deaths per million PPE as well (62 per
richest nations. million).

A similar pattern of inequality is revealed by EM-DAT data For populations affected by climate-related disasters (Figure
on the proportion of people directly affected (but not killed) 18), in absolute numbers more people were affected by
by disasters (Figure 16). While the largest absolute numbers climate-related disasters in upper-middle income countries
lived in upper-middle income countries (almost two billion), than any other group (a cumulative total of 1.9 billion people)
by far the highest percentage of people directly affected per followed by lower-middle income countries (1.5 billion).
PPE lived in low income countries.
Once again, though, when adjusted to a percentage of the
Again the contrast is sharpest between low income countries, PPE, a relentless increase is evident in the average percentage
where 7.8% of the PPE were on average directly affected by of people affected by climate-related disasters as incomes
each disaster, against just 1.3% in high income countries. decline. In the richest nations, just one per cent of the PPE
Thus, people in the poorest countries were on average six were affected, a figure that rises to 4.7% for upper-middle
times more likely than people in rich nations to be injured, income countries, 5.4% for lower-middle income states and
or lose their home, be displaced or evacuated, or require 7.6% for low income nations.
immediate medical assistance, food or shelter, and suffer the
consequences of damage to critical infrastructure including
the loss of public utilities, damaged schools, health facilities
and places of employment.
ACTION POINT

In order to establish the size and location of

vulnerable populations, accurate reporting of
the extent and type of disaster is of primary
importance. In this regard, low income countries
are leading the way.

Figure 17

Climate-related disaster deaths in absolute numbers

per million population potentially exposed (PPE) 2000-2017

n Deaths n Deaths/million

300,000 70
62
60
242,660 60
250,000
50
50
200,000
40
150,000
30
98,649 101,006
100,000
16 20
51,458
50,000 10

0 0
High Upper-middle Lower-middle Low
income income income income

Economic Losses, Poverty and Disasters 1998-2017 | 21

 Figure 18

Climate-related disaster affected totals in absolute numbers and

percentage of population potentially exposed (PPE) 2000-2017

n Affected (million) n % Affected

2,000 1,874 9.0

1,800 7.6 8.0
1,600 1,497
7.0
1,400
5.4 6.0
1,200 4.7 5.0
1,000
4.0
800
3.0
600
400 2.0
1.0 260
200 124 1.0

0 0
High Upper-middle Lower-middle Low
income income income income

BOX 4
Focus on storms

In 1998-2017, hurricanes that swept through the Caribbean Outside the Americas, Cyclone Pam cost Vanuatu 61% of the
and made landfall on the Americas caused the greatest island’s GDP in 2015 and set back development for years. In
economic losses of all climate-related disasters both in 2004, American Samoa recorded losses equivalent to 29%
absolute terms, and as a percentage of GDP (Tables 5 & 6). of GDP due to Cyclone Heta, while Cycle Waka in 2001 cost
Tonga 28% of GDP.
Three out of the five most costly storms in the past 20 years
occurred last year. The huge damage inflicted by these three In Puerto Rico last year, Hurricane Maria caused damage
storms - Hurricane Harvey (US$ 95 billion), Hurricane Maria equivalent to 69% of the island territory’s GDP, helping to
(US$ 70 billion) and Hurricane Irma (US$ 81 billion) - was push it into the top ten ranking for absolute losses in the
dwarfed only by Hurricane Katrina in 2005, which cost the past 20 years and also for losses as a percentage of GDP
USA a total of US$ 156 billion. (Figure 1). In Honduras in 1998, losses due to Hurricane
Mitch amounted to an even greater 73% of GDP.
Overall, storms cost more than any other type of climate-
related disaster in high and low income countries, accounting Insurers MunichRe16 called the 2017 hurricane season a
for 75% and 61% of their reported losses respectively. In wake-up call and a taste of things to come if meteorologists
middle income countries, storms came second only to floods are right in their predictions about the sort of extreme
in terms of absolute losses. weather events that climate change could make more
frequent. Munich Re calculated that disaster losses totaled
The heaviest cost of storms relative to the size of their US$ 340 billion last year, the second-highest annual figure
economies fell on small island states in the Caribbean. All ever, with insurance payouts at a record US$ 138 billion,
of the top 10 worst affected countries/territories in terms despite the share of insured losses remaining well below the
of losses as a percentage of GDP were small Caribbean 50% mark.
countries/territories.

Hurricane Irma in September 2017 cost Sint Maarten the 16 

Jeworrek T. (2018) Natural disasters in 2017 were a sign of things to
equivalent of 797% of GDP; the French part of the same come – New coverage concepts are needed. Available at : https://www.
island 584% of GDP, and British Virgin Islands 309%. That munichre.com/topics-online/en/2018/topics-geo/2017-was-a-wake-
same month Dominica recorded losses equal to 259% of GDP up-call
due to Hurricane Maria.

22 | Economic Losses, Poverty and Disasters 1998-2017

 Table 5

Top 10 climate-related disasters

for absolute losses 1998-2017

Name and date Countries/territories Sum of Total Damages

affected (billion US$)

Hurricane Katrina – Sep.2005 USA 156.3

Hurricane Harvey – Aug. 2017 USA 95.0
Hurricane Irma – Sep.2017 USA & Caribbean Islands 80.8
Hurricane Maria – Sep.2017 Caribbean Islands& USA 69.7
Hurricane Sandy – Oct. 2012 USA & Caribbean Islands 53.5
Flood – July & Aug. 1998 China 44.9
Flood – Aug.2011 to Jan. 2012 Thailand 43.4
Hurricane Ike – Sep.2008 USA & Caribbean Islands 36.3
Hurricane Ivan – Sep.2004 USA, Caribbean Islands & Venezuela 29.9
Hurricane Wilma – Oct.2005 USA, Mexico, Belize, Honduras & 25.0
Caribbean Islands

Table 6

Top 10 climate-related disasters

for losses as a percentage of GDP 1998-2017

Countries/territories Economic Economic

Name and date affected losses losses
(billion US$) (%GDP)

Hurricane Irma – Sep.2017 Sint Maarten 2.50 797

Hurricane Irma – Sep.2017 Saint Martin 4.10 584
Hurricane Irma – Sep.2017 British Virgin Islands 3.00 309
Hurricane Maria – Sep.2017 Dominica 1.46 259
Hurricane Ivan – Sep.2004 Grenada 1.15 148
Hurricane Ivan – Sep.2004 Cayman Islands 4.43 129
Hurricane Georges – Sep.1998 Saint Kitts and Nevis 0.60 110
Hurricane Erika – Aug. 2015 Dominica 0.50 90
Hurricane Mitch – Oct. & Nov. 1998 Honduras 5.68 73
Hurricane Maria – Sep.2017 Puerto Rico 68.00 69

Economic Losses, Poverty and Disasters 1998-2017 | 23

Conclusion: reducing
disaster losses is key
to eradicating poverty
It is just over three years since UN Member States adopted possibility that a natural hazard will turn into a humanitarian
the Sendai Framework for Disaster Risk Reduction 2015- catastrophe. More people are affected by extreme weather
2030, the global plan to reduce disaster losses, which events, especially floods and drought, than by any other type
is fundamental to the success of the 2030 Agenda for of natural hazard.
Sustainable Development.
While early warning systems and timely evacuations have
Today, the latest and most detailed disasters data available, led to reduced loss of life, economic losses continue to
compiled for this report by CRED from EM-DAT, the most grow, impeding a number of nations’ graduation from least
comprehensive disasters database in the world, underline developed country (LDC) status to middle income status.
the gross inequality of burden sharing between the richest
and poorest nations on earth, with the lowest income Many of those countries that suffer the most from economic
countries shouldering the greatest human cost of disasters. losses are small island developing states. Vanuatu, which
was devastated by Cyclone Pam in 2015 as the Sendai
Reducing disaster risk is a cross-cutting issue for all the Framework was being adopted, will not graduate from LDC
Sustainable Development Goals (SDGs), especially SDG 1, status until 2020.
on the eradication of poverty in all its forms, everywhere.
Disasters are a major contributor to entrenched poverty in The latest figures from the Internal Displacement Monitoring
low- and middle-income countries attempting to recover from Centre show the extent of the problem. Last year there were 18
extreme weather events amplified by the effects of climate million new weather-related displacements. Floods accounted
change. for 8.6 million displacements, storms 7.5 million, and drought
1.5 million. The challenge of reducing displacement from
Since the Sendai Framework was adopted, some 60 extreme weather events is universal. Due largely but not solely
million people in over 100 countries have been displaced to a very active Atlantic hurricane season in 2017, the USA
by disasters, mainly floods, storms and droughts. These had more people displaced by weather-related hazards than
adverse events often take place in environments exposed to Bangladesh, India or Somalia last year.
natural and man-made hazards, poverty, lack of protective
ecosystems, and weak institutional capacity to prepare for, The scale of economic losses, and the continuing high
and respond to them. numbers of people who are affected and often internally
displaced by disasters, should be incentive enough to
Population growth, economic development, and rapid and accelerate the implementation of the Sendai Framework, and
often risk-blind urbanization place more people in harm’s to ensure that these efforts are fully inclusive of the needs of
way than ever before in earthquake zones, flood plains, those groups and communities that are most at risk.
coastlines, dry lands and other high-risk areas, increasing the

24 | Economic Losses, Poverty and Disasters 1998-2017

Annexes
List of Acronyms
CRED: Centre for Research on the Epidemiology of Disasters
DRR: Disaster Risk Reduction
EM-DAT: Emergency Events Database
FAO: Food and Agriculture Organization of the United Nations
GAUL: Global Administrative Unit Layers
GDP: Gross Domestic Product
IRDR : Integrated Research on Disaster Risk
OFDA: Office of US Foreign Disaster Assistance
PPE: Population Potentially Exposed
SDGs: Sustainable Development Goals
UN: United Nations
UNISDR: United Nations Office for Disaster Risk Reduction

Chapter | 1
This report is based on EM-DAT data as of the date of the 14th of August 2018.

EM-DAT classification of natural dissaters

EM-DAT’s classification of natural disasters aligns with the IRDR’s “Peril Classification and Hazard Glossary” which in
turn is based on pre-existing work by Munich RE and CRED.
The IRDR’s classification is available online: http://www.irdrinternational.org/2014/03/28/irdr-peril-classification-
and-hazard-glossary/

The natural disaster category is divided into 6 sub-groups, which in turn cover 17 disaster types and more than 30 sub-
types. The EM-DAT classification is available on this page: https://www.emdat.be/classification

Definitions
https://www.emdat.be/Glossary

Disaster  Injured
Situation or event, which overwhelms local capacity, People suffering from physical injuries, trauma or an
necessitating a request to national or international level illness requiring immediate medical assistance as a direct
for external assistance (definition considered in EM-DAT); result of a disaster.
An unforeseen and often sudden event that causes great  
damage, destruction and human suffering. Though often Affected
caused by nature (Natural Hazard-Induced Disaster), People requiring immediate assistance during a period
disasters can have human origins. of emergency, i.e. requiring basic survival needs such as
food, water, shelter, sanitation and immediate medical
Total deaths  assistance.
In EM-DAT, it is the sum of deaths and missing.
Homeless
Death  Number of people whose house is destroyed or heavily
Number of people who lost their life because the event damaged and therefore need shelter after an event.
happened.
Economic losses (US$)
Missing
The amount of damage to property, crops, and livestock.
The number of people whose whereabouts since the disaster
For each disaster, the registered figure corresponds to the
is unknown, and who are presumed dead (official figure
damage value at the moment of the event, i.e. the figures
when available).
are shown true to the year of the event. The economic
Total affected losses were adjusted at 2017 US$ value using Consumer
In EM-DAT, it is the sum of the injured, affected and left Price Index for United States (2010 = 100) from the World
homeless after a disaster. Bank (at the date of June 2018).

Economic Losses, Poverty and Disasters 1998-2017 | 25

 Main Sources used in EM-DAT (non exhaustive)

Source Type Source Type of disasters covered

Information

United Nations OCHA Natural disasters

IRIN Natural and technological disasters (Africa)
WFP Drought/Famine
WMO Natural disasters
WHO/OMS Epidemics
FAO Drought/Famine

National National
Governments Governments Natural and technological disasters

US Governments FEMA Natural disasters (America)

NOAA Natural disasters
OFDA Natural and technological disasters
USGS Earthquakes
Smithsonian Volcanoes
DFO Floods, slides and windstorms
CDC Epidemics

IFRC International Federation of
Red Cross and Red Crescent
Societies Natural and technological disasters

Inter-Governmental
Organizations World Bank Major natural disasters

Reinsurance Companies SwissRe Natural and technological disasters

MünichRe Natural disasters
AON Benfield Natural disasters

Press AFP Natural and technological disasters

List of countries/territories per income group (World Bank, 2018)

For this report, CRED has adopted the World Bank revised classification of the world’s economies
based on estimates of GNI per capita for 2017.
https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups

• High income: $12,056 or more

• Upper-middle income: $3,896 to $12,055
• Lower-middle income: $996 to $3,895
• Low income: $995 or less

Small States in this report are the countries/territories marked with a * in the following table.
Small States are defined by the World Bank as those with populations of 1.5 million or less.

26 | Economic Losses, Poverty and Disasters 1998-2017

High income Upper-middle Lower-middle Low income
income income

Andorra Albania Angola Afghanistan

*Anguilla Algeria Bangladesh Benin
*Antigua and Barbuda *American Samoa *Bhutan Burkina Faso
Argentina Armenia Bolivia Burundi
*Aruba Azerbaijan *Cabo Verde Central Afr. Rep.
Australia Belarus Cambodia Chad
Austria *Belize Cameroon *Comoros
*Azores Islands Bosnia - Herzegovina Congo Congo Dem. Rep.
*Bahamas *Botswana Côte d’Ivoire Eritrea
*Bahrain Brazil *Djibouti Ethiopia
*Barbados Bulgaria Egypt Gambia
Belgium China El Salvador Guinea
*Bermuda Colombia Georgia Guinea-Bissau
*Brunei Darussalam Costa Rica Ghana Haiti
Canada Cuba Honduras Korea Dem. Peop. Rep.
Canary Is. *Dominica India Liberia
*Cayman Islands Dominican Rep. Indonesia Madagascar
*Channel Islands *Equatorial Guinea Kenya Malawi
Chile Ecuador *Kiribati Mali
*Cook Islands *Fiji Kyrgyzstan Mozambique
Croatia *Gabon Lao Peop. Dem. Rep. Nepal
*Curaçao *Grenada Lesotho Niger
*Cyprus Guatemala Mauritania Rwanda
Czech Republic *Guyana *Micronesia Fed. States Senegal
Denmark Iran Islam Rep. Moldova Rep. Sierra Leone
*Estonia Iraq Mongolia Somalia
*Faroe Isl. Jamaica Morocco South Sudan
Finland Jordan Myanmar Syrian Arab Republic
France Kazakhstan Nicaragua Tajikistan
*French Guiana Lebanon Nigeria Tanzania, Un. Rep.of
*French Polynesia Libya Pakistan Togo
Germany The Former Yugoslav Palestine, State of Uganda
*Gibraltar Republic of Macedonia Papua New Guinea Yemen
Greece Malaysia Philippines Zimbabwe
*Greenland *Maldives *São Tomé and Principe
*Guadeloupe *Marshall Islands *Solomon Islands
*Guam *Mauritius Sri Lanka
Hong Kong Mexico Sudan
Hungary *Montenegro *Swaziland
*Iceland Namibia *Timor-Leste
Ireland *Nauru Tunisia
*Isle of Man Paraguay
Israel Peru
Italy Romania
Japan Russian Federation
Korea Rep. *Saint Lucia
Kuwait *St Vincent and the
Latvia Grenadines
*Liechtenstein *Samoa
Lithuania Serbia
*Luxembourg (Serbia Montenegro)
*Macao South Africa
*Malta *Suriname
*Monaco Thailand
*Martinique *Tonga
*Montserrat Turkey
Netherlands Turkmenistan
*New Caledonia *Tuvalu
New Zealand Venezuela
*Niue
*Northern Mariana Isl.
Norway

Economic Losses, Poverty and Disasters 1998-2017 | 27

 High income Upper-middle Lower-middle Low income
income income

Oman
*Palau
Panama
Poland
Portugal
Puerto Rico
Qatar
*Réunion
*Saint Barthélemy
*St Helena, Ascension
and Tristan da Cunha
*Saint Kitts and Nevis
*Saint Martin (French)
*San Marino
Saudi Arabia
*Seychelles
Singapore
*Sint Maarten (Dutch)
Slovakia
Slovenia
Spain
Sweden
Switzerland
Taiwan (China)
*Tokelau
*Trinidad and Tobago
*Turks and Caicos Isl.
United Arab Emirates
United Kingdom
United States
Uruguay
*Virgin Isl.(British)
*Virgin Isl. (U.S.)
*Wallis and Futuna

Chapter | 2
Calculation of economic losses related to GDP

To calculate this proportion, only GDP data for the year and country where a value for economic losses was available
were taken into account. Thus if for certain years the economic losses are nil or unknown, GDP for this country is not
taken into account.

The percentage calculated is equal to the sum of economic losses for a year ‘j’, multiplied by 100, and divided by the GDP
for the same year ‘j’ for each country. The final percentage for the country for the period 1998-2017 is the average of the
previous calculated percentages.

Economic losses as % of GDP for i = average

x = economic losses ( x > 0)

( xij * 100
GDP ij )
i = country
j = year (from 1998 to 2017)

Same methodology is used for the calculation by income group:

Economic losses as % of GDP for z = average

z = income group
( xij * 100
GDP ij )
28 | Economic Losses, Poverty and Disasters 1998-2017
Chapter | 3
Precision of EM-DAT georeferencing

Ideally, georeferencing is taken to the 2nd Administrative Unit. But this level of precision is not always available within the
reported data. The quality of EM-DAT geo-location is therefore dependent on the accuracy and detail of the reporting by the
different data sources used by EM-DAT, as well as the type of the disaster and the region where the event occurred.

Unlike most disaster data, the proportion of events reported at the more detailed 2nd level Administrative Unit is highest
in low income countries (64%) and lowest in high income countries (40%). Because precision is less in higher income
countries, both the area potentially affected by the disaster and the affected PPE are both overestimated for higher income
countries. Were the PPE for higher income countries available to the more accurate 2nd level, the PPE percentages affected
would increase. This statistical phenomenon must be taken into account when interpreting these data.

Percentage of disaster reports by national income

groups at GAUL 1st and 2nd unit levels 2000-2017

1st Admin 2nd Admin

Unit /Low precision Unit /High precision

High income 60% 40% 100%
Upper-middle income 56% 44% 100%
Lower-middle income 38% 62% 100%
Low income 37% 64% 100%

For more information on Georeferencing, see CredCrunch 47 , CredCrunch 43 & CredCrunch 36:

1% of the disasters (68 on a total of 6,636) had a ratio >100%, which means that the population affected was superior to the
PPE. This is due to an under-reporting of the areas exposed to the disaster. The reliability of the PPE and calculated ratios
is highly dependent on the quality of the georeferencing, and therefore a precise reporting of the areas affected by the
disasters is crucial. The lowest income countries have a better precision in the area affected by disasters. (See table).
Those 68 ratios were modified and put as equal to 100% to avoid a bias in the calculation of the average ratios.

Calculation of population affected and number of deaths related to the population

To calculate this proportion, only population data for the year and country where a value for number of affected/deaths
was available were taken into account. Thus if for certain years the number of people affected/deaths are nil or unknown,
population for this country is not taken into account.

The percentage calculated is equal to the sum of number of people affected/deaths for a year ‘j’ multiply by 100 for the
rate related to affected and by 1 million for the rate related to deaths, and divided by the population for the year ‘j’ for
each country. The final rate for the country for the period 1998-2017 will be the average of the previous calculated rates.

Number of people affected per 100 inhabitants for i = average

( xij * 100
Population ij )
Number of deaths per 1 million inhabitants for i = average

X = population affected ( x > 0)

( xij * 1,000,000
Population ij )
i = country
j = year (from 1998 to 2017)

Economic Losses, Poverty and Disasters 1998-2017 | 29

Acknowledgements

This report is the outcome of a fruitful collaboration between

the United Nations Office for Disaster Risk Reduction
(UNISDR) and the Centre for Research on the Epidemiology of
Disasters (CRED), part of the Institute of Health and Society
(Université catholique de Louvain).

The analysis in this report was undertaken by Pascaline

Wallemacq, who also coordinated and supervised the
realization of this study.

The text was rewritten and edited by Rowena House, to whom

we are grateful for the quality of her work.

Denis McClean, UNISDR, provided insightful inputs and helped

refine the text.

Thanks to Regina Below who has maintained the EM-DAT

database for 20 years and who contributed with her past
work and ideas.

Mardi SPRL did the layout and infographics, and we would

like to thank them for their valuable work.

Debarati Guha-Sapir

30 | Economic Losses, Poverty and Disasters 1998-2017

Contact
CRED UNISDR

•M
 ail: • Mail:
Pascaline Wallemacq: Denis McClean:
[email protected] [email protected]
Regina Below:
• Phone:
[email protected]
+41 22 917 8897
•P
 hone:
• Postal Address:
+32 2 764 3327
9-11 Rue de Varembé
CH 1202, Geneva
•P
 ostal Address:
SWITZERLAND
School of Public Health
Institute of Health and Society (IRSS)
www.unisdr.org
Université catholique de Louvain
Clos Chapelle-aux-Champs, 30,
Bte B1.30.15
1200 Brussels, BELGIUM

www.emdat.be

design: www.mardi.be

@CREDUCL

We gratefully acknowledge partial support for this analysis

from the United States Agency for International Development
(USAID) as well as the Université catholique de Louvain (UCL),
Institute of Health and Society (IRSS).

The contents of this report remain the responsibility of the

authors alone.