ArticlePDF Available
Natural Disasters Trends
Nicolas Boccard
October 2018
We analyze the sparse publicly available statistics on natural disasters. After recalling the
pitfalls this endeavor entails, we argue for using risk ratio rather over absolute figures. Be-
tween 1970 and 2017, the risk of dying in a natural disaster is found to be without trend at
world level but falling in the US and Bangladesh. The global financial risk, the ratio of losses
to wealth, is found to be trending up at world level but once we compute this ratio at coun-
try level and aggregate using population as weights, we obtain a socioeconomic risk which
is statistically trend-less. Over the long run, financial losses amount to 2‰ of the gross na-
tional income while socioeconomic ones are twice as large. Lastly, we show that natural
disasters hit developing economies three times harder than the developed ones, both for in-
dividual fatality risk and for property destruction. Singular assessments for Bangladesh and
the USA are given. Data sources are publicly available at
Keywords: Natural Disaster; Trend; Economic Cost; Economic Growth; Developing Coun-
JEL codes : Q54; D81; G22; O20
The risk of dying in a natural disaster is without trend
Financial losses per unit of wealth are trending upward
Socioeconomic losses per unit of wealth are without trend
Financial losses amount to 2‰ of the long run Gross National Income
Socioeconomic losses double financial ones
Individual and socioeconomic risk in developing countries triple the OECD levels
Economics Department, Univeristy of Girona, Spain. This work was supported by the Generalitat de Catalunya
through contract AGAUR (SGR 1360) & the XREPP network as well as by the Spanish Ministerio de Economía y
Competitividad through contract ECO2016-76255-P.
1 Introduction
In 2017, hurricanes Harvey, Irma and Maria sowed devastation in the Caribbean and the south-
ern United States, renewing the interest of western media for these evils which have been known
to predominantly strike poor people living in developing countries (cf. Sawada and Takasaki
(2017)). As the world population keeps multiplying and the global economy continues to grow,
natural disasters are bound to find more to destroy on their path. At the same time, our modern
societies are becoming acutely aware that impacts are not only shaped by climatic events but
also by our own design choices regarding zoning, infrastructures or river management, to name
a few. Compounding this complexity is the fact that weather patterns display a long term regu-
larity, so that stricken zones will, in all likelihood, be hit again in the future. It is for that reason
that post-disaster decisions can have major consequences on future vulnerability. At the outset,
the question whether natural disasters are deadlier, more damaging, more frequent or already
under climate change influence, form a nexus of issues that has been a matter of controversy in
the academic literature.
The Stern (2007) review on climate change concludes that disaster losses are increasing
faster than might be explained by changes in wealth, population, and inflation so that climate
change influence must be present. Pielke (2007), among others responses, views this claim as
exaggerated because it relies on mere assumptions. The authoritative SREX-IPCC (2012) report
on climate extremes and disasters offers a rejoinder upon concluding that rising population and
increased capital at risk are the key drivers behind the observed increase in natural disaster
losses. Empirical work regarding trends in disasters has continued unabated but even the fairly
recent works of Loayza et al. (2012) and Visser et al. (2014) draw from datasets ending respec-
tively in 2005 and 2010. The time is thus ripe for an update of some essential statistics regarding
natural disasters.
The next section takes on the thorny question of defining natural disasters so as to delin-
eate our field of investigation. As we discuss next in the third section dedicated to information
sources, the high quality data needed for a trustworthy analysis at the world level did not be-
gin before 1980, so that even a modest addition of data points improves the statistical power of
econometric tests, and thus, the confidence we may place in the conclusions that emerge from
them. Section 4 then updates the values achieved by the basic indicators which feature in the
media. We then propose in §5, some better indicators capturing the notion of risk to lives and
material properties; we study their long-term behavior. Beyond updating well-known indica-
tors, we propose a novel one regarding economic losses that accounts for the heterogeneous
purchasing power of countries.
2 Nature vs. Society
Basically, a natural disaster occurs when a natural hazard1generates a human crisis but as
Quarantelli (1985) cheekily recalls, practitioners have long claimed that “a disaster is easier to
recognize than to define“. At any rate, the early definition of UNDRO (1980), as well as that fea-
turing in the SREX-IPCC (2012) report, is staunchly centered on physics; it came under challenge
by sociologists and psychologists who insisted on taking a broader view to account for the long
run consequences for communities affected by disasters. The acceptation of this expanded un-
derstanding is most apparent when looking at the various management stages relative to natu-
ral disasters; ex-ante activities include (long-term) prevention, mitigation and (short-term) pre-
paredness. During and after the natural event onslaught, immediate responses feature relief
and rescue; afterwards, ex-post activities involve recovery and reconstruction.
The consensual definition of UNISDR (2017) thus speaks of a crisis generated by the threat of
or the actual impact of relatively sudden natural agents such as earthquakes, floods, hurricanes,
volcanic eruptions, tornadoes, and tsunamis. The crisis must have a significant adverse social
impact, disrupting community life so strongly that local resources are overwhelmed and thus fail
to handle the requirements of the situation i.e., cope and respond to the crisis.
O’Keefe et al. (1976) even went to suggest that we ought to “take naturalness out of natural
disasters“ because nature is only a trigger for the human-felt disaster. Plainly, a strong clima-
tological or geophysical event taking place over an area void of human settlement (e.g., island,
forest or mountain) will not make it into the disaster rosters for two reinforcing reasons: there
was no one in harms way and no one to report the event. Everyone will agree that a disaster is
the interface of an extreme physical phenomenon and a vulnerable human population. These
authors then emphasize the role of uneven development among and within countries as the
real causes of the increasing frequency of disasters. It is now broadly agreed, at least within aca-
demic circles, that the root causes of natural disasters are the many socioeconomic and political
processes that push people into vulnerable situations.
In this respect, famines, epidemics and droughts (FEDs) constitute a class of ancient phe-
nomena featuring, on the one hand, a strong social component2and on the other hand, a
very distinct natural development when compared to meteorological disasters; specifically, on-
slaught is slower, longer lasting and broader in terms of geographic area stricken. The long lead
time to an actual disaster makes intervention and mitigation feasible, thus not only does the
social component play a role in the development of the crisis, it can also, through anticipation,
play a role in its resolution and the ultimate adverting of a disaster. These characteristics lead
Quarantelli (2001) to question the “disaster“ labeling of FEDs from a theoretical, research, plan-
1An extreme event occurring naturally and causing harm to humans, though less than a disaster.
2Even a drought contains a social component since usage of the scarce available water is a social decision.
ning and operational viewpoint. Instead, he views FEDs as social problems, involving chronic
stress that, when combined with an acute weather event, may lead to a disastrous outcome.
Drought-induced famine, then, should be seen as a continuous developmental problem, in-
timately linked to poverty, corruption and institution building (as shown recently by Irogbe
The point may seem moot for famines which have essentially disappeared3but it remains
forceful for epidemics. For instance, the recent Ebola virus outbreaks are counted as disasters by
the multi-lateral institutions whereas the long-running AIDS epidemic is not, even though AIDS
appears to possess all the characteristics of a disaster, at least from the perspective of African
countries. The difference of legal and symbolic treatment may lie with the once severe impact
of AIDS in the block of advanced economies. Another example of odd treatment are illegal drugs
in the US. Meanwhile these killed only minorities, authorities simply treated the problem as a
police issue4but now that the disease has spread to the main population with drug overdose
deaths killing more than guns, car crashes, or AIDS5it has been labeled an epidemic (cf. Case
and Deaton (2017)). Shouldn’t we then count these fatalities as natural disasters victims ?
To avoid an endless discussion about such a sensitive issue and enhance the coherency of
our work, we shall follow on the footsteps of the commercial rosters by solely counting the farm
losses from drought, ignoring all other FEDs. When working with the EM-DAT roster, we thus
exclude famines and epidemics.
3 Information Sources
Newspapers and TV channels report about natural disasters using the communication pieces re-
leased by the proprietary rosters of reinsurance companies MunichRe,SwissRe and Aon-Benfield
which do not divulge their databases nor the exact methods and sources to populate them.
We draw information from the reports published by commercial rosters in the first trimester
of 2018 (covering 2017 and previous years). Academics and multilateral institutions favor the
open-access EM-DAT roster operated by the Centre for Research on the Epidemiology of Disas-
ters at UCLouvain, with support from USAIDs Office of Foreign Disaster Assistance (OFDA). We
3World Peace Foundation (2018) reveals that, except for Sudan in 1984, the sole famine cases in recent decades
are war-induced, thus reducing the natural component to a minimum. Even the “risk-of-famine“ declared by the
UN in 2017 regards countries enduring wars, unrest or civil strife (cf. O’Brien (2017)).
4A wave of heroin-addicted Vietnam soldiers popularized the heroin drug in the US within the (social) housing
projects, leading a few years later to a peak of 15 overdose deaths per million people. The crack epidemic which was
again limited to ethnic minorities took place during the 1980s; it was a worse endemic, with a death rate reaching
almost 20 overdose deaths per million.
5The death rate is now above 200 overdose deaths per million US population as CDC (2018) estimates 72 300
drug overdose deaths in 2017.
queried the EM-DAT database in August 2018(cf. online dataset).
Regarding the quality of the information at hand, the OECD (1994) manual offers a harsh
judgment when noting “partial coverage and the lack of internationally agreed definitions and
protocols for collecting disaster statistics means that none of the existing databases currently
provides a satisfactory basis for the global analysis of the occurrence and impact of the principal
disaster types.“ In response, an effort has been made at developing common rules and methods
among MunichRe and EM-DAT (cf. Below et al. (2009)). Still, the United Nations report UNISDR
(2015) dedicated at improving data collection notes that all natural disaster rosters suffer from
endogeneity problems, thus calling for a cautious interpretation of analytic results. Let us review
these problems.
Firstly, a natural disaster is assessed only when researchers become aware through the news
channels or the insurance industry, a process that biases heavily against older events from a
time when no one was scrutinizing recurring disasters such as floods or storms. Hoeppe (2016),
the research director at MunichRe, explains that their collection contains about 26 catastrophes
(aka major disasters) per year for the period 1900-1950. Over 1951-1980, most natural disasters
occurring in western countries are accounted for so that about 100 natural disasters are added
to the database every year. Finally, starting 1980, non major disasters from all over the world
start being assessed and the number of records rises to about 700 per year. This staggering
increase in the apparent frequency of disasters is thus unrelated to the physical laws governing
the underlying natural processes. In agreement with SREX-IPCC (2012) (cf. ch.4), the year 1980
is taken to be an adequate starting point for the study of the economic dimensions of natural
Secondly, global catastrophe rosters are biased against areas with low insurance coverage
(e.g., the poorer countryside within a nation) because insurance companies and their field agents
are the most reliable source of information relative to disasters. Deficient information transmis-
sion also create a bias against events occurring in non-english speaking countries and in devel-
oping nations whose local administration and news networks are still weakly developed. To
correct somehow for these deficiencies and guarantee a proper assessment of candidate events,
stringent selection criteria are applied: Guha-Sapir et al. (2017) report that EM-DAT requires 10
casualties or 100 affected people or a state of emergency declaration, SwissRe (2017) requires at
least either 20 casualties, 100 M$ of economics losses or 50 M$ of insured losses while Aon Ben-
field (2017) requires half of these figures. MunichRe (2018) used selection criteria in the past but
now purports to record all natural disasters, solely discarding those with losses inferior to 3 M$
6A group of MunichRe researchers, in Faust et al. (2006), further explain that high quality reports were about
10% throughout the 1980s and only started rising during the 1990s to reach 30% in 2005 (cf. figure 1). This selection
bias in favor of recent years and against older times cannot be estimated but it certainly makes a rising trend (in
casualties or damages) even more unlikely.
in rich countries. We thus observe that the set of criteria has either not been changed since their
inception or changed to incorporate more events. A mechanical positive selection bias is thus
in motion, picking-up more disasters every year as population and income keep rising around
the world.
Next is the issue of insurance. Commercial rosters and the US National Oceanic and Atmo-
spheric Administration (NOAA) have in the past acknowledged that uninsured losses were sim-
ply estimated to be identical to insured ones as if insurance coverage was a uniform 50% across
land and time. The damage figure of a natural disaster was thus twice the amount that the local
pool of insurers paid to its insured clients in the stricken area. The current methodology pa-
pers published by all rosters claim to carefully assess each event in order to estimate properly
uninsured losses although there is no way to check its veracity. At any rate, the aforementioned
practice which was used for decades has underestimated the total damages in developing coun-
tries where the rate of insurance coverage is much lower than in the US. This is why we separate
in a later section the group of (rich) OECD countries from the rest of the world.
A further point, quite difficult to quantify, let alone test, is exposed by Quarantelli (2001)
who notes, from anonymous interviews, that some authorities have an incentive to overstate
the human consequences of natural hazards in order to obtain more aid or better international
finance deals while other governments tend to understate the impact of natural disasters to
mitigate the local political reaction.
Looking back at the origin of natural disaster studies also sheds light on some of the afore-
mentioned biases. Hewitt and Sheehan (1969) appears to be the first effort at recording major
natural disasters occurring worldwide. Their selection criteria was either 1 M$ of losses (7 M$
of today) or 100 casualties or 100 injured people which means that solely major disasters were
recorded. The Anglo-Saxon bias was acknowledged by authors as their sources were the New
York Times, Encyclopedia Britannica, Collier’s Encyclopedia, American People’s Encyclopedia
and Keesing’s Contemporary Archives. The follow-up by Dworkin (1974) and Thompson (1982)
use the exact same sources and selection criteria with an adjustment for inflation. Glickman
et al. (1992) on the other hand expand the recordings towards man-made disasters and concen-
trate on loss of life with a minimum threshold of 25 casualties.7
Even though disaster rosters display serious problems, we nevertheless see them useful for
a trend analysis. Indeed, inso far as the collection method has not been modified, any selection
bias will carry on without altering the advanced indicators we plan to use. If for instance, solely
30% of true natural disasters are recorded in India because the 10 fatalities threshold is too low
for some local events to make it to the national scene, it is as if the threshold in India was some
larger figure, say 50. Hence, whenever a disaster kills more than 50 people in India, it will in all
likelihood be recorded. It is this inter-temporal stability that matters when seeking to identify a
7The statistical analysis of this database reveals a very good match with EM-DAT which most likely included it.
trend. Note though that this disaster count series is not comparable to an equivalent measure
constructed in Belgium because in this smaller country, 10 fatalities are always a major news
4 Basic Indicators
The media, and too often academia, almost exclusively talk about natural disaster on the basis
of the following basic indicators: the yearly count of natural disasters, the associated casualties8
and the monetary valuation of damages (aka. financial losses, updated for inflation). Table 1 dis-
plays for each roster the three basic indicators averaged over the last decade. Opinions converge
for casualty counts but differ regarding damages.9The reasons explaining such a divergence are
impossible to give here since the commercial rosters do not divulge their database, nor their
assessment methods. We may nevertheless observe that, being in the insurance business, they
gain from showing off the largest possible figures to impress potential clients such as pension
funds seeking diversification or the very governments of the countries hit by disasters. At any
rate, these divergences regarding the magnitude of natural disaster losses puts into doubt the
quality of this information and thus the trust we should put into it (as previously argued).
Roster Events Casualties Losses
EM-DAT 327 65,336 bn$ 159
SwissRe 177 64,193 bn$ 204
MunichRe 621 59,227 bn$ 190
AON 287 64,342 bn$ 241
yearly average, computed over 2008-2017
Table 1: Natural Disaster summaries
To increase statistical confidence for an inter-temporal analysis of this basic information, we
take the average over the four rosters. The temporal evolution of the three indicators is displayed
on Figure 1. The number of events recorded (black line) evidently grows with time. Fatalities are
more challenging to plot because this statistic changes by two orders of magnitude from one
year to another. We thus display a multiple of the logarithm (red line); it does not show any clear
trend. Lastly, the inflation-adjusted monetary value of damages (blue line) is growing.
8Rosters also report the number of people affected by a natural disaster but as this murky category lacks an
established definition, it is not statistically reliable and won’t be used in this article.
9AON and SwissRe use the US CPI to publish time series of damages expressed in constant US dollars (USD)
whereas EM-DAT and MunichRe publish nominal USD damages; we apply upon the latter two the US CPI to derive
real damage series, so as to make all four estimates comparable.
The gist of this simple analysis is well known and updated every year in the natural disasters
reports published by the commercial rosters. It is then expounded in the media, leaving readers
with the impression that the danger or risk posed by natural disasters are growing. Sadly, too
many academic reports are also content with this simple raw aggregated information; the basic
indicators are the almost exclusive source for the encyclopedia entry of Keiler (2013) as well as
the infographics found in the official and influential UNISDR (2018) report.
1970" 1975" 1980" 1985" 1990" 1995" 2000" 2005" 2010" 2015"
"Events"" Log(fatalities)" Loss"(2017bn$)"
Figure 1: Natural Disaster basic indicators
The econometric analysis10 of the time series shown in Table 2 reveals that the positive slope
parameter is statistically significative at the 1‰ level for events and losses. Using the IPCC
(2014) jargon detailed in Mastrandrea et al. (2011), we may say that it is “virtually certain“ that
both events and losses are growing with time. On the other hand, the negative slope parameter
for casualties cannot be distinguished from zero since its statistically significance is very low.
Slope Estimate Standard Error t-Statistic P-Value
Events 7.66 0.32 583 0.000
Casualties -289 907 0.10 0.751
Loss (2017bn$) 4.44 0.65 47.2 0.000
Table 2: Econometric test for the Basic Indicators
5 Risk Indicators
In 1988, the UN proclaimed an International Decade for Natural Disaster Reduction to draw me-
dia attention towards the increasing losses caused by natural hazards and to promote actions
to reduce their impacts. Within this endeavor, basic indicators constitute a first step towards
10For a time series (Xt)t0, the equation Xt=α+βyeartis estimated with ordinary least squares (OLS).
understanding the impact of natural disasters but being absolute figures, they bear no relation
to socioeconomic dimensions such as population growth, urbanization, construction zoning,
economic activity (e.g., infrastructure building) and the environment (e.g., river management).
Knowledgeable about this shortcoming, the UN has developed the notion of disaster risk de-
fined as “the likelihood of loss of life, injury or destruction and damage from a disaster in a
given period of time“.11 The UN Sendai Framework for Disaster Risk Reduction precisely seeks
to achieve seven global targets that include the risk indicators we compute in this article. Oddly
enough, when moving from theory to practice, the UN publications broach on disaster risk al-
most exclusively through the exposition of dramatic events. Chapter 2 of the voluminous report
UNISDR (2016) takes stock of the progress achieved in reducing disaster risk and, except for the
last figure (#2.8), insists on talking about basic indictors which is rather uninformative if not
outright distorting.12
We thus feel necessary to relate the absolute figures of disaster reports with the growth of
population and economic activity, employing the following indicators of natural disaster risk:
Frequency ratio of events per billion population ν
Individual risk number of fatalities per million population ζ
Financial risk ratio of losses to gross national income (GNI) ϕ
Frequency, though intuitive at first glance, may be the least useful indicator because in a
densely populated world, a major natural hazard will almost always strike a human community,
thus being counted as a natural disaster. Now, even if climate change is slowly increasing the
frequency of natural disasters, their yearly absolute number will be more or less stable over a
decade whereas population will keep growing, thus mechanically forcing the ratio νdown. The
notion of individual risk is self-explaining and already appears in Dworkin (1974), Thompson
(1982), Glickman et al. (1992) and all UN documents. Lastly, when considering the replacement
cost of the assets such as property or infrastructure destroyed by a disaster, one would ideally
compare this loss magnitude to the existing stock of wealth. Neumayer and Barthel (2011) ex-
plain and defend the generalized use of the flow variable GNI/GDP as a acceptable proxy.13 One
may further interpret ϕas the share of economic activity “wiped out“ by disasters during a year.
Suppose indeed that a disaster strikes a country in early January generating damages as large as
11cf. almost identical definition in of SREX-IPCC (2012).
12We read p47, “Since 1990, around 85% of internationally reported earthquake mortality has occurred in low
and middle-income countries“ while p49 claims “over 60% of internationally reported economic losses are con-
centrated in OECD and other high-income countries, reflecting the concentration of economic assets.“ In both
cases, the obvious question is whether the impacted population is at, below or above 85% and likewise for the
concentration of assets wrt. the 60% threshold. The report does not provide an answer.
13This usage is so common that SREX-IPCC (2012) §5.4.2 does not even bother to define the financial risk ratio.
1% of the country GNI. Since losses are computed as the replacement cost of damaged assets, it
is as if 1% of the workers spend their year repairing and rebuilding the damaged infrastructures
(or making new machines and appliances to replace the broken ones). Hence, only 99% of the
year’s economic activity has been truly a novel addition of wealth to the country.
The population figures used here are sourced from the UN while the world GNI is sourced
from the World Bank.14 Figure 2 displays the risk indicators as well as the trend line for each.
The unit for financial risk ϕis the basis point (with 100 bp =1%).
1970" 1975" 1980" 1985" 1990" 1995" 2000" 2005" 2010" 2015"
events/bn.pop." fatalities/M.pop." Loss/GNI"(φ)"
Figure 2: Natural Disaster risk indices
We observe that frequency (black line) and financial risk (blue line) are rising; the econo-
metric test shown in Table 3 (cf. next section) indeed confirms that, in each case, the slope
parameter is statistically significative at the 1‰ level. The mean yearly real financial losses to
natural disasters over the last decade (2008-17) were 198 bn$ while the mean yearly real GNI was
78 tn$, so that the financial risk was 2.5‰ i.e, for every hundred dollars of wealth generated by
economic activity, 25 cents must be aside to replace the losses from natural disasters. This figure
looks small at world scale but may be several times over the GNI for a small country or a region
(within a large country) stricken by a major disaster; this is why the mutualization of disaster
risk is of paramount importance both within and between countries (cf. Linnerooth-Bayer and
Hochrainer-Stigler (2015)).
Regarding human casualties, the individual risk (red line) appears to be trending down but
since the relationship is statistically significative solely at the 20% level (“likely“ per IPCC), stan-
dard statistical usage leads us to conclude to an absence of trend. There is also a possible trend
14 We use the constant 2010 US$ GNI at market prices (WB code NY.GNP.MKTP.KD) since losses, a replacement
cost for damaged assets, are expressed in real terms and market prices (rather than purchasing power parities). The
world GDP estimate from the International Monetary Fund is used to estimate by splicing the GNI for 2017.
reversal (also not significative) if we only consider the more recent 1980-2017 period where data
achieved a greater quality. It is thus safest to conclude to an absence of trend at the global level
for the individual risk. As shown later, a downward trend is identified for some specific coun-
6 Socioeconomic Risk
The financial risk ϕis a useful indicator for the global insurance industry as it tracks changes
in their underlying business line. From a social perspective however, ϕis inadequate because
damage figures underestimate the harm inflicted by natural disasters upon developing economies.15
Consider two contrasting examples. A pair of tropical storms destroyed a staggering 118 bn$ of
value on the US gulf coast during 2017; because this country is also the first economic power, it
may be said that solely 6‰ of its GNI was lost to natural disasters.16 At the other extreme, the
2010 Haiti earthquake destroyed a more limited 8 bn$ of value but this amounted to 120% of the
GNI because Haiti is extremely poor. To assess the socioeconomic impact of natural disasters
properly, one should ideally account for the difference in exposure, resilience and construction
cost. A simple proxy for this task is the financial risk, computed at the lowest geographical level,
since it matches the replacement cost of losses (computed at market prices) with the ability to
underwrite those losses, the local GNI (also computed at market prices). Due to the paucity of
regional macroeconomic data, we shall perform this analysis at the country level.
An alternative avenue is to normalize losses for socioeconomic change and/or differences
of development. This idea from Pielke and Landsea (1998) has been frequently used in the lit-
erature but is however criticized by Estrada et al. (2015) for being ad-hoc and unable to distin-
guish spurious trends from hypothesized ones, such as the climate change contribution. At any
rate, normalization requires high quality information which may not be available in developing
countries, thus impeding its development beyond the US.
Our method, though simpler, is more robust at it solely relies on the GNI, a trusted socioeco-
nomic piece of information available across the globe. We then associate each person with the
risk figure from his/her country; summing over the world population, we obtain the socioeco-
nomic risk φof natural disasters. Going back to our previous examples, the low US risk figure will
thus be counted 325 millions times while the high Haitian figure will appear 10 million times.
Formally, for country iand year t, we let pi,tbe the population, Wi,tthe (current $) GNI and Li,t
15For instance, in order to rebuild a 100$ property damage in India, the owner must renounce a certain bundle
of goods and services. For a US property owner to renounce the same bundle, he should have suffered a property
damage to the tune of 900$ (due to the difference of the purchasing power of 1$ in both countries).
16At the local level, hurricane Harvey wiped out a quarter of Houston’s GDP (some 100 bn$ out of 400 bn$) which
is considerable given the economic power of this metropolis.
the (current $) financial losses. We have
where σi,t=pi,t
Pipi,tis the country’s share of world population. The socioeconomic risk φis thus
the expected loss of property to natural disasters expressed as a proportion of purchasing power
(because for each country we divide losses by the local GNI). To perform the empirical calcula-
tion, we use EM-DAT which is the only roster allowing to go down to the country level. Figure 3
displays jointly the financial (ϕ) and socioeconomic (φ) risks, both expressed in basis points on
a logarithm scale (due to the large year-to-year variations).
1970" 1975" 1980" 1985" 1990" 1995" 2000" 2005" 2010" 2015"
financial"risk" socioeconomic"risk"
Figure 3: Natural Disaster Monetary Risk (in basis points)
We observe that in some years (e.g., 1976), φis a multiple of ϕbecause destruction was
wrought in very poor and populous countries (at the time) and was therefore highly damaging
to a large number of people. In recent years, we observe several rank inversions of φand ϕbe-
cause natural disasters were predominantly impacting rich countries which are relatively better
prepared to withstand disasters (so that a smaller share of the local wealth is lost).
The econometric analysis has already shown that the financial risk ϕis trending up. The
socioeconomic risk φhas an empirically positive slope but since its statistically significance is
no better than 42%, we conclude to an absence of trend. Over the 1980-2017 period where data
quality is highest, the financial risk ϕaverages17 1.8‰ of the world GNI whereas the socioeco-
nomic risk φaverages twice higher at 3.6‰ of the world GNI. This empirical finding means that
natural disasters hit harder the developing countries because hazards destroy a greater share of
the local wealth, most likely because they lack the defensive infrastructures put in place by rich
countries.18 We have thus extended to damages a result well known for individual risk (cf. next
17This figure is smaller than the previous one at 2.5‰ because commercial rosters inflate losses wrt. EM-DAT. In
the abstract, we report a rounded value of 2‰.
18Our interpretation is buttressed by the UNISDR (2016) claim that reserving 1‰ of future infrastructure invest-
ment for disaster-risk-reduction would generate a very large reduction of future losses to natural disasters.
section). Note that the discrepancy between ϕand φover the last decade has shrunk because
more damages have been inflicted onto advanced economies (per the randomness of mother
nature). Table 3 displays all the estimation results relative to the trend for the 4 risk indicators.
Note that ϕ1is computed with the commercial roster information while ϕ2and φ2solely use the
EM-DAT open source.
1970-2017 period Slope Estimate Standard Error t-Statistic P-Value
events per billion population (ν) 0.869 0.073 142.0 0.000
fatalities per million population (ζ) -0.256 0.195 1.7 0.195
ratio of losses to GNI (ϕ1) 0.436 0.101 18.8 0.000
socioeconomic risk (φ) 0.233 0.284 0.7 0.417
ratio of losses to GNI (EM-DAT ϕ2) 0.294 0.105 7.9 0.007
1980-2017 period Slope Estimate Standard Error t-Statistic P-Value
events per billion population (ν) 0.491 0.074 44.1 0.000
fatalities per million population (ζ) 0.123 0.155 0.6 0.433
ratio of losses to GNI (ϕ1) 0.400 0.156 6.5 0.015
socioeconomic risk (φ) -0.241 0.389 0.4 0.539
ratio of losses to GNI (EM-DAT ϕ2) 0.201 0.163 1.5 0.224
Table 3: Econometric Trend estimation
In a related work, Noy (2016) applies the Disability-Adjusted Life Year (DALY) technique of
the World Health Organization (WHO) to create a synthetic absolute disaster measure whereby
each casualty is allocated as many lost “lifeyears“ as the number of years the median citizen is
expected to live; furthermore, each affected person is allocated 2 month of lost “lifeyears“ and
a damage equal to the “country GDP per capita“ is counted as 9 months of lost “lifeyears“. At
world level, we find that the ratio of total lost “lifeyears“ to population is well correlated with
socioeconomic risk (ρ=0.42). This is likely a consequence of the historical strong correlation
between per-capita wealth and longevity (also known as the correlation between wealth and the
UN human development index).
7 Regions
It makes little sense from a geophysical point of view to distinguish countries as nature knows no
borders but once we broach the disaster problem from a socioeconomic angle, it starts making
sense since each country has a different level of development, a specific disaster culture and a
set of prevention policies shaped by an idiosyncratic political arena.
7.1 USA
With respect to the frequency of natural disasters and their associated financial losses, this
country appears to be an oversampling outlier.19 Indeed, the US landmass occupies 6% of the
world emerged land so that the share of hazards passing over the US should be a nearby figure.
At the same time, the US makes up a quarter of the world GDP over recent decades; one would
then expect its share of natural disaster losses to be in between these figures. Yet, the US occu-
pies more one half of the entries in SwissRes yearly list of 20 most costly insurance losses (over
the last decade 2008-2017). This excess, in all likelihood, is due to the greater penetration of in-
surance in the US. Looking further back in time, the share of US natural disasters in the EM-DAT
roster has been greater than its share of population or share of potential hazards. A probable rea-
son for this over representation is the combination of the following facts: natural disaster trend
studies started in US universities, the US statistical apparatus is far reaching and the newspa-
per density is elevated. Researchers have thus been more efficient at identifying disasters in the
US than in other jurisdictions. This finding makes any roster of world disasters unbalanced, a
feature alluded to in the beginning of the paper; at the same time, the higher quality of natu-
ral disaster coverage in the US offers an opportunity to test for trends with a greater statistical
All our information is sourced at NOAA. Flood casualties and losses are extracted from the
Hydrologic Information Center and the National Weather Service (2017). Losses being expressed
in 2014$ using with the Construction Cost Index of consultancy business “Engineering News-
Record“, we use this very index to reconstruct a nominal dollar losses series. For storms, we
borrow the list of losses (in nominal dollars) published by Pielke et al. (2008). For missing years,
we extract losses from the reports published by NOAA (2018)’s National Hurricane Center as well
as updates for previous years (published between 2011 and 2014). As may be appreciated on Fig-
ure 4, the flood individual risk (ratio of flood casualties to population) has been trending down
even though some years featured dramatic events with a large number of casualties. The econo-
metric analysis is insensitive to the starting year, whether the low 1932 or the high 1934.20 The
total over all natural disasters (excluding heat and cold waves victims) reveals a similar picture
of a statistically significant downward trend.
We then look at the financial risk of floods and storms (including hurricanes) using the US
nominal GDP sourced at the measuringworth website. Figure 5 displays the 5-year moving aver-
age to smooth out the curve as many years feature zero hurricane losses.21 Whereas the financial
19Individual risk is not the issue since US natural disaster casualties are lesser than the country’s share of the
world population.
20The data goes back to 1903 but several years feature zero victims which is highly suspicious. We thus set the
starting point to the 1930s when statistics start to display a greater regularity.
21Note though that the econometric analysis is performed upon the raw time series.
1938$ 1948$ 1958$ 1968$ 1978$ 1988$ 1998$ 2008$
Casualties)per)million)population) $Flood$ $Natural$Disaster$
Figure 4: Natural Disaster Individual Risk in the USA
Slope Estimate Standard Error t-Statistic P-Value
All Hazards -0.0105 0.00224 -4.684 105
Flood -0.0096 0.00286 -3.356 0.001
Table 4: Econometric estimation for the US individual risk
risk from floods has been falling (statistically significative at 4%), that of storms (mostly rare but
devastating hurricanes) has gone up (statistically significative at 1%). Interestingly, the sum of
those two risks which is probably quite close to the total for natural disasters in the US, is statis-
tically trend-less because the positive slope parameter is only significative at the 16% level.
Interpreting this latter result is challenging. For one, Americans have moved from the inland
areas bordering rivers to the coast lines,22 thus moving their riches from flood prone areas to
hurricane prone ones. At the same time, the impact of flood, a very old problem, has probably
been mitigated, like in most advanced countries, by prevention efforts such as restrictive zoning
for new construction and defensive infrastructure building. It may also be the case, as the media
reveal, that no such effort was undertaken in coastal areas, thus precipitating the catastrophic
effect of hurricanes. Disentangling these effects will require using county level socioeconomic
information and geo-localized hazard data. At the outset, we may be observing a typical eco-
nomic phenomenon whereby storms are indeed becoming costlier but, since “all thing else“ are
not equal, the overall share of American wealth destroyed every year by natural disasters is not
increasing (yet).
22Office for Coastal Management (2017) reveals that between 1970 and 2010, coastal shoreline counties added
125 persons per square mile (now reaching 446) while the figure for the entire USA was 36 (now reaching 105). A
quick calculation shows that these areas added population 5 times faster than in non watershed counties (cf. also
Nordhaus (2010) footnote 7).
Figure 5: Natural Disaster Financial Risk in the US
Slope Estimate Standard Error t-Statistic P-Value
Flood -0.007 0.004 4.146 0.045
Storm 0.021 0.008 6.448 0.013
Flood+Storm 0.004 0.008 0.254 0.616
Table 5: Econometric Estimation for the US financial risk
7.2 Bangladesh
This country is also an outlier within disaster rosters for a completely different reason; it suffered
two of the modern times worst catastrophes that notably distort world statistics and trends.23
Hossain (2018) recounts the political landscape of the year 1970 when the Bhola cyclone took
out more than quarter million lives of what was then “East Pakistan“; this drama turned out to
be a trigger for the emancipation movement that ultimately lead to the creation of Bangladesh
in 1971. It was also in response to this devastating event that the government established, with
external financial and technical support, early warning systems and the so-called cyclone pre-
paredness programs (e.g., building concrete shelters). The second dramatic event is cyclone
Gorky (even stronger than Bhola) which struck Bangladesh in 1991 causing an estimated 140000
fatalities. Although this was still a major catastrophe with a death toll of 1.3‰ of the population,
it was three times less than two decades before (individual risk was 4.5‰ for Bhola). Paul (2009)
explains how many lives were saved in 2007 upon the passage of cyclone Sidr thanks to the suc-
cessful application of early warning, mobilization and evacuation plans which had markedly im-
proved after the 1991 cyclone (e.g., shelters in coastal Bangladesh increased from 500 to 4000).24
The death toll fell to a low exact figure of 3406 casualties, reflecting improvements on both safety
as well as administrative record keeping.
23The Bangladesh’s coastline, known as the Bay of Bengal, is one of the world’s most active cyclone area with
about five events every year and certainly the deadliest with an overwhelming share of world casualties to such
24Although Sidr was comparable in strength with Gorky, its landfall was also into a less inhabited area.
The aforementioned catastrophic figures for 1970 and 1991 are so large that they literally
drive the trends. The equation ln(ϕ)=α+βtis estimated below and shows that casualties from
natural disasters in Bangladesh have been falling with virtual certainty at the yearly rate of 6%,
as illustrated with Figure 6.
Figure 6: Death per million to natural disasters in Bangladesh
Estimate Standard Error t-Statistic P-Value
α131.0 27.09 4.836 0.00001
β-0.065 0.014 -4.759 0.00001
Table 6: Econometric Trend estimation for Bangladesh
7.3 Developing vs. Developed
We previously argued that insurance penetration is a key to the correct estimation of disaster
losses; this feature clearly separates developing and developed nations since insurance penetra-
tion is closely related to overall wealth. We select the 1980-2017 period where data coverage of
natural disasters occurring in the developing world is adequate25 and associate the “developed“
label solely with the members of the Organisation for Economic Co-operation and Development
(OECD), for lack of a better discriminating criteria.
For each of our three risk ratio, Table 7 displays the long run average, the slope parameter
and the probability that it is non zero (so-called P-value). Using the IPCC jargon, we may say that
the individual risk νat world level is likely to be falling, driven by the even stronger likely fall of
νwithin the developing world. At world level, natural disasters claimed 11 people per million
population over this long period but in a very unequal fashion as the risk in the developing
nations triples that of the OECD (at 4 vs. 12 fatalities per million population).
Financial losses wrought by natural disasters consume 20 basis points or 2 ‰ of the world
GNI (in the long run) but the figure rises to 4 ‰ once we relate local losses to local wealth. Per
25Doing so, we only drop 300 records out of 4300 with positive damages in the EM-DAT roster.
Indicator Individual Risk Financial Risk Socioeconomic Risk
Region νslope Prob ϕslope Prob φslope Prob
Developing 12.30 -0.37 0.27 35.01 -0.47 0.14 45.26 -0.47 0.32
World 10.99 -0.28 0.31 19.84 0.15 0.37 40.59 -0.34 0.39
OECD 4.04 0.12 0.55 15.89 0.28 0.19 15.59 0.18 0.35
OECDUS JP 5.09 0.13 0.70 9.91 -0.22 0.12 11.02 -0.31 0.05
Table 7: Econometric analysis developed vs. developing
IPCC terminology, it is “likely“ that financial losses are falling in the developing world and rising
in the OECD (probabilities of 14% and 19% respectively). The more meaningful socioeconomic
loss indicator however dampens our assurance that trends exist since the statistical confidence
falls (probabilities of 32% and 35% respectively). That the financial risk is twice larger in de-
veloping countries (wrt. the rich world) is a foregone conclusion but one that masks an even
starker reality since the socioeconomic risk φis trice greater in the developing world compared
to the OECD (4.5‰ vs. 1.6‰).
As Japan is situated in a zone prone to earthquakes, this country is a geophysical outlier in
our global sample. The econometric test for this country reveals no individual risk trend but
a clear upward trend for the financial risk. If we now consider the OECD without the US nor
Japan, that is to say Western Europe, then the socioeconomic risk is found to trend downward
(as shown in the last line of Table 7). Note also the significant level difference since the financial
risk is 1.8‰ in the US and Japan while it is almost half smaller at 1‰ in Europe where the climate
is milder (cf. Boccard (2018)).
8 Conclusion
This study has used the sparse publicly available data on natural disasters to confirm and am-
plify some known facts but also to dispel some incorrect statements reported in the media and,
needlessly, in the UN leaflets. We have recall the difficulties inherent in defining and delineating
natural disasters as well as the paucity of trustable sources of information. Even so, we firmly
believe that if any statistic should be produced out of these data, it ought to be done properly.
We thus argue for a focus on risk ratios which, incidentally, scarcely feature in diffusion reports
by the media or even academia.
After recalling the last trends for the classical absolute indicators of natural disasters, we
show that over the long run, the risk for a random earth inhabitant of dying in a natural disaster
is without trend. Next, the ratio of losses to wealth, the so-called financial risk is found to be
trending up at world level over the long run. To better account for fundamental differences of
development between countries, we compute this ratio at country level and aggregate figures
using population as weights. We obtain a so-called socioeconomic risk which is shown to be
without a trend. Over the long run, financial losses consume about 2‰ of the gross national
income while socioeconomic ones are twice as large because poor countries suffer about three
times more destruction for a comparable natural hazard.
Lastly, we tackle two outlier countries which illustrate the variability of country cases, and
ultimately the need to handle global statistics carefully. The USA is overrepresented in disaster
databases but allows analysis over a long time frame. In this most advanced country, individual
risk is found to be falling. Regarding the risk of losses, the flood risk is found to trend down
while the storm risk is rising; interestingly, the sum of these two major risks fails to display a
trend. We last expound the case of Bangladesh, a country subject to recurrent cyclonic activity.
Having been victim of devastating catastrophes, Bangladesh has been able to prepare better
and drive down individual risk to an acceptable level. The striking difference in the statistical
and insurance apparatus in these examples pushes us to divide the world into developed and
developing (for the entire 1980-2017 period). We confirm and amplify, for the most recent data,
some well know facts, namely that individual risk and socioeconomic risk is about three times
larger in the developing world.
Aon Benfield. Weather, climate & catastrophe insight. Technical report, Aon Benfield, 2017.
Below R., Wirtz A., and Guha-Sapir D. Disaster category classification and peril terminology for
operational purposes. Technical report, CRED, 2009.
Boccard N. Natural disasters over france: An inter-temporal assessment.Weather and Climate
Extremes, to appear, 2018.
Case A. and Deaton A. Mortality and morbidity in the 21st century. Technical report, Brookings
Institution, 2017.
CDC. Overdose death rates. Technical report, Centers for Disease Control and Prevention, 2018.
Dworkin J. Global trends in natural disasters 1947 - 1973. Technical report, Natural Hazards
Research and Applications Information Center, University of Colorado Boulder, 1974.
Estrada F., Botzen W., and Tol R. Economic losses from us hurricanes consistent with an influ-
ence from climate change.Nature Geoscience, 8, 2015.
Faust E., Höppe P., Wirtz A., and Schmidt S. Trends in natural catastrophes – potential role of
climate change. In Workshop on Climate Change and Disaster Losses: Understanding and
Attributing Trends and Projections, 2006.
Glickman T. S., Golding D., and Silverman E. D. Acts of god and acts of man – recent trends in
natural disasters and major industrial accidents. Technical Report 92-02, Resources for the
Future, Washington DC., 1992.
Guha-Sapir D., Hoyois P., and Below R. Annual disaster statistical review. Technical report,
CRED, 2017.
Hewitt K. and Sheehan L. A pilot survey of global natural disasters of the past twenty years.
Technical report, Natural Hazards Research and Applications Information Center, University
of Toronto, 1969.
Hoeppe P. Trends in weather related disasters – consequences for insurers and society.Weather
and Climate Extremes, 11(Supplement C):70 – 79, 2016.
Hossain N. The 1970 bhola cyclone, nationalist politics, and the subsistence crisis contract in
bangladesh.Disasters, 42(1):187–203, 2018.
IPCC. Managing the risks of extreme events and disasters to advance climate change adaptation.
Technical report, Intergovernmental Panel on Climate Change, 2012.
IPCC. Climate change 2014: Impacts, adaptation, and vulnerability. Technical report, Intergov-
ernmental Panel on Climate Change, 2014.
Irogbe K. The persistence of famine in sub-saharan africa.The Journal of Social, Political, and
Economic Studies, 38(4):441–461, 2013.
Keiler M. Worldwide trends in natural disasters, pages 1111–1114. Springer Netherlands, Dor-
drecht, 2013.
Linnerooth-Bayer J. and Hochrainer-Stigler S. Financial instruments for disaster risk manage-
ment and climate change adaptation.Climatic Change, 133(1):85–100, 2015.
Loayza N. V., Olaberría E., Rigolini J., and Christiaensen L. Natural disasters and growth: Going
beyond the averages.World Development, 40(7):1317–1336, 2012.
Mastrandrea M. D., Mach K. J., Plattner G.-K., Edenhofer O., Stocker T. F., Field C. B., Ebi K. L.,
and Matschoss P. R. The IPCC AR5 guidance note on consistent treatment of uncertainties: a
common approach across the working groups.Climatic Change, 108(4):675, 2011.
MunichRe. NatCatSERVICE methodology. Technical report, NatCatSERVICE, 2018.
National Weather Service. Natural hazard statistics. Technical report, NOAA, 2017.
Neumayer E. and Barthel F. Normalizing economic loss from natural disasters: A global analysis.
Global Environmental Change, 21(1):13–24, 2011.
NOAA. Tropical cyclone reports. Technical report, National Hurricane Center, 2018.
Nordhaus W. The economics of hurricanes and implications of global warming.Climate Change
Economics, 1:1–20, 2010.
Noy I. A global comprehensive measure of the impact of natural hazards and disasters.Global
Policy, 7(1):56–65, 2016.
O’Brien S. Statement to the security council on missions to yemen, south sudan, somalia and
kenya and an update on the oslo conference. Technical report, United Nations Office for the
Coordination of Humanitarian Affairs, March 2017.
OECD. Guidelines for aid agencies on disaster mitigation. Technical report, OECD, 1994.
Office for Coastal Management. National coastal population report. Technical report, NOAA,
O’Keefe P., Westgate K., and Wisner B. Taking the naturalness out of natural disasters.Nature,
260, 1976.
Paul B. K. Why relatively fewer people died? the case of bangladeshs cyclone sidr.Natural
Hazards, 50(2):289–304, 2009.
Pielke R. Mistreatment of the economic impacts of extreme events in the stern review report on
the economics of climate change.Global Environmental Change, 17(3):302–310, 2007.
Pielke R. A. and Landsea C. W. Normalized hurricane damages in the united states: 1925–95.
Weather Forecast, 13:621–631, 1998.
Pielke R. A., Gratz J., Landsea C. W., Collins D., Saunders M. A., and Musulin R. Normalized
hurricane damage in the united states: 1900–2005.Natural Hazards Review, 9(1):29–42, 2008.
Quarantelli E. L. What is disaster? Technical report, Disaster Research Center, University of
Delaware, 1985.
Quarantelli E. L. Statistical and conceptual problems in the study of disasters.Disaster Preven-
tion and Management: An International Journal, 10(5):325–338, 2001.
Sawada Y. and Takasaki Y. Natural disaster, poverty, and development: An introduction.World
Development, 94:2–15, 2017.
Stern N. The economics of climate change the stern review. Cambridge University Press, 2007.
SwissRe. Methodology of sigma explorer data. Technical report, SwissRe, 2017.
Thompson S. A. Trends and developments in global natural disasters, 1947 to 1981. Technical re-
port, Natural Hazards Research and Applications Information Center, University of Colorado
Boulder, 1982.
UNDRO. Natural disasters and vulnerability analysis : report of expert group meeting. Technical
report, Office of the United Nations Disaster Relief Coordinator, 1980.
UNISDR. Global assessment report on disaster risk reduction. Technical report, United Nations
Office for Disaster Risk Reduction, 2016.
UNISDR. Infographics. Technical report, United Nations Office for Disaster Risk Reduction,
UNISDR. Terminology. Technical report, United Nations Office for Disaster Risk Reduction,
UNISDR. Indicators to monitor global targets of the sendai framework for disaster risk reduc-
tion 2015-2030: A technical review. Technical report, United Nations Office for Disaster Risk
Reduction, 2015.
Visser H., Petersen A., and Ligtvoet W. On the relation between weather-related disaster impacts,
vulnerability and climate change.Climatic Change, 125(3-4):461–477, 2014.
World Peace Foundation. Famine trends dataset, 2018.
... All those aspects considered, researchers and analysts suggest the years 1970 or 1980 as the years to be considered as an adequate starting point for the study of the dimensions of natural disasters at a global level, for instance from an economic point of view (IPCC 2012;CRED 2017;Boccard 2018;SwissRe2019). ...
... When the sources of the data are international or intergovernmental agencies, the problem often may be the lack of information transmission (bottom-up process), due to higher or lower cooperative predisposition of local/national administrations, or, when the work is a data collection from a central service, the problem may be related to non-English speaking countries, even more so when its news networks are poorly developed. In addition, sometimes, mechanisms are influenced by the fact that some countries or territories voluntarily overestimate data on the consequences of disasters as a strategy, for example, in an attempt to obtain more financial aid, whereas some kinds of political regime may prefer to underestimate data on consequences of disasters for reasons of internal national consensus (Boccard 2018). ...
This book addresses key issues across the field of sustainable urban planning with the aim of providing a unique reference tool for planners, engineers, architects, public administrators, and other experts. The evolution of cities and communities is giving rise to pressing energy and environmental problems that demand concrete solutions. In this context, urban planning is inevitably a complex activity that requires a sound analytical reading of ongoing developments, multidisciplinary analysis of the available tools and technologies, appropriate political management, and the ability to monitor progress objectively in order to verify the effectiveness of adopted policies. Green Planning for Cities and Communities is exceptional in the breadth of its coverage and the way in which it highlights the interactions among different elements. Individual sections focus on strategies and tools for green planning, energy efficiency and sustainability in city planning, sustainable mobility, rating systems, and the smart city approach to improve urban-scale sustainability. The authors draw on extensive practical experience to provide operational content to the theoretical and methodological elements covered in the text, and each section is enriched by informative case studies.
... All those aspects considered, researchers and analysts suggest the years 1970 or 1980 as the years to be considered as an adequate starting point for the study of the dimensions of natural disasters at a global level, for instance from an economic point of view (IPCC 2012;CRED 2017;Boccard 2018;SwissRe2019). ...
... When the sources of the data are international or intergovernmental agencies, the problem often may be the lack of information transmission (bottom-up process), due to higher or lower cooperative predisposition of local/national administrations, or, when the work is a data collection from a central service, the problem may be related to non-English speaking countries, even more so when its news networks are poorly developed. In addition, sometimes, mechanisms are influenced by the fact that some countries or territories voluntarily overestimate data on the consequences of disasters as a strategy, for example, in an attempt to obtain more financial aid, whereas some kinds of political regime may prefer to underestimate data on consequences of disasters for reasons of internal national consensus (Boccard 2018). ...
In green planning, the correct management of energy in towns and cities is a critical issue and is the topic of this chapter. In the first part, a general overview is provided, identifying the new paradigms to be pursued according to the needs of reducing, and possibly eliminating, the emission of greenhouse gases and those gases that have an impact on the environment, also considering the technological evolution currently in progress and opportunities offered by the market of energy generation and management market. In the second part of the chapter, a methodological approach to Green Energy Planning on a territorial scale is proposed and discussed. The chapter is completed by an in-depth analysis of the European project which promotes an approach to bottom-up energy planning, the Covenant of Mayors: its methodologies are discussed and a case study is analysed.
... Estimates show that the frequency and intensity of flooding have increased over the past several decades partly due to climate variability and 1 3 change; low adaptive capacity and inadequate infrastructure to deal with these disasters (Gain et al. 2015a, b;Shah et al. 2019a, b). As the world population and global economy continue to grow at the backdrop of increasing frequency and severity of floods, the number of people at risk is also expected to increase especially in developing countries (Gain et al. 2015a, b;Boccard 2018;Munyai et al. 2019). ...
Full-text available
Flood risk studies remain focused on understanding causation, impacts, perceptions and coping and adaptation measures and there remain many inappropriate measures taken to reduce household risk. Through a case study approach, this study assessed rural household vulnerability to flooding which helped to determine the level of risk to flood hazards in Mbire District of Zimbabwe. Three hundred and thirty-five (335) households were randomly selected to participate in the survey. The composite index approach was used to calculate vulnerability and risk indices. Results indicated that more than 70% of the households were highly vulnerable to flooding and about 50% were in the medium to very high categories of flood risk. Household vulnerability was not only a question of the natural flood event, but a result of the intersection of poor socio-economic conditions and the physical exposure to floods. The socio-economic determinants of vulnerability included high levels of unemployment resulting in low income and poverty; low levels of education; and over-dependence of the community on rain-fed smallholder farming. Physical exposure of the community mainly resulted from materials used to construct houses. Pole and dagga houses had a higher degree of exposure compared to houses with bricks and cement. Therefore, we recommend that instead of focusing on the whole ward/district the institutional services should first target the most vulnerable and at risk households in order to increase their capacity in dealing with flood hazards.
... Data EMDAT menunjukkan bahwa tujuh dari sepuluh negara dengan kejadian bencana paling sering terjadi adalah negara-negara berkembang seperti Indonesia, Afganistan, Filipina, dan Nigeria (Guha-Sapir, 2018). Menurut Boccard (2018) tren kejadian bencana alam yang melanda negara-negara di dunia dipengaruhi oleh beberapa faktor, yaitu, kondisi fisik wilayah serta kerentanan sosial, ekonomi, dan kultural. ...
Full-text available
This paper aims to further examine the opportunities and challenges of the industrial revolution 4.0 in disaster preparedness management. The method used is qualitative by analyzing articles obtained from Google Scholar with keywords industrial revolution 4.0 and disaster preparedness. The results of an analysis of documents related to the industrial revolution 4.0 found that cloud computing technology, artificial intelligence, big data, and the internet of things are useful in collecting data, mapping, and design of disaster risk information in real time and up to date. These conditions make management of disaster management more effective and efficient because the data is presented accurately so as to minimize errors or data distortion. Challenges that may be faced in the integration of the industrial revolution 4.0 technology and disaster preparedness management are also discussed.
... Natural Beyond the average levels achieved over the recent decades, we may consider the evolution of economic losses due to disasters. As shown in Boccard (2018), natural disaster losses are growing over the period period 1970-2017 while man-made ones appear to be receding (cf. ...
Full-text available
We tackle the incidence of accidents within the energy supply chain and firstly extend the analysis from severe accidents to smaller ones. We are then able to go beyond fossil fuels technologies and estimate the hazard rate (ratio of casualties to energy) of wind power, the electricity network and the nuclear sector (for latent victims). Technologies are ranked, separately in the developed and developing worlds. In a second part, we compute the risk rate (ratio of casualties to population) for a variety of countries, accounting for the energy mix and imports; differences are found to be less glaring than for hazard rates. Lastly, we compare this risk of energy supply with the negative health impacts of energy consumption such as atmospheric pollution and road accidents. We find that for every casualty within the energy supply chain, there is a hundred more casualties among end-users in the developed countries and a thousand more in the developing ones. These stark differences call for giving priority to policies aimed at reducing the negative externalities of energy production and consumption.
... As detailed in Boccard (2018), the natural disaster rosters featuring prominently in the literature, namely EM-DAT and MunichRe, focus on major disasters (aka catastrophes), eschewing hundreds of less dramatic events that are no less consequential for their victims. 4 At issue is the small sample size since natural catastrophes are unfrequent events. ...
Full-text available
Using an exhaustive administrative database, we assess the impact of extreme weather events over French cities during the last three decades. We identify numerous non catastrophic disasters, thereby improving coverage in the existing literature. Counting residents of cities stricken by a disaster, we find that in the long run, there were 23 residents affected every month per thousand population. This risk factor has been falling by 4 fewer people with every passing decade.France has thus improved its preparedness to natural disasters even though the seaboard regions fare worse than the northern region, most likely because of heightened urban pressure in risky areas. Tropical territories are more at risk than the temperate European mainland, from a different mix of events. The full economic cost of natural disasters at 65€ per capita, 22€ of which is currently uninsured, is a small fraction of property insurance premiums. Safe areas residents are currently subsidizing those living in riskier areas. Preventive investments should be directed at the main cities to be most effective.
Conference Paper
Full-text available
The Fourth Industrial Revolution or industry 4.0 is the entry point of the digitalization era which refers to the current situation of life style changes and the behavior of individuals and organizations today. This study aims to discuss an increasingly complexp public relations strategy after under going a transformation asitenters industrial era 4.0. By using literature review method, this research is proposed descriptively to analyze public relations strategy that carries digitalization strategy and mobilization. The results found in this study are, the current community relations industry has been adopting a number of strategies that prioritize funding efficiency, high mobilization, and greater target achievement. Soit can be concluded the public relations strategy today reflects the pattern that arises because of the disruption of industry 4.0. Keywords: Communication Digital, Revolution Industri 4.0, Lifestyle
This chapter illustrates the need for the inclusion in sustainable urban planning of the problem of the impact of disasters on cities and communities: disasters that will continue to occur and may well even become more frequent because of urban expansion, increasing population densities and climate change effects. To achieve this result, first of all some basic tools about disaster terminology, disaster trends and disaster risk computation are provided. Subsequently, the latest international decisions at global level to face disasters and climate change and to promote sustainability are explained and elucidated, showing their connections and synergies. In the final part, some originally selected elements on how to put in place strategies and those actions which are no longer deferrable are proposed.
Full-text available
Using an exhaustive administrative database, we assess the impact of extreme weather events over French cities between 1982 and 2017. We identify numerous non-catastrophic disasters, thereby improving coverage wrt. the existing literature. Counting residents of cities stricken by a disaster, we find that in the long run, there were 22 residents affected every month per thousand population. This risk factor has been falling by 5 fewer people with every passing decade. France has thus improved its preparedness to natural disasters even though the seaboard regions fare worse than the northern region, most likely because of heightened urban pressure in hazardous areas by the seaside. Tropical territories are more at risk than the temperate European mainland, from a different mix of events. The full economic cost of natural disasters is estimated at 22 € per capita per year and represent a small fraction of property insurance premiums. Residents from safer areas currently subsidize those living in riskier areas. To be more effective, preventive investments should be directed towards the main cities.
Full-text available
The devastating Bhola cyclone in November 1970 is credited with having triggered the political events that led to the division of Pakistan and the creation of Bangladesh in 1971. A callous response to the disaster by the Pakistani regime resulted in a landslide electoral victory for Bengali nationalists, followed by a bitter and bloody civil war. Yet, despite its political momentousness, the Bhola cyclone has been the subject of little political analysis. This paper examines the events, arguing that its extraordinary political significance put disaster management on the nationalist agenda; the famine of 1974 confirmed its centrality, producing a social contract to protect the population against disasters and subsistence crises on which the country’s acclaimed resilience to the effects of climate change rests. The Bhola cyclone also drew international attention to this neglected, little-known region, and in general can be seen as foundational for the subsequent developmental achievements of Bangladesh.
Technical Report
Full-text available
Executive Summary In 2015, 376 natural triggered disasters were registered. After the lowest number since the beginning of the century in 2014 (330), this increase could be a sign of a reversal in the trend to decline in the annual number of disasters since 2005, even if the 2015 number remains below its average annual for the period 2005-2014 (380). Last year natural disasters made still 22,765 deaths, a number largely below the annual average for years 2005-2014 (76,416), and made 110.3 million victims worldwide, also below the 2005-2014 annual average (199.2 million) (see Figure 1). Like the other indicators, with estimates placing economic damages at US$ 70.3 billion, natural disasters costs were, in 2015, significantly below their decennial average of US $ 159.7 billion. The increase in the number of reported natural disasters in 2015, was mostly due to a higher number of climatological disasters: 45 compared with the 2005-2014 annual average of 32, an increase of 41%. The number of meteorological disasters (127) was 2% above its decadal average (125) while, inversely, the number of hydrological disasters (175) and of geophysical disasters (29) were, both, 9% below their 2005-2014 annual average of, respectively, 192 and 32. As each year since 2005, the number of hydrological disasters still took by far the largest share in natural disaster occurrence in 2015 (46.5%, for a mean proportion of 50.6% for the period 2005-2014), followed by meteorological disasters (33.8% versus a decadal mean proportion of 32.7%), while climatological disasters (12% versus an annual mean proportion of 8.3%) overpassed geophysical disasters (7.7% for a 2005-2014 mean proportion of 8.4%) Over the last decade, China, the United States, India, the Philippines and Indonesia constitute together the top 5 countries that are most frequently hit by natural disasters. In 2015, with 36 natural disasters reported, China experienced its third highest number of natural disasters of the last decade, 20% above its 2005-2014 annual average of 30. The country was affected by a variety of disasters types, including 17 storms, 13 floods and landslides, 5 earthquakes and one drought. The number of natural disasters in the United States (28) was as high as in 2013, and 33% above its decadal annual average of 21. With 21 disasters, its third highest number since 2005, India is 24% below its 2005-2014 annual average of 27. Inversely, with respectively 15 and 10 natural disasters, the Philippines and Indonesia knew their 4th and 2nd lowest numbers since 2005, below their respective annual average of 18 and 14. In 2015, the number of people killed by disasters (22,765) was the lowest since 2005, way below the 2005-2014 annual average of 76,416 deaths which, however, takes into account two years with more than 200,000 people reported killed, each time mostly attributable to major catastrophes: the cyclone Nargis in Myanmar in 2008 (138,366 deaths) and the earthquake in Haiti in 2010 (225,570 deaths). But even after exclusion of these disasters, the number of deaths in 2015 remains below a recomputed 2005-2014 annual average of 40,022 deaths. At a more detailed level, it appears that, in 2015, earthquakes and tsunamis killed the most people (9,526) however far below a 2005-2014 annual average of 42,381. Extreme temperatures made 7,418 deaths, the second highest number since 2005 but far below the peak of 2010 (57,064). Inversely, the number of deaths from floods (3,449) and storms (1,260) were, both, the lowest since 2005, far below their 2005-2014 annual averages (5,933 and 17,769, respectively). Amongst the top 10 countries in terms of disaster mortality in 2015, six countries are classified as low-income or lower-middle income economies (see World Bank income classification), and accounted for 67.6% of global reported disaster mortality. Four disasters killed more than 1,000 people in 2015: the Gorkha earthquake in Nepal of April (8,831 deaths) and three heat waves in France between June and August (3,275 deaths), in India in May (2,248 deaths) and in Pakistan in June (1,229 deaths). The number of victims in 2015 (110.3 million) was the second lowest since the decade, far below its 2005-2014 annual average (196.3 million). It must be noted that the four years with the lowest number of victims since 2005 are the four last years, 2012 to 2015, far below the 200 million victims reported between 2007 and 2011. This decrease is mainly explained by the lower human impact of floods, whose number of victims (36.1 million) was the second lowest since 2005, 58.4% below its 2005-2014 annual average (86.9 million) and of storms with a number of victims (10.4 million) 70.2% below its decade’s average (34.9 million). The number of victims of climatological disasters (54.3 million) was near its 2005-2014 average (56.7 million). Geophysical disasters made 8.1 million victims, a number lightly below the 8.6 million annual average, but however the second highest since 2005, after the very high peak of 2008 (47.7 million). Nine countries of the top ten countries in terms of number of victims were low or lower-middle income countries, accounting for 69.9% of the victims of 2015. The natural events that accounted for more than 10 million victims were two droughts in DPR Korea in June and July (18 million victims) and in Ethiopia, from September (10.2 million) and floods in India in July and August (13.7 million). Twenty other disasters (10 droughts, 5 floods, 4 storms and one earthquake) had severe human impacts ranging from 1 to 9 million victims. The estimated economic losses from natural disasters in 2015 (US$ 70.3 billion) was the third lowest since 2005 and 56 % below the annual 2005-2014 damages average (US$ 159.8 billion). The lowering in the amount of damages come from geophysical (US$ 6.7 billion; -86.0% compared to the 2005-2014 average), meteorological disasters (US$ 33.4 billion; -51.7% compared to the 2005-2014 average) and hydrological disasters (US$ 21.3 billion; -38% compared to the 2005- 2014 average). Damages from earthquakes were the second lowest since 2005, and represent 8.7% of all disaster costs. Those from storms and floods were, both at their third lowest since 2005, contributing, respectively, to 47.4 and 30.3% of all disaster costs. These three disaster types are at the origin of almost all these costs. On their side, damages from climatological disasters (US$ 8.9 billion) were, in 2015, very near their 2005-2015 annual average (US$ 8.8 billion), however if in this disaster category, damages from droughts and from wildfires were, both, the fourth lowest since 2005, costs of droughts (US$ 5.8 billion) were slightly below their decadal average (US$ 6.4 billion) while those from wildfires (US$ 3.1 billion) were 27.9% above their 2005- 2014 annual average. In the top ten countries for economic damages, six were high or upper-middle income countries which accounted for 70.7% of the total economic losses while the share of the four low and lowermiddle income countries in this total was of 17.6%. The costliest natural disaster in 2015 was the Gorkha earthquake, in Nepal, which cost US$ 5.7billion to the country, while typhoon Mujigae impacted China for a total of US$ 4.2 billion economic losses. Twenty-one other disasters (9 storms, 7 floods, 3 droughts and 2 wildfires) accounted for damages ranging from US$ 1 to 3 billion. The total costs of these 23 disasters represent 61.2% of all reported damages in 2015. Looking at the distribution of disasters across continents, it appears that Asia was most often hit (44.4%), followed by the Americas (25.5%), Africa (16.5%), Europe (7.2%) and Oceania (6.4%). This regional distribution of disaster occurrence is, in 2015, not very different from the profile observed between 2005 and 2014. However, the share of Europe in the distribution is half its 2005-2014 mean proportion, while the share of Oceania is, in 2015, twice its average. Asia accounted in 2015 for 62.7% of worldwide reported disaster victims (against 80.6% for the 2005-2014 decade’s average), while Africa accounted for 28.0% (against 13.1% on average for the 2005-2014 period) and the Americas for 7.0% (against 5.8% on average for 2005-2014). Oceania accounted for 2.2% of all natural disasters victims (against 0.1% for 2005-2014 average) and Europe for only 0.21% (against 0.35% according to the 2005-2014 average). With 49.1% of worldwide natural disaster reported costs, Asia suffered the most damages in 2015, followed by the Americas (36.7%) and Europe (6.8%). A share of 5.1% of global disaster damages was reported for Oceania and of 2.4% for Africa. In spite of some differences in the proportions, the ranking of the continents according to their contribution to the total reported damages is similar from the one observed over the last decade, where Asia had the most damages, followed by the Americas and Europe. However, when compared to its 2005-2014 average, the amount of damages in Africa was significantly above its 2005-2014 annual average of 0.34%. EM-DAT’s global approach to the compilation of disaster data continuously provides us with valuable information and trends on the occurrence of natural disasters and their impacts on society. However, the development of guidelines and tools for the creation of national and subnational disaster databases; for the compilation of standardized, interoperable disaster occurrence and impact data remain priorities for the strengthening of tools helping to benchmark and orientate effective disaster risk reduction programs.
Building on our earlier research (Case and Deaton 2015), we find that mortality and morbidity among white non-Hispanic Americans in midlife since the turn of the century continued to climb through 2015. Additional increases in drug overdoses, suicides, and alcohol-related liver mortality- particularly among those with a high school degree or less-are responsible for an overall increase in all-cause mortality among whites. We find marked differences in mortality by race and education, with mortality among white non- Hispanics (males and females) rising for those without a college degree, and falling for those with a college degree. In contrast, mortality rates among blacks and Hispanics have continued to fall, irrespective of educational attainment. Mortality rates in comparably rich countries have continued their premillennial fall at the rates that used to characterize the United States. Contemporaneous levels of resources-particularly slowly growing, stagnant, and even declining incomes-cannot provide a comprehensive explanation for poor mortality outcomes. We propose a preliminary but plausible story in which cumulative disadvantage from one birth cohort to the next-in the labor market, in marriage and child outcomes, and in health-is triggered by progressively worsening labor market opportunities at the time of entry for whites with low levels of education. This account, which fits much of the data, has the profoundly negative implication that policies-even ones that successfully improve earnings and jobs, or redistribute income-will take many years to reverse the increase in mortality and morbidity, and that those in midlife now are likely to do worse in old age than the current elderly. This is in contrast to accounts in which resources affect health contemporaneously, so that those in midlife now can expect to do better in old age as they receive Social Security and Medicare. None of this, however, implies that there are no policy levers to be pulled. For instance, reducing the overprescription of opioids should be an obvious target for policymakers.
This paper examines the problem of famine in postindependence sub-Saharan Africa with a view toward determining appropriate remedies. Some scholars place the problem on the legacy of imbalanced economic development under colonialism, on climate change, on extremely rapid population growth resulting from global disease control measures, on excessive urbanization, and on natural disasters such as drought. This paper takes a holistic approach by contending that food shortages in post-independence sub-Saharan Africa have been caused by not only these factors but also by a shortage of modern agricultural machinery, by governmental mismanagement and corruption, by wars, by a brain drain to more prosperous countries, and even by food aid which serves as a disincentive to local production. It is further argued that the pervasiveness of famine will continue until a genuinely transparent, accountable, and responsive governmental system that recognizes the values of the rule of law and good governance has been firmly established throughout the region.
This paper introduces the special issue, "Natural Disaster, Poverty, and Development." We examine the macro-level nexus between natural disasters and poverty, discuss prospects for formal insurance against disasters, and review the micro-development literature on informal insurance against risk. We develop a conceptual framework for microeconomic analyses on the disaster-poverty nexus, highlighting asset loss/recovery and asset-dependent private coping, disaster aid and its link with private mechanisms, and broad/persistent impacts of disasters and coping responses. We synthesize the main findings of the nine articles, revealing the critical importance of the complementarity among markets, governments, and communities for successful pro-poor disaster policies.