Download PDF Download

A joint initiative of Ludwig-Maximilians-Universität and the Ifo Institute for Economic Research
VOLUME 11, NO. 2
Forum
SUMMER
2010
Focus
NATURAL DISASTERS
Wolfgang Kron
Stéphane Hallegatte and
Valentin Przyluski
Eduardo Cavallo and
Ilan Noy
David Hofman
Hideki Toya and
Mark Skidmore
Yasuhide Okuyama
Makena Coffman and
Ilan Noy
Peter Gordon,
James E. Moore II,
Jiyoung Park and
Harry W. Richardson
Bradley T. Ewing,
Jamie B. Kruse and
Mark A. Thompson
Jacob Vigdor
Mario Mazzocchi,
Francesca Hansstein and
Maddalena Ragona
Specials
A EURO RESCUE PLAN
Wolfgang Franz,
Clemens Fuest,
Martin Hellwig and
Hans-Werner Sinn
RECENT FISCAL DEVELOPMENTS
AND EXIT STRATEGIES
Vito Tanzi
Spotlight
GOVERNMENT DEBT
IN
EUROPE
Christian Breuer and
Matthias Müller
Trends
STATISTICS UPDATE
CESifo Forum ISSN 1615-245X
A quarterly journal on European economic issues
Publisher and distributor: Ifo Institute for Economic Research e.V.
Poschingerstr. 5, D-81679 Munich, Germany
Telephone ++49 89 9224-0, Telefax ++49 89 9224-98 53 69, e-mail [email protected]
Annual subscription rate: n50.00
Single subscription rate: n15.00
Shipping not included
Editors: John Whalley ([email protected]) and Chang Woon Nam ([email protected])
Indexed in EconLit
Reproduction permitted only if source is stated and copy is sent to the Ifo Institute
www.cesifo-group.de
Forum
Volume 11, Number 2
Summer 2010
_____________________________________________________________________________________
Focus
NATURAL DISASTERS
Natural Catastrophes: Do We Have to Live with Them?
Wolfgang Kron
3
The Economics of Natural Disasters
Stéphane Hallegatte and Valentin Przyluski
14
The Aftermath of Natural Disasters: Beyond Destruction
Eduardo Cavallo and Ilan Noy
25
Mitigating the Impact of Natural Disasters on Public Finance
David Hofman
36
Natural Disaster Impacts and Fiscal Decentralization
Hideki Toya and Mark Skidmore
43
Globalization and Localization of Disaster Impacts: An Empirical Examination
Yasuhide Okuyama
56
A Hurricane Hits Hawaii: A Tale of Vulnerability to Natural Disasters
Makena Coffman and Ilan Noy
67
Short-Run Economic Impacts of Hurricane Katrina (and Rita)
Peter Gordon, James E. Moore II, Jiyoung Park and Harry W. Richardson
73
Measuring the Regional Economic Response to Hurricane Katrina
Bradley T. Ewing, Jamie B. Kruse and Mark A. Thompson
80
What Should the World Do about Port-au-Prince? An Economic Assessment
Jacob Vigdor
86
The 2010 Volcanic Ash Cloud and Its Financial Impact on the European Airline Industry
Mario Mazzocchi, Francesca Hansstein and Maddalena Ragona
92
Specials
A Euro Rescue Plan
Wolfgang Franz, Clemens Fuest, Martin Hellwig and Hans-Werner Sinn
101
Comments on Recent Fiscal Developments and Exit Strategies
Vito Tanzi
105
Spotlight
Government Debt in Europe
Christian Breuer and Matthias Müller
111
Trends
Statistics Update
113
Focus
NATURAL DISASTERS
age), the snow damage in China in early 2008 (more
than 21 billion US dollars) and droughts like the one
in Southeast Asia in 2009–2010 – let alone the many
‘silent’ disasters in Africa – which go unnoticed
because their onset is not sudden. In poor countries,
natural catastrophes often do not produce high monetary loss numbers in absolute terms, and sometimes
not even high death tolls, but nevertheless they may
be more severe and momentous for the country
affected than, for instance, hurricane Katrina for the
United States.
NATURAL CATASTROPHES:
DO WE HAVE TO LIVE WITH
THEM?
WOLFGANG KRON*
Introduction: recent catastrophic events
Hardly a year passes in which at least one country
around the world suffers from a large natural catastrophe. The first months of 2010 have already seen
a series of severe earthquakes in different parts of
the world such as Haiti, Chile, Indonesia, China and
Mexico, and a volcanic eruption in Iceland. Winter
storms in Europe (Kyrill 2007, Klaus 2008, Xynthia
2009) remind us regularly that billion-dollar loss
events are a continuing threat to Europe, as are
widespread floods. The ones in Britain in 2007, along
the lower Danube in 2006, in the Alps in 2005, and in
central Europe in 2002 all set new loss records in the
regions where they occurred. In 2004 and 2005, hurricanes in the North Atlantic – Katrina, Wilma, Rita
(2005) and Ivan (2004) to mention just a few names
– did the same with respect to their number and
monetary losses. Nargis, a tropical cyclone in the
Gulf of Bengal in 2008, devastated the Irrawaddy
Delta through wind and storm surge and cost the
lives of more than 140,000 Burmese. The Philippines
were drenched in 2009 by enormous amounts of
rainfall. Countless numbers of flash floods all over
the world, of which only events like those in Istanbul
(September 2009), Madeira (February 2010) and Rio
de Janeiro (April 2010) became generally known,
claim lives and cause huge destruction practically
every day.
On the other hand, events which cause hardly any
physical damage – like the eruption of the Eyjafjallajökull volcano in Iceland (April 2010) – may
produce costs running into several hundred million
US dollars a day by interrupting private and business
lives and the flow of goods.
From the above, it becomes clear that disasters can
assume different forms: in terms of scale (regional
intensity or large-scale impact), number of fatalities,
monetary losses and impact on the local economy. It
is without doubt, however, that natural catastrophes,
especially weather-related events, have been
increasing dramatically in frequency and intensity.
The 20 greatest natural catastrophes in terms of
monetary losses and the ten deadliest catastrophes
since 2000 are listed in Tables 1 and 2 respectively
(see below).
From hazard to risk – from event to catastrophe
It is important to understand the circumstances
under which natural catastrophes happen. Some
people even claim that there is no such a thing as a
natural catastrophe. Why? Because nature alone
does not produce catastrophes, it only produces
natural extreme events. We regard catastrophes
from the point of view of their impact on man, so an
extreme event like the one that caused the extinction of the dinosaurs is not a catastrophe in this
sense. For this reason, even a very strong earthquake in an uninhabited region without human
property cannot result in a catastrophe. Similarly, a
Sometimes, very expensive events – recent examples
being two hailstorms in Australia (Melbourne and
Perth, March 2010) – are hardly noticed outside the
country hit. But there are also less spectacular catastrophes such as the extreme heatwave in Europe in
2003 (70,000 deaths and over USD 10 billion dam-
* Munich Reinsurance Company.
3
CESifo Forum 2/2010
Focus
The overall losses in Chile are estimated as exceeding 30 billion US dollars (as at 7 June 2010). This
shows that well-prepared regions may still face high
repair and reconstruction bills, but the loss of life, the
number of people injured, and the interruption to
regular life are definitely less severe than at highly
vulnerable locations. To be fair, one must consider
the previous costs for precautionary measures as
well. It follows from this that better precaution does
not necessarily result in reduced overall costs, but
there is no indication either that they are higher than
if nothing is done. To sum up, precaution, even if
costly, pays off.
strong earthquake in a well-prepared region may
not be catastrophic. In a poorly prepared region,
however, even a moderate tremor may cause a devastating disaster.
A natural catastrophe happens if people and/or their
possessions are affected so severely that a society’s
life is disrupted. A well-prepared society is not likely to experience a catastrophe as easily as one where
many aspects of preparedness, from education and
knowledge to building codes, and from functioning
governance to availability of financial means are
missing, making it vulnerable to impacts from
nature. Catastrophes are hence not only products of
chance but also the outcome of interaction between
political, financial, social, technical and natural circumstances.
Reconstruction costs in Haiti will also be well above
10 billion US dollars. The difference to Chile is that
Haiti’s buildings and infrastructure will be raised to
a higher level of safety by (hopefully) proper planning and construction, i.e. it will have its vulnerability significantly reduced. This must be considered
when comparing the material losses in Haiti with
those in Chile; the term ‘costs’ rather than ‘losses’
would therefore be more appropriate.
The earthquakes in 2010 clearly support this statement. Haiti’s capital Port-au-Prince was razed by a
magnitude 7.0 quake at 13 km depth on 12 January.
The energy set free by the event was relatively moderate, but concentrated on a small area. Additionally, the disadvantageous underground conditions amplified the shaking. Presumably more than
220,000 people died as their dwellings were not the
least designed for earthquake forces (the US
Embassy was well designed and suffered practically
no damage, by the way). The catastrophe happened
because of the high concentration of humans at the
epicentre and the extreme vulnerability. The physical parameters of the earthquake in Baja California,
Mexico on 4 April were very similar, with a magnitude of 7.2 at 10 km depth. Two people died, the
damage is expected to be less than 1 billion US dollars. Here, the quake hit an almost unpopulated area
with only one larger settlement (Mexicali) at some
50 km distance. The lower population density and
values and better building standards made the difference to Haiti.
The above statements hold, in a similar way, for natural catastrophes caused by windstorm, flood, tsunami, etc. Whether a location is risky depends on
(a) the likelihood that a natural event may occur;
(b) the presence of people/items; and (c) their vulnerability (Kron 2005). Where there are no people or
values that can be affected by a natural phenomenon, there is no risk. Vulnerability can refer to
human health (human vulnerability), structural
integrity (physical vulnerability), or personal wealth
(financial vulnerability). Insurance’s contribution to
risk control addresses the last of these factors. All
three components have been and still are increasing
unabated. Rising sea levels, increased tropical
cyclone intensities (wind and rain), unprecedented
flood experiences, on the one hand, and megacities
with exploding populations and industrial development, on the other, are making many regions ever
riskier, in particular those on coasts. The overall risk
is determined by computing the integral over all possible threatening events and their consequences. The
thus quantified risk is identical to the expected average annual loss.
The magnitude 8.8 Chile earthquake on 27 February was about 500 times stronger than the
Haitian, and 250 times stronger than the Mexican.
While the depth of 35 km, the distance of 100 km
from the city of Concepción, and the vast area
which was exposed to the shaking certainly played
a role with respect to the impacts, the fact that less
than 300 people died (plus some 200 due to a tsunami) can clearly be attributed to the higher standard
of construction. Hence, the governing parameter
for this catastrophe was the high magnitude, i.e. the
hazard parameter.
CESifo Forum 2/2010
Natural catastrophe statistics and trends
Munich Re has been systematically collecting information on natural catastrophes for more than
4
Focus
Table 1
Overall and insured losses of the 20 costliest natural catastrophes since 2000
Rank
Date
Event
1
Aug. 2005
2
3
4
5
6
May 2008
Sep. 2008
Feb. 2010
Oct. 2004
Sept. 2004
Hurricane Katrina,
storm surge
Earthquake
Hurricane Ike
Earthquake
Earthquakes
Hurricane Ivan
7
Oct. 2005
Hurricane Wilma
8
9
10
11
Aug. 2002
Jan.–Feb. 2008
Aug. 2004
Sep. 2005
12
13
14
15
Jul. – Aug. 2003
Jul. 2007
Sep. 2004
Dec. 2004
Jan. 2007
Floods, severe storms
Winter damage
Hurricane Charley
Hurricane Rita,
storm surge
Heatwave, drought
Earthquake
Hurricane Frances
Tsunami
Winter storm Kyrill
Losses in billion US$
Affected country/regiona) (original values, not adjusted
for inflation)
Overall
Insured
USA*
125
62
China (Sichuan)
Caribbean, USA*
Chile*
Japan (Niigata)*
Caribbean, USA*
85
38
30
28
23
0.3
18.5
8
0.76
13.8
Caribbean, Mexico,
USA*
Europe
China
Caribbean, USA*
USA*
22
12.5
21.5
21.1
18
16
3.4
1.2
8
12.1
13.8
12.5
12
10
10
2
0.335
5.5
1
5.8
Europe
Japan (Niigata)*
Caribbean, USA*
Indian Ocean*
West and Central
Europe*
USA (Midwest)
Caribbean, USA*
Caribbean, USA*
Japan*
Jan. 2008
Floods
10
0.5
Sep. 2008
Hurricane Gustav
10
3.5
19
Sep. 2004
Hurricane Jeanne
9.2
5
20
Sep. 2004
Typhoon Songda
4.7
9
a)
Area near the coast is marked with *.
Notes: Losses from the Haiti earthquake are not included. The extent of incurred damage is not likely to
exceed 9 billion US dollars, although estimates of the reconstruction costs exceed 10 billion US dollars.
Source: Munich Re.
35 years. The firm’s NatCatSERVICE database, the
near a coast (15 of 20), and are mostly caused by
world’s largest with respect to losses from natural
weather events (15 of 20). Despite the fact that
catastrophes, contains more than 28,000 entries for
showing original values not adjusted for inflation
the period 1970 to 2009. US dollars are used as the
might give a biased picture, normalisation would not
lead currency in the database. This means that all
change this picture fundamentally, which shows:
losses are converted from the local currency/ies into
(a) natural catastrophes have never been so expen-
US dollars, applying the exchange rate at the time
sive; (b) losses in the two-digit billion dollar range
the event occurs.
have become more frequent; and (c) loss potentials
have reached new dimensions. This is not only true
The analyses conducted by Munich Re’s Geo Risks
for overall economic losses but also for the insur-
Research department deliver the most accurate esti-
ance industry’s share. 11 events in Table 1, i.e. in just
mates of the total economic and insured losses
the past ten years (all those with insured losses larg-
caused by any kind of natural peril. The results and
er than 4 billion US dollars), belong to the top 18
conclusions of these analyses are not only used for
all-time insured losses from a single event.
determining insurance premiums but also made
available to governments and non-governmental
In Table 2 (deadliest events in the past ten years), we
organisations to assist them in better planning and
also see that many events (five) are related to coasts,
developing prevention measures against natural
but – with the exception of Europe in 2003 – hit poor
catastrophes.
regions. If one considers the past 60 years instead of
only the past ten, the events ranked 9th and 10th
Table 1 reveals that catastrophes with high mone-
with less than 10,000 fatalities are no longer among
tary losses occur in well developed countries, often
the top ten; they fall back to ranks beyond 50. This
5
CESifo Forum 2/2010
Focus
Table 2
The 10 deadliest natural catastrophes since 2000
Rank Date
Event
1
Jan. 2010
Earthquake
2
Dec. 2004
Tsunami
3
May 2008
Cyclone Nargis, storm surge
4
Oct. 2005
Earthquake
5
May 2008
Earthquake
6
July – Aug. 2003 Heatwave, drought
7
Dec. 2003
Earthquake
8
Jan. 2001
Earthquake
9
May 2006
Earthquake
10
Nov. 2007
Cyclone Sidr, storm surge
a)
Area near the coast is marked with *.
Source: Munich Re.
reflects the fact that many catastrophes with very
high death tolls occurred in the second half of the
20th century, although the population density everywhere then was less than nowadays. This development testifies to the efficiency of modern technological achievements and efforts in disaster reduction,
initiated for instance by the International Decade
for Natural Disaster Reduction (IDNDR), which
was proclaimed by the United Nations in the 1990s.
However, the top six events in Table 2, all recorded
in the past seven years, would still rank among the
top 14 since 1950 and hence suggest that it is by no
means guaranteed that catastrophic death tolls are
becoming more and more limited.
Fatalities
222,500
220,000
140,000
88,000
84,000
70,000
26,200
14,970
5,749
3,360
take into account that these values are changing
over time. The least that can be done is to adjust for
inflation. Currently Munich Re, together with various scientific institutions such as the London
School of Economics (LSE), is intensely studying
the potential effects of other parameters that may
affect comparability and bias results, such as
changes in population and wealth (e.g. the development of the value of assets relative to GDP), as well
as implications of improved precautionary measures, etc.
An illustration comparing the parameters, number of loss events, fatalities, overall losses and
insured losses for GNCs caused by different hazard categories is presented in the pie charts of
Figure 1. They show that, for the past 60 years,
72 percent of loss events were weather-related
(meteorological, hydrological and climatological
events). Of this portion, tropical cyclones and
extratropical winter storms make up more than
half. If it comes to fatalities, earthquakes (including tsunamis) are responsible for roughly half of
all deaths, followed by meteorological causes
(including storm surges), which account for about
40 percent.
Tables 1 and 2 clearly reveal the distinct difference
between rich and poor countries. Large financial losses – in absolute terms – occur in the developed world,
which is also quite well insured. The actual impact on
a country must be measured in relative terms though.
While a one billion dollar loss in the United States or
in certain European countries is not outstanding, it
may cripple the economy of a small, poor country.
Munich Re therefore defines the so-called Great
Natural Catastrophes (GNCs) using relative criteria
(see note to Figure 1). This definition also helps when
it comes to trend analyses. Looking only at GNCs
means the reporting bias introduced by the development of communication technology in the past
60 years is largely eliminated. While GNCs are likely
to be documented in the records of the affected countries and their consequences can be assessed, smaller
events are not. In contrast to that, we quickly learn in
today’s internet age of almost any minor local loss
event, no matter where it happens.
The charts for monetary losses in the lower part of
Figure 1 reveal distinct differences between insured
and overall losses. Weather events are, with more
than 90 percent, by far the costliest ones for the
insurance industry, while the overall losses are
more evenly distributed among the three main categories.
We found that the annual average monetary losses from GNCs have increased dramatically over
the past 60 years (Figure 2). Even if a high fluctu-
For statistical analyses, in particular for trend analyses, one cannot use original loss values, but must
CESifo Forum 2/2010
Affected country/regiona)
Haiti*
Indian Ocean*
Myanmar*
Pakistan, India (Kashmir)
China (Sichuan)
Europe
Iran (Bam)
India (Gujarat)
Indonesia (Java)*
Bangladesh *
6
Focus
Figure 1
GREAT NATURAL CATASTROPHES 1950 TO 2009
PERCENTAGE DISTRIBUTION ORDERED BY TYPE OF EVENT
285 loss events
2 million fatalities
6%
7%
4%
28%
25%
53%
36%
41%
Overall losses USD 2000 bn (in 2009 values)
Insured losses USD 415 bn (in 2009 values)
7%
6%
4% 9%
32%
23%
81%
38%
Geophysical events
Meteorological events
Hydrological events
(Earthquake, tsunami, volcanic
eruption)
(Storm)
(Flood, mass movement)
Climatological events
(Extreme temperature, drought,
forest fire)
Note:
GreatMunich
natural catastrophes
are those where the affected region’s ability to help itself was overtaxed and interSource:
Re.
national assistance required because thousands were killed, hundreds of thousands made homeless, or substantial
economic or insured losses (relative to the regional situation) sustained.
Source: Munich Re.
ation is observed from year to year, the overall
trend is obvious. As weather-related catastrophes
predominate in terms of losses, this rise is mainly
attributable to windstorms and floods. The question is: what are the driving factors of this development?
People – land use – risk perception
At the moment, about 6.8 billion people require a
place to live on the earth, up from 3 billion in 1960
and heading for nine billion in the 2050s. Adequate
and safe areas are limited by natural circumstances,
but also by economic needs. While in the past settlement areas could be chosen because of their sheltered location with respect to adverse natural conditions, nowadays any available piece of land has to be
used. Furthermore, new residents are often unfamiliar with the local hazard situation or lulled into a
sense of security by trusting in the technical controllability of the forces induced by nature.
Reasons for increasing losses
There is no doubt that the main reasons for the
increase of catastrophe losses are global population
growth, the settlement and industrialisation of
regions with high exposure levels, and the fact that
modern technologies are highly susceptible to external disturbances. Due to its complexity, international
trade is even more susceptible. Changing environmental conditions, in particular climate change and
the lack of adequate risk perception, are additional
features.
Coastal areas, in particular, have been very attractive
to people. Today, already one tenth of the world’s
population lives within 5 km of the coast, one-third
within 50 km and two-thirds within 300 km, and 15 of
the 20 largest cities in the world are located on
7
CESifo Forum 2/2010
Focus
worldwide problem, there are
some striking examples such as
areas along the Yangtze River in
China that were devoted and
designed for flood retention in
the 1950s, but can no longer be
used for this very purpose as several hundred thousand people
live there today.
Figure 2
GREAT NATURAL CATASTROPHES 1950 TO 2009
OVERALL AND INSURED LOSSES WITH TREND
250
in 2009 values
USD billion
200
Overall losses
Insured losses
Trend overall losses
Trend insured losses
150
100
Complexity – wealth –
susceptibility
50
0
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
Source: Munich Re.
coasts. And this trend is still unbroken. According to
an OECD study (Nicholls et al. 2007), 113 million
people will live in the flood-prone neighbourhoods
of the 20 most populated coastal cities in 2070, an
almost five-fold increase in today’s number. The
same study predicts that assets in the 20 cities with
the highest concentration of flood-exposed values
will increase from the current 2.2 billion US dollars
to about 27 billion. In addition to permanent residents, millions of tourists choose coastal regions as
their holiday destination.
The state of Florida in the United States, which has
always had a high hurricane exposure, is a good illustration of how socio-economic factors act as natural
catastrophe loss drivers. The population has grown
from three million in 1950 to the current 19 million
(plus 86 million tourists every year). This means that
a hurricane making landfall in Florida today will
have a multiple of people and their belongings in its
path compared with the past.
2005
Wealth has increased in practically all regions of the
world. At developed locations, even in poor countries, buildings have expensive features such as glass
facades and sensitive claddings for architectural reasons – not just walls and windows. The potential for
destruction from shaking, wind pressure, hail and
flying debris is high, while more and more expensive
items can be found inside buildings as was the case in
the past. In general, modern equipment is highly vulnerable. Almost everything contains electric or electronic components, and these items often suffer
severely when exposed to vibrations, heat, sand and
dust, water or even humid, salty air. Whereas years
ago water-damaged items had simply been dried and
re-used, they are now discarded.
In the interior of countries, river plains are – if one
neglects the flood hazard – also well suited for development, and preferably used for this purpose. While
flood-control measures prevent frequent losses and
inconvenience, this effect is counterbalanced by the
feeling of security it creates, leading people to
expose more and more objects of increasing value to
the risk of flood. This sense of security is transmitted
not only by dykes and embankments, early-warning
systems, and the availability of disaster-relief organisations, but also by the intentional or unintentional
transmission of false information and by local
authorities or groups with a vested interest (e.g. the
tourist trade) playing down the risk. While this is a
CESifo Forum 2/2010
Several outages in recent years
in Europe, North America and
Asia have shown how badly we
depend on functioning power
supply and telecommunication
networks. Given a complex infrastructure (e.g. traffic and power networks), a failure in one place may
cause a domino effect that brings the whole system
to a standstill. It is scarcely conceivable what would
happen if a large-scale and long-term power breakdown were to occur that turned off the (electrical)
lights completely in Europe and/or North America
for weeks and even months. This could come about,
for instance, in a strong electro-magnetic event
(‘sun storm’) which destroys several large, systemrelevant transformers. Replacing/repairing these
may take months – as might the outage. Another
recent event, actually one with minor catastrophic
potential, has had a dramatic impact: the ash of
Eyjafjallajökull volcano that kept aircraft on the
ground led to chaotic situations at airports and in
hotels, interfered with business activities, and even
started to lead to shortfalls in supplies – and this
just after a few days.
2000
8
Focus
the last 100 years; the last nine years were among the
11 hottest in recorded history; 2010 has a high potential to become the warmest year ever. Climate
change is taking place and it is mainly caused by
human activities.
Urban concentration and environmental changes
It is not only the development of hazard-prone
regions, but additionally the concentration of commercial and industrial centres which attracts people. More than 50 percent of the world’s population
lives in urban areas compared to just 30 percent in
1950 – and the percentage is still increasing. Half a
century ago, there were eight megacities in the
world with a population exceeding five million.
Today the number of megacities has grown to over
60. It is obvious that the chance of a severe natural
event hitting one of these high-concentration
regions is steadily growing.
The fourth status report of the Intergovernmental
Panel on Climate Change (IPCC 2007) regards the
link between global warming and the greater frequency and intensity of extreme weather events as
probable. The report finds, for instance, with more
than 66 percent probability, that climate change
already produces more heatwaves, heavy precipitation, droughts and intense tropical storms. The
expected rise in global average temperatures of
between 1.6 and 6.4°C by the end of the century,
depending on future emissions of greenhouse gases,
significantly increases the probability of short-term
record temperatures (heatwaves). Warmer sea surface temperatures enhance evaporation and warmer
air can hold more water vapour, thus increasing precipitation potential. Combined with more pronounced convection processes, in which warm and
moist air rises to form clouds, this results in more frequent and more intense precipitation events.
Particularly over dense urban areas – i.e. areas with
high concentrations of values – the more intense
convection may lead to local severe weather events
that often involve a high density of lightning strokes,
hailstorms and gale-force gusts, sometimes even tornadoes. On account of the large proportion of impervious surfaces in urban areas, the torrential rain runs
straight into the drainage systems, which are not
designed to cope with such volumes, with the result
that underpasses, cellars and sometimes subway tunnels are flooded. A striking example is the 2005
Mumbai flooding: within 24 hours, 944 mm of rain
fell in the Indian industrial metropolis on 26 July,
that is more than 40 percent of the city’s average
annual rainfall (2170 mm). The flood losses added up
to 5 billion US dollars, of which 770 million US dollars were insured.
Large cities not only represent huge value accumulations, but are also very vulnerable to disturbances.
Many dwellings are being built in an uncontrolled
way, i.e. on unstable ground (e.g. the favelas in South
American metropolitan areas) or near bodies of
water, and they are certainly not erected according
to any construction code. Infrastructure (roads, electricity, water, sewage, etc.) is added later on (if at all)
as a kind of emergency measure rather than in a
planned and designed fashion. That makes these systems highly unreliable and vulnerable. On top of
that, little attention is paid to governance in poor
neighbourhoods. Early warning and evacuation in
the event of imminent severe threats can therefore
be extremely difficult.
Colonisation of land has always the consequence of
changing the environment. Forests are removed,
rivers have to be tamed or diverted, and the local
micro-climate is possibly altered. Most of these measures improve the situation in the short term but
deteriorate it in the long term. One fairly common
consequence of urban growth is subsidence, caused
for instance by groundwater extraction or sometimes
simply by the weight of buildings erected on relatively soft and unconsolidated coastal soil. A few
decimetres of subsidence can create huge problems
in already low-lying port cities and deltas, some areas
suddenly finding themselves below a given flood
level or even below the local sea level. The most
severe – and globally effective – change of the environment, however, concerns our climate.
The fact that the average number of great weather
catastrophes has tripled since 1950 is again evidence
of causal links between global warming and increasing frequencies and severity of natural events. Socioeconomic factors like the ones mentioned above
cannot explain the rise of catastrophes in total. It is
highly likely that climate change also is responsible
for a part of it and that the number of severe, weather-related natural catastrophes will further increase
in the long term as a result of continuing climate
Climate change
The scientific facts are clear: the global average temperature in the atmosphere has risen by 0.74°C over
9
CESifo Forum 2/2010
Focus
quake catastrophes and some – particularly for the
insurance industry – very costly weather events, it
has become likely that the year will be an outstanding one with respect to natural disasters. But
despite unfavourable loss trends, the insurance
industry continues to offer a wide range of natural
peril covers whilst trying, at the same time, to
encourage its clients to focus more on loss prevention. It is also making strenuous efforts to control
its own loss potentials with the help of modern
geoscientific methods. It is still difficult, however,
to predict in quantitative terms the effects that
future changes will have on the frequency and
intensity of extreme events.
change. This, combined with the trend towards higher value concentrations in exposed areas, will
increase loss potentials dramatically.
Even before publication of the Stern Review by
Lord Nicholas Stern (2006), it was clear that climate change is not just an ecological problem; it is
also an economic issue. Stern predicted that climate change could cost 5 to 20 percent of the
worldwide GDP annually until 2050. These costs
could be reduced to 1 percent by proper and timely actions. If damage costs continue to rise, this also
affects industry and primarily, of course, insurance
companies.
For Munich Re as a leading reinsurer, the natural
catastrophe trends of recent years have resulted in
three action strategies. Firstly, we call for effective
and binding rules on CO2 emissions in the international debate, so that climate change is curbed and
future generations do not have to live with weather
scenarios that are difficult to control. Secondly, with
our expertise we develop new business opportunities
in the context of climate protection and adaptation
measures. And thirdly, in our core business we only
accept risks at risk-adequate prices.
The role of the insurance industry
The insurance industry plays an important role in
raising awareness and coping with natural hazards:
it quantifies risk by means of adequate premiums
and thus makes risks transparent. Therefore, it creates incentives for reasonable behaviour and prevention, and so reduces the losses for the society.
The insurance industry also has tremendous potential for promoting climate protection and climate
change adaptation, and thus positively influencing
future losses, by taking account of such issues in its
products, investments, sponsoring activities and
communications.
In the same way as private individuals, insurance
companies try to avoid volatility in their payments.
Natural perils insurance is highly volatile. Large single losses (from one event) can be reduced by transferring part of the risk to the reinsurance sector, in
which companies often do business worldwide.
When catastrophic losses occur in one country, they
are distributed all over the world, thus relieving the
local insurance market and possibly even preventing
its collapse.
Driven by high losses from weather-related catastrophes, 2008 was the third most expensive year
on record, taking inflation into account. It has only
been exceeded by the hurricane year of 2005 and
by 1995, the year of the Kobe earthquake. Overall
economic losses totalled some 200 billion US dollars (2007: 82 billion US dollars), which is not too
far from the record set in 2005 (232 billion US dollars in current values). Worldwide insured losses in
2008 rose to 45 billion US dollars, about 40 percent
higher than the average in the previous ten years
(32 billion US dollars). The large number of tropical cyclones and the earthquake in Sichuan made
2008 also one of the deadliest years on record;
98 percent of the fatalities occurred in Asia
(Munich Re 2009).
The insurance industry’s natural catastrophe risk
models have already been adjusted in the light of
the latest findings. For instance, they now incorporate sea temperatures that remain above the longterm average due to the ongoing cyclical warm
phase in the North Atlantic; the effects of this warm
phase are enforced by global warming. We can
expect the above-average water temperatures to
increase further the intensities of cyclones. There
are two model types that are applied in the context
of insuring natural catastrophes: (a) those to determine the individual risk of a given object (or a portfolio of objects) to be insured, and (b) those to
assess expected losses, in particular for accumulation losses of extreme events.
The following year, 2009, was practically free from
large disasters and spectacular record losses on a
global scale, but the sum of all losses was only marginally below average (Munich Re 2010). 2010 has
again started differently. With the various earth-
CESifo Forum 2/2010
10
Focus
same as a probability distribution of losses. From
this curve, the expected loss for a given return period, or the return period of a historical event with a
known loss, can be read off. While such models for
windstorm and earthquake have been available for
many years, flood loss models have only become
operational for the past few years as they require
considerably more detail.
Hazard zonation and premium calculation
Premiums for the various hazards should reflect the
individual exposure. For the bulk of business, i.e. for
private homes and small businesses and their contents, the effort required to assess the exposure of a
particular building has to be seen in the context of
the annual premium income for one such property,
which may be in the order of less than 100 US dollars in low-risk areas. Since individual assessment of
the risk and calculation of an individual premium for
such properties are impossible, the premium has to
be fixed on the basis of a flat-rate assumption. For
this purpose, zones with a similar hazard level
(storm, flood, earthquake, landslide, etc.) have to be
identified and/or defined, premiums (per unit value)
being constant within a given zone. In most developed countries, hazard zone data of this kind are
available.
The partnership for risk reduction
Risk and loss minimisation call for an integrated
course of action. The risk must be borne by several
shoulders: the state, the people and enterprises
affected, and the financial sector, in particular the
insurance industry. Only when they all cooperate
with each other in a finely tuned relationship, in the
spirit of a risk partnership, can disaster prevention
really be effective.
Modelling probable expected losses
The job of public authorities (i.e. the state or the government) is primarily to reduce the underlying risk
to society as a whole. They provide access to observation and early-warning systems, build river dykes
and sea defences, determine the framework for the
use of exposed areas by enacting statutory provisions, and prepare emergency plans, including programmes to alleviate recovery (temporary housing,
financial assistance, tax relief, etc.). In some countries, insurance programmes are state-run. Unlike in
the case of earthquake and windstorm, where homeowners themselves are responsible for ensuring their
houses are properly protected, the responsibility for
protection from flooding is largely shifted to public
authorities.
Insurance and especially reinsurance companies
must be prepared to pay large amounts of money
after major events. One example: Munich Re faced
claims in the order of 2 billion US dollars after hurricane Katrina in 2005. The company is not threatened in its existence even by such enormous
amounts. However, volatility is expensive. Money for
payments must be made available very quickly and
cannot be placed in long-term – and thus profitable
– investments. With increasing single losses, the
whole financial market, including banks, loan institutions and investors, is becoming more and more
involved in covering risks.
Assessing probable maximum (accumulation) loss
(PML) and holding adequate reserves are crucial to
an insurer’s economic survival. PML models are
based on the definition in the second section of this
paper where the expected loss is a function of the
hazard, the values at risk and their vulnerability.
The values at risk are represented by the portfolio
under consideration, i.e. the distribution of
(insured) values in a country. Typical (average) loss
rates for different types of buildings and given
loads (wind speed, water level, ground acceleration) are applied and account for vulnerability. The
hazard is introduced by simulating a storm/earthquake/flood event in the area. By simulating a large
number of – stochastically generated – events and
arranging the resulting losses according to size, one
obtains an (empirical) PML curve, which is the
Those immediately affected (individuals, companies,
communities) have a great potential for loss reduction. The crucial point is whether they keep their risk
awareness alive. Even those people who do not
neglect the danger of a natural peril from the very
beginning often quickly forget about it, especially if
nothing happens for some time. They rely on technological protection systems and at the same time
make their property more and more valuable by
adding additional items that are often susceptible to
damage. These people must be informed and educated to build in an appropriate manner, control the
exposure of their values, and be ready to take action
in an emergency. This includes preparing for catastrophic losses by taking financial precautions, e.g.
buying insurance.
11
CESifo Forum 2/2010
Focus
surance Initiative (MCII). The idea of MCII is to collect money through global emissions trading and pay
it into a pool. In this way, the largest CO2 emitters
would ultimately finance the insurance solutions.
The MCII has already introduced this idea into the
negotiations for a successor to the Kyoto protocol,
which is absolutely vital. To restrict global warming
to the rise of 2°C that experts estimate to be just
about controllable, well-conceived measures are
needed quickly. This can only succeed if an international agreement with clear emission reductions and
supported by all the major CO2 emitters is adopted
as a successor to the Kyoto Protocol, which expires
in 2012.
The true task of insurance companies is to compensate financial losses that would have a substantial
impact on insureds or even bring about their ruin.
They carry the financial risk from events that have
such a low probability that they cannot be considered foreseeable. Insurance redistributes the burden
borne by individuals among the entire community of
insureds, which is ideally composed in such a way
that they all have a chance of being affected – even
if the degrees of probability differ. Furthermore,
they perform educational and public relations services, e.g. by publishing brochures in which they
draw attention to hazards and explain ways of dealing with them (e.g. Munich Re 2010).
Other measures being initiated in emerging countries include microinsurance. Munich Re recently
launched a pilot product offering low-income households in the Indonesian capital Jakarta the opportunity to insure against the direct economic losses and
social risks caused by severe flooding. The product –
the first microinsurance flood product worldwide – is
trigger based, depending on the height of the flood at
a specific, public river gauge.
What can and should be done?
Impacts from natural disasters are not as devastating
to rich societies as to those in less developed parts of
the world. There, whole countries are sometimes
thrown back in their development for years. Rich
states, on the other hand, have a significant financial
burden not only from catastrophe losses, but also
from the costly precautionary measures that citizens
demand from their governments to protect themselves and their properties.
Chances and opportunities
While we cannot stop or even reverse climate
change, there are many opportunities to mitigate it
and to adapt to it. New technologies and innovative
products will open up extraordinary economic
opportunities for companies, sectors and countries
leading the process.
Disaster prevention and risk reduction has several
levels. Starting from protecting the global climate
from becoming more and more threatening, they
range via forecast, warning, and technical control
systems to the individual’s (person, company) behaviour and provisions to make sure she/he/it will not be
ruined by an extreme event. While there is no discussion that loss of life must be prevented by all
means, the costs of efforts for prevention of monetary losses should not be completely out of balance
with the value of the protected items.
For whole countries, the mitigation of climate
change offers exceptional chances. New technologies are emerging and specific sectors can expand.
Especially economies with a high percentage of
technological industries have enormous potential
with regard to mitigating and adapting to climate
change.
Emerging countries
Climate change is a global problem and can therefore only be solved globally. The biggest challenge is
to quickly create and implement a global action
plan that includes both the largest emitters of greenhouse gases and the developing countries, which are
affected the most. Industrialised countries have to
take the lead as their emissions have mainly created
the problem. If we do not take ambitious action, the
effects of global warming may become unmanageable. The longer we wait, the harder it will be for
future generations to cope with climate change and
Effective and economical solutions have to be found
for emerging countries. These countries – especially
in Africa or Southeast Asia – are often faced with a
higher risk of natural disasters. At the same time, the
level of insurance density is generally very low, as
these markets are less industrialised and people have
less available income to insure themselves against
existential risks.
An attempt to provide assistance for climatechange-related risks is the Munich Climate In-
CESifo Forum 2/2010
12
Focus
result of the effects of extreme natural events and
the response to those events. Effective safeguards
are both achievable and indispensable, but they will
never provide complete protection. The determining factor is awareness that nature can always come
up with events against which no human means can
prevail.
the more expensive it will become. From the economic point of view, it definitely makes sense to act
now: investments in climate protection come much
cheaper than paying for the damage that would ultimately occur. And in the end, it is people in general
that have to bear the costs of natural catastrophes –
not least because a large portion of the damage is
always uninsured, as Munich Re’s statistics show
abundantly. Overall, the costs of not acting will certainly be vast.
References
Frisch, M. (1981), Der Mensch erscheint im Holozän, Frankfurt:
Suhrkamp.
However, it is important to keep the debate on an
objective level and to clarify some misunderstandings. The world will not immediately be destroyed
by climate change and we will not be overwhelmed
by a sudden endless series of destructive weather.
In connection with global warming, we are talking
of statistical changes in natural events. No single
event can serve as a proof for the change. However,
we still need to adapt to the consequences of climate change in order to lower the vulnerability, for
example through better building standards or flood
prevention measures, and avoid ever rising loss
potentials by uncontrolled development of hazardprone areas.
Intergovernmental Panel on Climate Change (IPCC, 2007), Climate
Change 2007: Fourth Assessment Report of the Intergovernmental
Panel on Climate Change,
http://www.ipcc-data.org/ddc_ar4pubs.html.
Kron, W. (2005), “Flood Risk = Hazard · Values · Vulnerability”,
Water International 30, 58–68.
Munich Re (2009), Topics Geo – Natural Catastrophes 2008,
Munich: Munich Reinsurance Company.
Munich Re (2010), Topics Geo – Natural Catastrophes 2009,
Munich: Munich Reinsurance Company.
Nicholls, R. J., S. Hanson, C. Herweijer, N. Patmore, S. Hallegatte, J.
Corfee-Morlot, J. Chateau and R. Muir-Wood, (2007), Ranking of
the World’s Cities Most Exposed to Coastal Flooding Today and in
the Future, OECD Environment Working Paper 1,
http://www.olis.oecd.org/olis/2007doc.nsf/linkto/env-wkp(2007)1.
Stern, N. (2006), Stern Review on the Economics of Climate Change,
HM Treasury, London, http://webarchive.nationalarchives.gov.uk/
+/http://www.hm-treasury.gov.uk/sternreview_index.htm.
Conclusion
The Swiss writer Max Frisch once stated (1981,
103): “only man knows natural disasters, so far as he
survives them. Nature does not know disasters”.
These have become more frequent and more
intense during the last decades. The main causes are
increasing global population and its need – and
sometimes wish – to settle in areas that are prone to
natural hazards, often in highly concentrated urban
agglomerations. At the same time, the amount and
susceptibility of possessions grow as risk awareness
fades. A further significant contribution is added by
climate change.
The latest IPCC report has clearly documented that
climate change is real and caused by human activities. It is today one of the greatest risks facing societies, and the rising number of severe weather-related natural catastrophes will cause higher loss burdens for economies in the future. At the same time,
it offers growth opportunities for innovative economies and businesses.
Great natural events are not avoidable. Great catastrophes are. Catastrophes are inevitably the net
13
CESifo Forum 2/2010
Focus
To clarify the situation, this article proposes a definition of the cost of a disaster, and emphasizes the
most important mechanisms that explain this cost. It
does so by first explaining why the direct economic
cost, i.e. the value of what has been damaged or
destroyed by the disaster, is not a good indicator of
disaster seriousness and why estimating indirect losses is crucial. Then, it describes the main indirect consequences of a disaster and of the following reconstruction phase, and discusses the methodologies to
measure them. Finally, it proposes a review of published assessments of indirect economic consequences, which confirm their importance and the
need to take them into account.
THE ECONOMICS OF
NATURAL DISASTERS
STÉPHANE HALLEGATTE* AND
VALENTIN PRZYLUSKI**
Introduction
Large-scale disasters regularly affect societies over
the globe, causing huge destructions and damages.
The 2010 earthquake in Port-au-Prince and the hurricane Katrina in 2005 have shown that poor as well
as rich countries are vulnerable to these events,
which have long lasting consequences on welfare,
and on human and economic development.
What is a disaster? How to define its cost?
For which purpose?
After each of these events, media, insurance companies and international institutions publish
numerous assessments of the ‘cost of the disaster’.
However these various assessments are based on
different methodologies and approaches, and they
often reach quite different results. Beside technical
problems, these discrepancies are due to the multidimensionality in disaster impacts and their large
redistributive effects, which make it unclear what is
included or not in disaster cost assessments. But
most importantly, the purpose of these assessments
is rarely specified, even though different purposes
correspond to different perimeter of analysis and
different definitions of the ‘cost’.
There is no single definition of a disaster. From an
economic perspective, however, a natural disaster
can be defined as a natural event that causes a perturbation to the functioning of the economic system, with a significant negative impact on assets,
production factors, output, employment, or consumption. Examples of such natural event are
earthquakes, storms, hurricanes, intense precipitations, droughts, heat waves, cold spells, and thunderstorms and lightning.
Disasters affect the economic system in multiple
ways and defining the cost of a disaster is tricky.
Pelling et al. (2002), Lindell and Prater (2003),
Cochrane (2004), Rose (2004), among others, discuss typologies of disaster impacts. These typologies usually distinguish between direct and indirect
losses.
This confusion translates into the multiplicity of
words to characterize the cost of a disaster in published assessments: direct losses, asset losses, indirect
losses, output losses, intangible losses, market and
non-market losses, welfare losses, or any combination
of those. It also makes it almost impossible to compare or aggregate published estimates that are based
on so many different assumptions and methods.
Direct losses are the immediate consequences of the
disaster physical phenomenon: the consequence of
high winds, of water inundation, or of ground shaking. Direct losses are often classified into direct market losses and direct non-market losses (also sometimes referred to as intangible losses, even though
non-market losses are not necessarily intangible).
Market losses are losses to goods and services that
* Centre International de Recherche sur l’Environnement et le
Développement, Paris and Ecole Nationale de la Météorologie,
Météo-France, Toulouse.
** Ecole Nationale de la Météorologie, Météo-France, Toulouse.
This research is supported by the European Community's Seventh
Framework Programme (FP7/2007-2013), ConHaz Project, Contract
No. 244159.
CESifo Forum 2/2010
14
Focus
lines are often possible. Moreover, in cases where
recovery and reconstruction does not lead to a
return to the baseline scenario, there are permanent
(positive or negative) disaster effects that are difficult to compare with a non-disaster scenario.
are traded on markets, and for which a price can easily be observed. Even though droughts or heat waves
affect directly the economic output (especially in the
agriculture sector), direct market losses from most
disasters (earthquakes, floods, etc.) are losses of
assets, i.e. damages to the built environment and
manufactured goods. These losses can be estimated
as the repairing or replacement cost of the destroyed
or damaged assets. Since building and manufactured
goods can be bought on existing markets, their price
is known. Direct market losses can thus be estimated
using observed prices and inventories of physical
losses that can be observed (as recorded, e.g. in the
EM-DAT database or insurance-industry databases)
or modelled (using, e.g. catastrophe models of the
insurance industry).
For instance, a disaster can lead to a permanent
extinction of vulnerable economic activities in a
region, because these activities are already threatened and cannot recover, or because they can move
to less risky locations. In that case, the disaster is not
a temporary event, but a permanent negative shock
for a region and it is more difficult to define the disaster cost. Also, reconstruction can be used to develop new economic sectors, with larger productivity,
and lead to a final situation that can be considered
more desirable than the baseline scenario. This
improvement can be seen as a benefit of the disaster.
It is however difficult to attribute unambiguously
this benefit to the disaster, because the same economic shift would have been possible in absence of
disaster, making it possible to get the benefits without suffering from the disaster-related human and
welfare losses.
Non-market direct losses include all damages that
cannot be repaired or replaced through purchases on
a market. For them, there is no easily observed price
that can be used to estimate losses. This is the case,
among others, for health impacts, loss of lives, natural asset damages and ecosystem losses, and damages
to historical and cultural assets. Sometimes, a price
for non-market impacts can be built using indirect
methods, but these estimates are rarely consensual
(e.g. the statistical value of human life always leads
to heated controversies).
Most importantly, defining the cost of a disaster cannot be done independently of the purpose of the
assessment. Different economic agents, indeed, are
interested in different types of cost. Insurers, for
instance, are mainly interested in consequences that
can be insured. Practically, this encompasses mainly
the cost of damages to insurable assets (e.g. damaged
houses and factories), and short-term business interruption caused by the disaster (e.g. the impossibility
to produce until electricity is restored).
Indirect losses (also labelled ‘higher-order losses’ in
Rose (2004)) include all losses that are not provoked by the disaster itself, but by its consequences.
Often, the term ‘indirect losses’ is used as a proxy
for ‘output losses’, i.e. the reduction in economic
production provoked by the disaster. Output losses
include the cost of business interruption caused by
disruptions of water or electricity supplies, and
longer-term consequences of infrastructure and capital damages. Indirect losses can also have ‘negativecost’ components, i.e. gains from additional activity
created by the reconstruction. Sometimes, non-monetary indirect consequences of disasters are also
included, like the impact on poverty or inequalities,
the reduction in collected taxes, or the increase in
national debt.
For affected households, insurable assets are also a
major component, but other cost categories are as
important. Primarily, loss of lives, health impacts and
perturbation to their daily life are crucial. But in
addition, households are concerned about their
assets but also about their income, which can be
reduced by business interruption or by loss of jobs,
and about their ability to consume, i.e. the availability of goods and services.
At the society scale, all these aspects are important,
but local authorities, governments and international
institutions are also interested in other points. First,
to manage the recovery and reconstruction period
and to scale the necessary amount of international
aid, they need information on the aggregated impact
on economic production, on unemployment and
Another difficulty in disaster cost assessment lies in
the definition of the baseline scenario. The cost of
the disaster has indeed to be calculated by comparing the actual trajectory (with disaster impacts) with
a counterfactual baseline trajectory (i.e. a scenario of
what would have occurred in absence of disasters).
This baseline is not easy to define and several base-
15
CESifo Forum 2/2010
Focus
so, Figure 1 (a), (b) and (c) show simplified representations of a post-disaster situation. Figure 1(a)
depicts the situation in which only output losses are
estimated: in this situation the disaster leads to a
temporary reduction in output during the reconstruction phase. We assume here that reconstruction
is a return to the baseline scenario (i.e. a no-disaster
counterfactual scenario). As already stated, this is
not always the case, but making an assumption on
the final state is necessary to define the cost of the
disaster, and the assumption of a return to the nondisaster baseline scenario is likely to be the most
neutral one for this type of assessment.
jobs, on the impact of inequality and poverty, on
local-businesses market-shares, on commercial balance, on collected taxes, etc. Second, to assess
whether investment in prevention measures are
desirable, they need the broadest possible assessments of the total disaster cost to the population, i.e.
an estimate of welfare losses.
Moreover, disaster impacts can have positive or negative ripple-effects at the global scale, as shown by
hurricane Katrina, which led to a significant rise in
world oil prices. Depending on the purpose and of
the decision-making spatial scale, the perimeter of
the cost analysis will be different. For instance, a
country may want to assess the losses in the affected
region, disregarding all out-of-the-region impacts, to
calibrate the financial support it wants to provide to
the victims. But it may also want to assess total losses on its territory, including gains and losses outside
the affected region, for example to assess the impact
on its public finance.
The sum of instantaneous output loss is what is often
referred to as the indirect loss. But reconstruction
needs in the disaster aftermath means that a significant share of the remaining production will have to
be devoted to reconstruction, as shown in
Figure 1(b). In other terms, the resources used to
rebuild damaged houses cannot be used to build new
houses, or to maintain existing ones. This reconstruction output is included in total output, and is not a
loss of output. But it is a ‘forced’ investment, in addition to the normal-time investment – consumption
trade-off. It causes, therefore, a loss of welfare. The
value of this forced investment is the replacement
value of damaged asset, which is referred to as the
direct losses. This is what is represented in
Figure 1(c): the sum of the output loss and of the
reconstruction output is the amount that cannot be
used for consumption and non-reconstruction
investment. This is referred to here as ‘total losses’.
Clearly, depending on the purpose of the assessment,
some of the cost components have to be included or
not in the analysis. In the following, we focus on the
economic cost for the affected region, with the aim
of informing decision-makers on post-disaster financial aid and prevention measure desirability. To do
so, it is obvious that the direct cost is an insufficient
measure, and that the loss of welfare is much more
relevant. Assessing a loss of welfare is complicated,
as it includes many economic and non-economic
components. Here, we focus on the economic component of welfare losses, and we define the economic cost of disaster as the lost consumption, considered as an important component and a good proxy
of economic-related welfare losses.1 Of course, this
article does not try to be comprehensive, and major
cost components are left out of the analysis, like loss
of lives, health consequences and loss of jobs. These
additional component are important for the population welfare and therefore for prevention measure
assessments.
In this framework, total costs are the sum of the indirect cost (i.e. the reduction of the total value added
by the economy due to the disaster) and the direct
cost (i.e. the portion of the remaining value added
that has to be dedicated to reconstruction instead of
normal consumption). Capital and output losses can
therefore be simply added to estimate consumption
losses.
Of course, Figure 1 shows a simplified situation in
which production has no flexibility. In this case,
reconstruction needs cannot be satisfied through
increased production and it has to crowd out other
consumptions and investments. Figure 2 depicts a
different case, in which there is a limited flexibility in
the production process: capital destruction leads to a
reduction in output; but unaffected capital can
increase its own production to compensate this
reduction, for instance through an increase in work
Consumption losses and output losses
This section explains how to assess consumption
losses from asset and output losses. More precisely, it
explains why the sum of asset losses and output losses is a good proxy for the loss of consumption. To do
1 In an utilitarist framework, what matters is not output and production, but consumption.
CESifo Forum 2/2010
16
Focus
Figure 1
DIRECT LOSSES, INDIRECT LOSSES AND TOTAL LOSSES WHEN THERE IS
NO FLEXIBILITY IN THE PRODUCTION PROCESS
a)
hours by workers at unaffected
factories and businesses. In practice, there are gross indirect losses, and gross indirect gains (due
to the stimulus effect of the
reconstruction). But there is still
a fraction of the remaining production that is used for reconstruction instead of normal consumption, even though this share
is smaller than in absence of production flexibility.
b)
In this situation also, the consumption loss is still the sum of
direct (asset) and indirect (output) losses (see Figure 2(c)),
making it necessary to estimate
output losses. But output losses
are not only the lost production
from the affected capital, but
also the output gains and losses
from unaffected capital in the
rest of the economy. It makes the
assessment of output losses
more complicated, since it depends on complex economic
mechanisms and trade-offs.
c)
In practice, moreover, the reduction in consumption can be mitigated or amplified by (i) changes
in prices; (ii) flexibility in the
production process; (iii) changes
in the saving-consumption tradeoff for the remaining production;
and (iv) the fact that the rebuilt
capital will be more recent than
before the disaster, with potential benefits (Hallegatte and
Dumas 2009). The following section will describe methodologies
to assess output losses and highlight the most important processes to take into account.
How to assess output losses?
If output losses represent an
important component of total
losses, it becomes essential to
develop methodologies to assess
them. To do so, we propose to
Source: Authors’ conception.
17
CESifo Forum 2/2010
Focus
Figure 2
DIRECT LOSSES, INDIRECT LOSSES AND TOTAL LOSSES WHEN THERE IS
A LIMITED FLEXIBILITY IN THE PRODUCTION PROCESS
a)
start by assessing the lost output
from the directly affected capital. In the second subsection, we
investigate the systemic impacts
of disasters, including the effect
on the capital that is not directly
affected by the disaster.
From asset losses to output losses
The first step in an assessment
of output losses is to estimate
how much output is lost because
of direct asset losses. Economic
theory states that, at the economic equilibrium and under
certain conditions, the value of
an asset is its expected future
production, and this equality has
been widely used to assess disaster output losses. Assuming this
equality is always verified, the
output loss caused by capital
loss is simply equal to the value
of the damages, capital losses
and output losses are simply
equal, and the sum of asset and
output losses is the double of
asset losses.
b)
The assumption that output losses are equal to capital losses is
however based on strong assumptions, which are not always
verified. In estimates of disaster
consequences, ‘asset loss’ is the
replacement value of the capital.
To have the equality of asset loss
and output loss, a double equality needs to be verified: replacement value has to be equal
to market value; and market
value has to be equal to the
net present value of expected
output. In an optimal economy at
equilibrium, these two equalities
are valid: first, the market value
of an asset is by definition the
net value of its output; second,
if market value were higher
(lower) than replacement value,
then investors would increase
(decrease) the amount of capital
to restore the equilibrium.
c)
Source: Authors’ conception.
CESifo Forum 2/2010
18
Focus
Another example is the health care system in New
Orleans. Beyond the immediate economic value of
the service it provided, a functioning health care is
necessary for a region to attract workers. After
Katrina landfall on the city, the absence of health
care services made it more difficult to reconstruct,
and the cost for the region was much larger than the
economic direct value of this service.
Therefore, in theory, there is no difference between
capital losses and the reduction in output from this
capital. But the assumption of the economics being
optimal and at equilibrium is questionable. First, for
the replacement value and the market value to be
equal, the economy needs to be at its optimum, i.e.
the amount of capital is such that its return is equal
to the (unique) interest rate. This is unlikely for the
capital that is affected by natural disasters, especially as infrastructure and public assets are heavily
affected. Since these assets are not exchanged on
markets, they have no market prices. Moreover, they
are not financed by private investors, but decided
about through a political process under the consideration of multiple criteria (e.g. land-use planning
objectives), and there is no reason for their purelyfinancial return to be equal to the (private) interest
rate. Practically, some assets may have an output
value lower than their replacement value (e.g. a secondary road that is redundant and does not provide
a significant gain of time or distance), while some
may have an output value much larger than their
replacement value (e.g. a bridge that cannot be
closed without large consequences for users).
The systemic impact of natural disasters
The equality between output losses and asset losses
is questionable for any economic shock, small or
large. The most important issues appear when considering very large shocks, or systemic events, which
are the events that perturb the functioning of the
entire economic system and affect relative prices. In
this case, output losses may be damped or amplified
by several mechanisms.
(a) Changes in prices
Figures 1 and 2 show output in real terms, i.e. with
no monetary effects. But output losses can be estimated assuming unchanged (pre-disaster) prices or
con-sidering the impact of the disaster on prices.
Both assumptions lead to the same result if the disaster has only a marginal impact on the economy,
with little impact of prices, but can be very different
in the opposite situation. In other terms, one can
assume that if a house is destroyed, the family who
owns the house will just have to rent another house
at the pre-disaster price. But this assumption is
unrealistic if the disaster causes more than a marginal shock. In post-disaster situations, indeed, a significant fraction of houses may be destroyed, leading to changes in the relative price structure. In this
case, the price of alternative housing can be much
higher than the pre-di-saster price, as a consequence
of the disaster-related scarcity in the housing market. Estimating the value of lost housing service
should then be done using this higher cost instead of
the pre-disaster one, which can lead to a significant
increase in the assessed disaster cost. Unfortunately,
it is difficult to predict ex ante the change in prices
that would be caused by a disaster, making loss
assessment more complicated.
Second, for market values to be equal to net present
value of expected output, expectations have to be
unbiased and markets need to be perfect. This is not
always the case especially in sectors affected by disasters, where expectations can be heavily biased (e.g.
in housing market).
Also, output losses are most of the time estimated
from a social point of view. The equality between
market value (for the owner) and expected output
(for the society) is valid only in absence of externalities. Some assets that are destroyed by disasters may
exhibit positive externality. It means that their value
to the society is larger than the value of the owner’s
expected output. Public goods have this characteristic, among which most infrastructures.
An example is provided by the San Francisco
Oakland Bay Bridge, which is essential to the economic activity in San Francisco and had to be closed
for one month after the Loma Prieta earthquake in
1989. Its replacement value has no reason to be
equal to the loss in activity caused by the bridge closure, because the bridge production is not sold on a
market, the bridge has no market value, and the
social return on capital of the bridge is unlikely to
exhibit decreasing returns and is likely to be much
higher than the interest rate.
The same reasoning is possible in all other sectors,
including transportation, energy, water, health, etc.
In extreme cases, reconstruction may even be impossible, at all prices. This is because markets are not at
equilibrium in disaster aftermath (i.e. price is not
19
CESifo Forum 2/2010
Focus
households are able to pay for reconstruction, but
cannot find workers and contractors to carry out the
work. The same is true for businesses and factories.
This explains why reconstruction often takes several
years, even for limited damages (e.g. the 2004 hurricane season in Florida – see McCarty and Smith
2005). Examples of constraint include the availability of equipment and qualified workers. For instance,
the limited availability of glaziers increased the cost
of reconstruction and lowered the reconstruction
pace after the 2001 chemical explosion in Toulouse
(France), despite glaziers coming from all the country to carry out the work.
such that demand equals production). The ‘if I can
pay it, I can get it’ assumption is not valid in post-disaster situations. In these situations, therefore, the
value of lost production cannot always be estimated
as the product of lost produced quantity and pre-disaster prices. Providing an unbiased estimate requires
an assessment of the disaster impact on prices.
Often considered as resulting from unethical behavior from businesses, which are thought to benefit
from the disaster, post-disaster price inflation can
also have positive consequences. This inflation,
indeed, helps attract qualified workers where they
are most needed and creates an incentive for all
workers to work longer hours, therefore compensating for damaged assets and accelerating reconstruction. It is likely, for instance, that higher prices after
hurricane landfalls are useful to make roofers from
neighboring unaffected regions move to the landfall
region, therefore increasing the local production
capacity and reducing the reconstruction duration.
Demand surge, as a consequence, may also reduce
the total economic cost of a disaster, even though it
increases its burden on house owners.
(c) Output gains and losses from the non-affected
capital
Damages in crucial intermediate sectors may lead to
negative ‘network effects’ in the economy, leading to
production losses even for businesses that are not
directly affected by the disaster. Water, electricity,
gas and transportation are the most critical sectors,
and their production interruption has immediate
consequences on the entire economic system. In past
cases, it has been shown that the indirect consequences of utility services had larger consequences
than direct asset losses, both on households
(McCarty and Smith 2005) and on business (Tierney
1997). Of course, when capital cannot produce
because of a lack of input (e.g. electricity, water),
input substitution, production rescheduling, and
longer work hours can compensate for a significant
fraction of the losses (see Rose et al. 2007). These
mechanisms can damp the output losses, and can
especially reduce the crowding-out effects of reconstruction on normal consumption and investment
(see Figure 2). But their ability to do so is limited,
especially when losses are large.
(b) Length of the reconstruction phase
Importantly, there is a large difference between losing a home for one day (in this case the total loss is
the reconstruction value, i.e. the direct loss) and losing a home for one year (in this case the total loss is
the reconstruction value, i.e. the direct loss, plus the
value of one year of housing services, i.e. the output
loss). Of course, the longer the reconstruction period, the larger the total cost of the disaster.
The reconstruction phase, and the economic recovery pace, will ultimately determine the final cost of
the natural disasters. The reconstruction pace is
linked to the constraints to the reconstruction
phase, which are of two types. First, they can be
financial. This concerns situations in which households and businesses can simply not finance the
reconstruction. This is of particular importance in
countries with limited resources (Freeman et al.
2002; Mechler et al. 2006).
There are many sources of flexibility in the economic system. First, production capacity is not fully used
in normal times and idle production capacity can be
mobilized in disaster aftermath to compensate for
lost production from lost assets. Second, behaviors
can change in disaster aftermath and workers can
increase their work hours in unaffected businesses to
help society cope with disaster consequences (and
sometimes benefit from increased prices). As a consequence, unaffected capital can often increase production to compensate for output loss from affected
capital. After mild disasters, net output gains can
even be observed, explained by the non-zero price
elasticity of production, and by the under-optimality
Constraints are also technical. Technical limits to the
ability to increase production are obvious in the construction sector, which experience a dramatic increase in demand after the disaster. In spite of this
demand, production does not follow, because there
are strong constraints on reconstruction. Many
CESifo Forum 2/2010
20
Focus
substitution can take place in the production system. On the opposite CGE models are considered
as too optimistic, since they assume that markets
function perfectly (even in post-disaster situations), and that optimal prices balance production
and demand and, act as signals to incentivize production of the most needed goods and services. The
reality probably lies somewhere in between these
two extremes, prompting the work on intermediate
models. These intermediate models are either IO
models with flexibility like those in Hallegatte
(2008) or CGE models with reduced substitution
elasticity like those in Rose et al. (2007).
of the pre-disaster situation that leaves some room
for increased production. In an economy that fully
uses all resources and cannot increase its production
over the short-term (whatever the price level), such
a gain would be impossible. In a more realistic economy that does not use efficiently all resources (with
underemployment, and imperfect allocation of capital), additional demand does not lead only to inflation, but also to increased output.
The ‘adaptability’ and ‘flexibility’ of the production
system and its ability to compensate for unavailable
inputs is largely unknown and largely depend on the
considered timescale. Over the very short term, the
production system is largely fixed and the lack of
one input can make it impossible to produce.
Moreover, over short timescales, local production
capacity is likely to be highly constrained by existing
capacities, equipments and infrastructure. Only
imports from outside the affected region and postponement of some non-urgent tasks (e.g. maintenance) can create a limited flexibility over the shortterm. This is what is represented in economic InputOutput model (e.g. Okuyama 2004), in which producing one unit of output requires a fixed amount of
all input categories.
(d) The stimulus effect of disasters
Disasters lead to a reduction of production capacity, but also to an increase in the demand for the
reconstruction sector and goods. Thus, the reconstruction acts in theory as a stimulus. However, as
any stimulus, its consequences depend on the preexisting economic situation, or the phase of the
business cycles. If the economy is in a phase of high
growth, in which all resources are fully used, the
net effect of a stimulus on the economy will be negative, for instance through diverted resources, production capacity scarcity and accelerated inflation.
If the pre-disaster economy is depressed, on the
other hand, the stimulus effect can yield benefits to
the economy by mobilizing idle capacities. This
complex interplay between business cycles and
natural disasters economics is analyzed in detail in
Hallegatte and Ghil (2008), who support the
counter-intuitive result that economies in recession
are more resilient to the effect of natural disasters.
This result appears to be consistent with empirical
evidence. For instance, the 1999 earthquake in
Turkey caused destructions amounting to 1.5 to
3 percent of Turkey’s GDP, but consequences on
growth remained limited, probably because the
economy had significant unused resources at that
time (the Turkish GDP contracted by 7 percent in
the year preceding the earthquake). In this case,
therefore, the earthquake may have acted as a
stimulus, and have increased economic activities in
spite of its terrible human consequences. In 1992
also, when hurricane Andrew made landfall on
southern Florida, the economy was depressed and
only 50 percent of the construction workers were
employed (West and Lenze 1994). The reconstruction needs had stimulus effects on the construction
sectors, which would have been impossible in a better economic situation.
Over the longer term and the entire reconstruction
period which can stretch over years for large-scale
events, the flexibility is much higher: relative prices
change, incentivizing production in scarce sectors;
equipments and qualified workers move into the
affected region, accelerating reconstruction and
replacing lost capacities; and different technologies
and production strategies can be implemented to
cope with long-lasting scarcities. The production system organization can also be adjusted to the new situation: one supplier that cannot produce or cannot
deliver its production (because of transportation
issues, for instance) can be replaced by another suppliers; new clients can be found to replace bankrupt
ones; slightly different processes can be introduced
to reduce the need for scarce inputs (e.g. oil-running
backup generator can be installed if electricity availability remains problematic). These types of substitution are represented in Calculable General
Equilibrium models (e.g. Rose et al. 2007), in which
the scarcity of one input translates into higher price,
and reduced consumption of this input, compensated
by larger consumption of other inputs.
IO models are often considered too pessimistic,
since they assume that prices are fixed and that no
21
CESifo Forum 2/2010
Focus
replace the destroyed capital by the same capital, in
order to restore production as quickly as possible,
even at the price of a lower productivity. In extreme
case, one may even imagine that reconstruction is
carried out with lower productivity, to make reconstruction as fast as possible, with a negative impact
on total productivity. Second, even when destructions are quite extensive, they are never complete.
Some part of the capital can, in most cases, still be
used, or repaired at lower costs than replacement
cost. In such a situation, it is possible to save a part
of the capital if, and only if, the production system is
reconstructed identical to what it was before the disaster. This technological ‘inheritance’ acts as a major
constraint to prevent a reconstruction based on the
most recent technologies and needs, especially in the
infrastructure sector.
(e) The productivity effect
When a disaster occurs, it has been suggested that
destructions can foster a more rapid turn-over of capital, which could yield positive outcomes through the
more rapid embodiment of new technologies. This
effect, hereafter referred to as the ‘productivity
effect’, has been mentioned, for instance, by AlbalaBertrand (1993), Stewart and Fitzgerald (2001),
Okuyama (2004), and Benson and Clay (2004).
Indeed, when a natural disaster damages productive
capital (e.g. production plants, houses, bridges), the
destroyed capital can be replaced using the most
recent technologies, which have higher productivities.
Examples of such upgrading of capital are: (i) for
households, the reconstruction of houses with better
insulation technologies and better heating systems,
allowing for energy conservation and savings; (ii) for
companies, the replacement of old production technologies by new ones, like the replacement of paperbased management files by computer-based systems;
and (iii) for government and public agencies, the
adaptation of public infrastructure to new needs, like
the reconstruction of larger or smaller schools when
demographic evolutions justify it. Capital losses can,
therefore, be compensated by a higher productivity
of the economy in the event aftermath, with associated welfare benefits that could compensate for the disaster direct consequences. This process, if present,
could increase the pace of technical change and
accelerate economic growth, and could therefore represent a positive consequence of disasters.
This effect is investigated in Hallegatte and Dumas
(2008) using a model with embodied technical
change. In this framework, disasters are found to
influence the production level but cannot influence
the economic growth rate, in the same way as the
saving ratio in a Solow-like model. Depending on
how reconstruction is carried out (with more or less
improvement in technologies and capital), indeed,
accounting for the productivity effect can either
decrease or increase disaster costs, but is never able
to turn disasters into positive events.
(f) Poverty traps
It is crucial to also take into account the possibility
that natural disasters increase poverty. In particular,
because they destroy assets and wipe out savings, they
can throw households into ‘poverty traps’, i.e. situation in which their productivity is reduced, making it
impossible for them to rebuild their savings and
assets.These micro-level poverty traps can also be created by health and social impacts of natural disasters:
it has been shown that disasters can have long-lasting
consequences on psychological health (Norris 2005),
and on children development (from reducing in
schooling and diminished cognitive abilities – see, for
instance, Santos (2007); Alderman et al. (2006)).
As an empirical support for this idea, AlbalaBertrand (1993) examined the consequences of 28
natural disasters on 26 countries between 1960 and
1979 and found that, in most cases, GDP growth
increases after a disaster and he attributed this
observation, at least partly, to the replacement of the
destroyed capital by more efficient one.
However, the productivity effect is probably not
fully effective, for several reasons. First, when a disaster occurs, producers have to restore their production as soon as possible. This is especially true for
small businesses, which cannot afford long production interruptions (see Kroll et al. 1991; Tierney
1997), and in poor countries where people have no
mean of subsistence while production is interrupted.
Replacing the destroyed capital by the most recent
type of capital implies in most cases to adapt company organization and worker training which takes
time. Producers have thus a strong incentive to
CESifo Forum 2/2010
These poverty traps at the micro-level (i.e. the
household level) could even lead to macro-level
poverty traps, in which entire regions could be stuck.
Such poverty traps could be explained by the amplifying effect reproduced in Figure 3. Poor regions
have a limited capacity to rebuild after disasters: if
they are regularly affected by disasters, they do not
22
Focus
output model that indirect ecoAMPLIFYING EFFECT ILLUSTRATING HOW NATURAL DISASTERS COULD
nomic losses in Louisiana after
Katrina amounted to 42 billion
BECOME RESPONSIBLE FOR MACRO-LEVEL POVERTY TRAPS
US dollars compared to 107 billion US dollars of direct ecoLimited reconstruction
nomic losses. More generally,
capacity
this analysis concludes that
regional indirect losses increase
nonlinearly with direct losses,
Reduced economic
Long reconstruction
Amplifying
development
period after disasters
suggesting the existence of
effect
threshold in the coping capacity
of economic systems. In this
analysis of Louisiana, indirect
Reduced accumulation of
Large economic cost of
losses remain negligible (or even
capital and infrastructure
natural disasters
negative) for direct losses below
50 billion US dollars, and then
Source: Authors’ conception.
increase nonlinearly to reach 200
billion US dollars for direct losshave enough time to rebuild between two events,
es of the same amount (see Figure 4). Also, indirect
and they end up into a state of permanent reconlosses decrease rapidly if it is possible to ‘import’
struction, with all resources devoted to repairs
reconstruction means (workers, equipment, finance)
instead of addition of new infrastructure and equipfrom outside the affected region. This result highments. These obstacles to capital accumulation and
lights the importance of considering the interregioninfrastructure development lead to a permanent disal interactions.
aster-related under-development. This effect has
been analyzed by Hallegatte et al. (2007) with a
reduced-form model showing that the average GDP
Conclusions
impact of natural disasters can be either close to zero
if reconstruction capacity is large enough, or very
This article highlights the main difficulties in definlarge if reconstruction capacity is too limited (which
ing, measuring and predicting the total cost of disasmay be the case in less developed countries).
ters. It focuses on indirect (or output) losses, considered as a major component of the total loss of popuThis type of feedback can be amplified by other
lation welfare. There are several methodologies to
long-term mechanisms, like changes in risk percepassess these indirect losses, but they are all based on
tion that reduces investments in the affected regions
questionable assumptions and modelling choices,
or reduced services that make
qualified workers leave the
regions. Because of these mechFigure 4
anisms, the consequences of a
INDIRECT (OUTPUT) LOSSES AS A FUNCTION OF DIRECT (ASSET) LOSSES
disaster can last much longer
IN LOUISIANA FOR KATRINA-LIKE DISASTERS
than what is normally considFigure 3
ered to be the recovery and reconstruction period.
Indirect losses
in billion US dollars
300
250
An example of assessment on
Katrina in New Orleans
For the landfall of Katrina on
New Orleans, the availability of
a large amount of data allowed
many
modelling
analyses.
Hallegatte (2008), for instance,
estimated using a regional input-
200
150
100
50
0
- 50
0
50
100
150
200
250
300
Direct losses
Source: Hallegatte (2008).
23
CESifo Forum 2/2010
Focus
McCarty, C and S. K. Smith (2005), Florida’s 2004 Hurricane Season:
Local Effects, Florida Focus, University of Florida,
http://www.bebr.ufl.edu/system/files/FloridaFocus1_3_2005_0.pdf.
and they can lead to very different results. The main
conclusion is of this article twofold.
Mechler, R., J. Linnerooth-Bayer and D. Peppiatt (2006),
Microinsurance for Natural Disasters in Developing Countries:
Benefits, Limitations and Viability, ProVention Consortium,
Geneva,
http://www.proventionconsortium.org/themes/default/pdfs/Microin
surance_study_July06.pdf.
First, it is impossible to define ‘the cost’ of a disaster,
as the relevant cost depends largely on the purpose
of the assessment. The best definition and method
obviously depend on whether the assessment is supposed to inform insurers, prevention measure costbenefit analyses, or international aid providers. A
first lesson from this article is that any disaster cost
assessment should start by stating clearly the purpose of the assessment and the cost definition that is
used. Following this recommendation would avoid
misleading use of assessments, and improper comparison and aggregation of results.
Norris, F. H. (2005), Range, Magnitude, and Duration of the Effects
of Disasters on Mental Health: Review Update 2005, Dartmouth
Medical School and National Center for PTSD, Hanover and
Boston.
Okuyama, Y. (2004), “Modeling Spatial Economic Impacts of an
Earthquake: Input-Output Approaches”, Disaster Prevention and
Management 13, 297–306.
Pelling, M., A. Özerdem and S. Barakat (2002), “The Macro-economic Impact of Disasters”, Progress in Development Studies 2,
283–305.
Rose, A. (2004), “Economic Principles, Issues, and Research
Priorities in Hazard Loss Estimation”, in: Okuyama, Y. and S. Chang
(eds.), Modeling Spatial and Economic Impacts of Disasters, Berlin:
Springer, 14–36.
Second, there are large uncertainties on indirect disaster costs, and these uncertainties arise both from
insufficient data and inadequate methodologies.
Considering the importance of unbiased estimates of
disaster cost, for instance to assess the desirability of
prevention measures, progress in this domain would
be welcome and useful. To do so, much more
research should be devoted to this underworked
problem.
Rose, A., G. Oladosu and S. Y. Liao (2007), “Business Interruption
Impacts of a Terrorist Attack on the Electric Power System of
Los Angeles: Customer Resilience to a Total Blackout”, Risk
Analysis 27, 513–531.
Santos, I. (2007) Disentangling the Effects of Natural Disasters on
Children: 2001 Earthquakes in El Salvador, Doctoral Dissertation,
Kennedy School of Government, Harvard University.
Tierney, K., (1997),“Business Impacts of the Northridge Earthquake”,
Journal of Continencies and Crisis Management 5, 87–97.
West, C. T. and D. G. Lenze (1994), “Modeling the Regional Impact
of Natural Disasters and Recovery: A General Framework and an
Application to Hurricane Andrew”, International Regional Science
Review 17, 121–150.
References
Albala-Bertrand, J. M. (1993), The Political Economy of Large
Natural Disasters with Special Reference to Developing Countries,
Oxford: Clarendon Press.
Alderman, H., J. Hodditnott and B. Kinsey (2006), “Long-term
Consequences of Early Childhood Malnutrition”, Oxford Economic
Papers 58, 450–474.
Cochrane, H. (2004), “Economic Loss: Myth and Measurement”,
Disaster Prevention and Management 13, 290–296.
Freeman, P. K., L. A. Martin, J. Linnerooth-Bayer, R. Mechler, S.
Saldana, K. Warner and G. Pflug (2002), Financing Reconstruction:
Phase II Background Study for the Inter-American Development
Bank Regional Policy Dialogue on National Systems for
Comprehensive Disaster Management, Washington DC: InterAmerican Development Bank.
Lindell, M. K. and C. S. Prater (2003), “Assessing Community
Impacts of Natural Disasters”, Natural Hazards Review 4, 176–185.
Hallegatte, S. (2008), “An Adaptive Regional Input-Output Model
and Its Application to the Assessment of the Economic Cost of
Katrina”, Risk Analysis 28, 779–799.
Hallegatte S., J. C. Hourcade and P. Dumas (2007), Why Economic
Dynamics Matter in Assessing Climate Change Damages: Illustration
on Extreme Events”, Ecological Economics 62, 330–340.
Hallegatte, S. and P. Dumas (2008), “Can Natural Disasters Have
Positive Consequences? Investigating the Role of Embodied
Technical Change”, Ecological Economics 68, 777–786.
Hallegatte, S. and M. Ghil (2008), “Natural Disasters Impacting a
Macroeconomic Model with Endogenous Dynamics”, Ecological
Economics 68, 582–592.
Kroll, C. A., J. D. Landis, Q. Shen and S. Stryker (1991), Economic
Impacts of the Loma Prieta Earthquake: A Focus on Small Business,
Studies on the Loma Prieta Earthquake, University of California,
Transportation Center.
CESifo Forum 2/2010
24
Focus
A recent pertinent example is the devastation that
recent earthquakes wrought in Haiti and Chile. The
January 2010 earthquake that struck Haiti’s densely
populated capital, Port-au-Prince, caused significant
loss of human life (between 200,000 and 250,000 fatalities), the displacement of more than a million and
severe damage to the country’s economic infrastructure (estimated over 100 percent of the country’s
GDP) – see Cavallo, Powell and Becerra (2010). In
contrast, the February 2010 earthquake in Chile
which was physically stronger and also struck a
densely populated area caused many fewer fatalities
(less than 500 people killed according to most recent
official estimates). And although direct economic
damages are expected to be substantial due to the
amount of wealth exposed, they are expected to be
far less than Haiti’s in relation to the size of the
economy.1 Clearly, these dissimilar outcomes originated from different policies, institutional arrangements and economic conditions.
THE AFTERMATH OF
NATURAL DISASTERS:
BEYOND DESTRUCTION
EDUARDO CAVALLO* AND
ILAN NOY**
Introduction
Recent catastrophic natural disasters, such as the
Indian Ocean tsunami of 2004, and the Haitian
earthquake of 2010, have received more international attention than previous disasters, yet our
rapidly evolving understanding regarding their
relevance to economic development and growth is
still in its infancy. Much research in the social sciences, and even more in the natural sciences, has
been devoted to increasing our ability to predict
disasters and prepare for them. Interestingly,
however, the economic research on natural disasters and their consequences is fairly limited. We
summarize here the state of this literature and
point to questions that we believe need further
probing.
Pelling et al. (2002) and ECLAC (2003) introduce a
typology of disaster impacts that we adopt here.
They distinguish between direct and indirect damages. Direct damages are the damage to fixed assets
and capital (including inventories), damages to raw
materials and extractable natural resources, and of
course mortality and morbidity that are a direct consequence of the natural phenomenon. Indirect damages refer to the economic activity, in particular the
production of goods and services, that will not take
place following the disaster and because of it. These
indirect damages may be caused by the direct damages to physical infrastructure, or because reconstruction pulls resources away from production.
These indirect damages also include the additional
costs that are incurred because of the need to use
alternative and potentially inferior means of production and/or distribution for the provision of normal
goods and services. At the household level, these
indirect costs also include the loss of income resulting from the non-provision of goods and services or
Sen (1981), in his seminal economic history of
famines, famously observed that starvation is the
characteristic of some people not having enough
food to eat. It is not the characteristic of there being
not enough food to eat. In Sen’s work, the central
emphasis is that the costs associated with what we
define as natural disasters are largely determined by
economic and social forces rather than predetermined by natural processes. Sen’s observation suggests that economics is important not only in understanding what happens after a disaster occurs, but
rather that the very occurrence of disasters is an economic event.
* Inter-American Development Bank.
** University of Hawaii.
The views and interpretations in this document are those of the
authors and should not be attributed to the Inter-American
Development Bank, or to any individual acting on its behalf. This
paper is an abridged version of Cavallo and Noy (2009). We thank
Oscar Becerra for excellent research assistance.
1
Chile’s vast experience with prior earthquakes, its prudent macroeconomic policies in the last two decades and its copper sovereign
wealth fund have all been used to motivate predictions about a
speedy recovery in the aftermath of the earthquake (Barrioneuvo
2010).
25
CESifo Forum 2/2010
Focus
only of direct damages (e.g. damage to infrastructure, crops and housing). An alternative but similar
source that is less extensive, and only parts of which
are publicly available, is the Munich Re dataset at
http://mrnathan.munichre.com/. A similar data collection effort with similar coverage but more limited access is maintained by another reinsurer, Swiss
Re. For an analytical review of selected data sets on
natural disasters, see Tschoegl et al. (2006).
from the destruction of previously used means of
production. These costs can be accounted for in the
aggregate by examining the overall performance of
the economy, as measured through the most relevant
macroeconomic variables. They can also be further
divided, following the standard distinction in macroeconomics, between the short run (up to several
years) and the long run (at least five years but sometimes also measured in decades). We use this distinction in the discussion that follows.
A few papers use other data sources. Most notable
are those that aim to estimate the impact of
storms/hurricanes. These papers use data on storm
intensity, typically measured by wind speed or storm
radius that are taken from the US National Oceanic
and Atmospheric Administration (NOAA) and the
Pielke et al. (2008) database.
The second section begins with a brief review of the
main data sources used in this largely empirical literature. The third section discusses the determinants of
the direct effects, while the fourth section examines
the indirect effects. The fifth section focuses on policy, while the sixth section describes several case studies of specific disasters and the insights gained from
them. The final section 7 summarizes and points to
several significant gaps in this literature.
Before reviewing the evidence on the impacts of natural disasters, it is useful to describe the stylized
facts. First, natural disasters, as defined in the EMDAT database, are fairly common events and their
reported incidence has been growing over time.
Figure 1 plots the average number of natural events
(including hydro-meteorological and geophysical
events) per country over the span of the last four
decades. The figure shows that the incidence of disasters has been growing over time everywhere in the
world. For example, in the Asia-Pacific region which
is the region with the most events, the incidence has
grown from an average of 11 events per country in
the 1970s to over 28 events in the 2000s.2 In other
regions, while the increase is less dramatic, the trend
is similar. However, these patterns appear to be driven to some extent by improved recording of milder
events, rather than by an increase in the frequency of
occurrence. Furthermore, truly large events – i.e.
conceivably more catastrophic – are rarer. Both of
these facts are shown in Figure 2, where the sample
is restricted to large events only, and where ‘large’ is
defined in relation to the world mean of direct damage caused by natural events.3
Data on disasters
Almost all the empirical work we survey here relies
on the publicly available Emergency Events
Database (EM-DAT) maintained by the Center for
Research on the Epidemiology of Disasters (CRED)
at the Catholic University of Louvain, Belgium
(http://emdat-be/). The database is compiled from
various sources, including UN agencies, non-governmental organizations, insurance companies, research
institutions and press agencies.
EM-DAT defines a disaster as a natural situation or
event which overwhelms local capacity and/or necessitates a request for external assistance. For a disaster to be entered into the EM-DAT database, at least
one of the following criteria must be met: (a) ten or
more people are reported killed; (b) 100 people are
reported affected; (c) a state of emergency is
declared; or (d) a call for international assistance is
issued. Disasters can be hydro-meteorological
including floods, wave surges, storms, droughts, landslides and avalanches; geophysical including earthquakes, tsunamis and volcanic eruptions; and biological covering epidemics and insect infestations (these
are much more infrequent in this database).
As is evident from the figure, there is no time trend
for the subset of large events in any region.
Moreover, the frequency of occurrence of ‘large’
disasters is significantly smaller than for all events.
For example, while there are more than 28 events
The data report the number of people killed, the
number of people affected and the dollar amount
of direct damages in each disaster. The amount of
material damage reported in the database consists
CESifo Forum 2/2010
2
The numbers corresponding to the decade of 2000 were adjusted
to account for the fact that there is one fewer year of reported data
in this decade.
3 A large disaster occurs when its incidence, measured in terms of
people killed as a share of population, is greater than the world
pooled mean for the entire sample period.
26
Focus
Figure 1
INCIDENCE OF NATURAL DISASTERS BY REGION 1970–2008
Total number of disasters by region
Hidro-meteorological and geological
30
Number of events per country
Africa
Asia-Pacific
C&E Europe
W Europe
LAC
the 1970s, and excluding the
most recent 2009–2010 events,
almost three million people
were reportedly killed by natural disasters in the three most
vulnerable regions.
25
20
15
10
5
0
1970s
1980s
1990s
2000s
Note: 2000s figures were adjusted to account for the fewer number of years in the decade.
Source: Authors' calculations based on data from EM-DAT database.
per country on average in the Asia-Pacific region
in the 2000s, the frequency of occurrence of large
events is only 0.5 episodes per country. This suggests that the threshold for what constitutes a disaster (and hence gets recorded in the dataset) is
quite low.
In summary, natural events are
frequent although ‘large’ events
– the ones that would typically
be considered catastrophic – are
rarer. The direct costs associated
with these events are huge, and
developing countries bear the
lion’s share of the burden, in
terms of both casualties and
direct economic damages.
Determinants of initial disaster costs
A spate of papers in the last several years has
attempted to understand the determinants of the initial direct costs of disasters. When evaluating the
determinants of disasters, most papers estimate a
model of the form:
The overwhelming majority of people affected and
killed by natural disasters reside in developing
countries, particularly in the Asia-Pacific region.
Figures 3 and 4 show that 96 percent of the people
killed and 99 percent of the people affected by
natural disasters over the period 1970–2008 were
in the Asia-Pacific region, Latin America and the
Caribbean, or Africa, whereas the combined population share of these three regions is approximately 75 percent of the world population. Since
DISit = + Xit + it .
(1)
where DISit is a measure of direct damages of a disaster in country i and time t; using measures of primary initial damage such as mortality, morbidity or
capital losses. Xit is a vector of control variables of
interest with each paper distinguishing different
independent variables; typically Xit will include a
measure of the disaster magnitude (i.e. Richter scale for earthFigure 2
quakes or wind speed for hurricanes) and variables that capINCIDENCE OF 'LARGE' NATURAL DISASTERS BY REGION 1970–2008
Total number of disasters by region
ture the ‘vulnerability’ of the
Over the mean – hidro-meteorological and geological
country to disasters (i.e. the conNumber of events per country
1.0
ditions which increase the susAfrica
Asia-Pacific
C&E Europe
W Europe
LAC
ceptibility of a country to the
0.8
impact of natural hazards).
Instead of estimating these pan0.6
els, several papers aggregate the
data across time and estimate
0.4
cross sections of country observations.
0.2
0.0
1970s
1980s
1990s
2000s
Note: 2000s figures were adjusted to account for the fewer number of years in the decade.
Source: Authors' calculations based on data from EM-DAT database.
27
One of the conditions that may
increase a country’s susceptibility to the impact of natural disas-
CESifo Forum 2/2010
Focus
Figure 3
DISTRUBUTION OF FATALITIES BY NATURAL DISASTERS 1970–2008
All disasters
are rare in less developed countries (see, for example, Freeman
et al. 2003).
Notwithstanding this, Kellenberg and Mobarak (2008) suggest a more nuanced, nonlinear
relationship between economic
LAC
8%
development and vulnerability
to natural disasters, with risk initially increasing with higher
incomes as a result of changing
behaviors, such as residents
locating to more desirable but
Africa
27%
more dangerous sites near coasts
and floodplains. Sadowski and
Sutter (2005) provide some confirmation for this view by examining hurricanes in the United
States and the ways in which better preparedness
leads to higher residential coastal concentrations
(where the risk from hurricane-associated wave
surges is actually higher).
C&E Europe 1%
W Europe 3%
North America 1%
Asia-Pacific
60%
Source: Authors' calculation based on data from EM-DAT database.
ters is its level of economic development. Kahn
(2005) estimates a version of equation (1) and concludes that while richer countries do not experience
fewer or less severe natural disasters, their death toll
is substantially lower. In 1990, a poor country (per
capita GDP < 2000 US dollars) typically experienced 9.4 deaths per million people per year, while
a richer country (per capita GDP > 14,000 US dollars) would have had only 1.8 deaths. This difference
is most likely due to the greater amount of resources
spent on prevention efforts and legal enforcement
of mitigation rules (e.g. building codes). In particular, some of the policy interventions likely to ameliorate disaster impact, including land-use regulations, building codes and engineering interventions,
Other papers focus on the political and institutional
factors that affect disaster impact. A consistent finding of several studies (i.e. Kahn 2005; Skidmore and
Toya 2007; Plümper and Neumayer 2009; Raschky
2008) is that better institutions – understood, for
instance, as more stable democratic regimes or
greater security of property rights – reduce disaster
impact. Anbarci et al. (2005) elaborate on the political economy of disaster prevention. They conclude
that inequality is important as a determinant of prevention efforts: more unequal
societies tend to have fewer
resources spent on prevention,
Figure 4
as they are unable to resolve the
DISTRIBUTION OF AFFECTED PEOPLE BY NATURAL DISASTERS
collective action problem of
1970–2008
All disasters
implementing preventive and
mitigating measures. In a similar
vein, Besley and Burgess (2002)
C&E Europe 0.5%
observe that flood impacts in
W Europe 0.3%
India are negatively correlated
North America 0.4%
with newspaper distribution;
LAC 3%
they attribute this effect to the
fact that when circulation is
Africa
Asia-Pacific
6%
90%
higher, politicians are more
accountable and the government
is more active in both preventing
and mitigating the impacts of
disasters. Eisensee and Strömberg (2007) reach similar concluSource: Authors' calculation based on data from EM-DAT database.
CESifo Forum 2/2010
28
Focus
sions regarding the response of US disaster aid to
media reports.
occurrence and sometimes a measure of the disaster
magnitude – either using physical criteria such as
wind-speed or earthquake magnitude or using measures of primary initial damage such as mortality,
morbidity or capital losses. Xit is a vector of control
variables that potentially affect Yit.
Healy and Malhotra (2009) add to this literature by
identifying the lack of political accountability for
elected public officials in the United States as an
explanation for inefficient allocation decisions.
Voters reward candidates for post-disaster aid but
not for well-funded prevention. Thus, the public sector under-invests in preventing these catastrophic
events, but readily spends on post-disaster reconstruction and aid.
In order to facilitate investigations into the interaction of the initial disaster impact with country-specific conditions, equations such as:
Yit = + Xit + DISit + DISit Vit + Vit + it
are used, where the Vit variables are the hypothesized
interactions of disaster impact with macroeconomic,
institutional or even demographic or geographic
characteristics. In these specifications, the coefficients
of interest are typically H and the vector E.
In summary, thinking of natural disasters as economic phenomena and not as purely exogenous events
has led researchers to seek to explain the fundamental structural determinants of the direct damages
incurred from disasters. While the damage caused by
disasters is naturally related to the physical intensity
of the event, the literature has identified a series of
economic, social, and political characteristics that
also affect vulnerability. A by-product of this analysis, of course, is that these characteristics are therefore potentially amenable to policy action.
Noy (2009) estimates a version of equation (3) and
finds that natural disasters have an adverse shortrun impact on output dynamics. In addition he
describes some of the structural and institutional
details that make this negative effect worse. He concludes that countries with a higher literacy rate, better institutions, higher per capita income, higher
degree of openness to trade, higher levels of government spending, more foreign exchange reserves,
and higher levels of domestic credit but with less
open capital accounts are better able to withstand
the initial disaster shock and prevent further
spillovers. Subsequently, Raddatz (2009) extends the
investigation to the impact of various types of natural disasters on countries in different income
groups. He concludes that smaller and poorer states
are more vulnerable, especially to climatic events,
and that most of the output cost of climatic events
occurs during the year of the disaster. His evidence
also suggests that, historically, aid flows have done
little to attenuate the output consequences of climatic disasters. More recent work adds more detail
by differentiating between different types of events
and different economic sectors (e.g. Hochrainer
2009; Loayza et al. 2009).
Cross-country studies of indirect impacts
A disaster’s initial impact causes mortality, morbidity and loss of physical infrastructure (residential
housing, roads, telecommunication and electricity
networks as well as other infrastructure). These initial impacts are followed by consequent impacts on
the economy (in terms of income, employment, sectoral composition of production, inflation, etc.).
Macroeconomics generally distinguishes between
the short run (usually up to three years) and the long
run (anything beyond five years is typically considered the long run). In what follows we summarize
the literature on the indirect economic effects of natural disasters.
When evaluating the determinants of these consequent impacts of disasters in a regression framework, most recent papers estimate a model of the
form:
Yit = + Xit + DISit + it .
(3)
Several papers pursue similar investigations, but
instead of relying on cross-country panels, they rely
on more detailed panels at the firm, county, region,
or the state level. Strobl (2008) uses differences in
hurricane impact on coastal counties in the United
States; Noy and Vu (2009) use provincial disaster
data from Vietnam; and Leiter et al. (2009) uses
European firm-level data.
(2)
where Yit is the measured consequent impact of
interest (e.g. per capita GDP). DISit is a measure of
the disaster’s immediate impact on country i at
time t; it is sometimes a binary indicator of disaster
29
CESifo Forum 2/2010
Focus
of disasters have been under-investigated.
Rodriguez-Oreggia et al. (2009) and Mechler
(2009) innovate by examining poverty and human
development (the World Bank’s Human Development Index, HDI) and consumption, respectively,
instead of the standard growth variables. The first
paper shows a significant increase in poverty and
a decline in the HDI in disaster-affected municipalities in Mexico; poverty increases by
1.5–3.6 percentage points. The second paper finds
a small decrease in household consumption for
low-income countries hit by disasters.
In summary, the emerging consensus in the literature is that natural disasters have, on average, a
negative impact on short-term economic growth.
Yet, the channels that are responsible for this economic slowdown have not been described methodically at all.
In investigations into the long-term impact of natural disasters, Skidmore and Toya (2002), and Noy and
Nualsri (2007) reach diametrically opposite conclusions, with the former identifying expansionary and
the latter contractionary disaster effects. More
recently, Jaramillo (2009) finds qualified support for
the Noy and Nualsri (2007) conclusion. Skidmore
and Toya (2002) explain their somewhat counterintuitive finding by suggesting that disasters may be
speeding up the Schumpeterian ‘creative destruction’ process that is at the heart of the development
of market economies. Cuaresma et al. (2008) attempt
to investigate this creative destruction hypothesis
empirically by closely examining the evolution of
R&D from foreign origin and how it is affected by
catastrophic risk. They conclude that the creative
destruction dynamic most likely only occurs in countries with high per capita income. For developing
countries, disaster occurrence is associated with less
knowledge spillover and a reduction in the amount
of new technology being introduced. Hallegatte and
Dumas (2009) critically examine the creative
destruction hypothesis using a calibrated endogenous growth model. They conclude that disasters are
never positive economic events and find that large
disasters that overwhelm local reconstruction capacity may actually lead to poverty traps.
The fiscal impact of natural disasters has also been
under-investigated. On the expenditure side, publicly
financed reconstruction costs may be very different
than the original magnitude of destruction of older
capital that occurred (Fengler et al. 2008). On the
revenue side of the fiscal ledger, the impact of disasters on tax and other public revenue sources has also
seldom been quantitatively examined. Using panel
VAR methodology, Noy and Nualsri (2008) estimate
the fiscal dynamics likely in an ‘average’ disaster;
however, they acknowledge that the impacts of disasters on revenue and spending depend on the country-specific macroeconomic dynamics occurring following the disaster shock, the unique structure of
revenue sources (income taxes, consumption taxes,
custom dues, etc.), and large expenditures.4
Borensztein et al. (2009) utilize data from Belize to
estimate in a calibrated model the likely fiscal insurance needs of a government. Barnichon (2008) calculates the optimal amount of international reserves
for a country facing external disaster shocks using a
similar methodology.5 Yang (2008) and Bluedorn
(2005) investigate the evolution of capital flows following disasters, and both conclude that disasters
generate some inflows (mostly international aid but
also other types of flows like remittances).
When compared to the short-run research, the literature on the long-run effects of natural disasters is scant
and its results inconclusive. Cavallo, Galiani, Noy and
Pantano (2010) provide the most recent attempt to
bridge that gap. They implement a new methodological approach based on a comparative event study
approach. The idea is to construct an appropriate
counterfactual – i.e. what would have happened to the
path of gross domestic product (GDP) of the affected
country in the absence of a natural disaster and to
assess the disaster’s impact by comparing the counterfactual to the actual path observed. The paper concludes that unless a natural disaster triggers a radical
institutional change; it is unlikely to affect economic
growth in the long run.
Case studies of disaster impacts
Several research projects have examined the economic impact of specific disaster events. Examples
are the 1995 Kobe earthquake in Japan (Horwich
2000), the 1999 earthquake in Turkey (Selcuk and
Yeldan 2001) and hurricane Katrina in 2005
(Hallegatte 2008; Vigdor 2008). Most of these are
descriptive, though some also construct calibrated
models that simulate the dynamics of the economy
Almost all existing research focuses on domestic
production (GDP) or on incomes; other impacts
CESifo Forum 2/2010
4
5
30
Lis and Nickel (2009) examine similar questions.
See also Cárdenas et al. (2007) and Mechler et al. (2009).
Focus
Besides policies that can reduce initial disaster damage, policies that can reduce the longer-term economic damage that disasters can wreak should also
be contemplated. We have already observed that
large disasters typically lead to reduced production
and incomes, even if the exact distribution of these
effects and their causes are not yet clear. Yet, as
Freeman et al. (2003) observe, some of the other likely macroeconomic impacts of disasters may be a
deteriorating trade balance, downward pressure on
the exchange rate, and upward pressure on prices.
How to deal with these likely dynamics is a policy
question that also needs to be asked.
after it is hit by the disaster and are therefore able
to tentatively evaluate various policy responses.
More recently, Cavallo, Powell and Becerra (2010)
estimate Haiti’s economic damage to fixed assets,
extractable natural resources and raw materials in
the aftermath of the earthquake that struck the
Caribbean country on 12 January 2010 – by far the
most catastrophic natural disaster in modern
records in terms of fatalities (relative to the country’s population).
These analyses were typically written not very
long after the event considered. In contrast,
Coffman and Noy (2009) investigate the long-term
impact of a 1992 hurricane on the economy of a
Hawaiian island. In this case, the long horizon
available, the unexpectedness of the event, and the
existence of an ideal control group subjected to
almost identical conditions but not the hurricane
itself, enables them to argue that in spite of massive transfers, it took nearly seven years for the
island’s economy to return to its pre-hurricane per
capita income level. The hurricane also resulted in
an out-migration of residents from which the
island’s population has not fully recovered. The
island permanently ‘lost’ about 15 percent of its
population as a result of the hurricane, even
though very few deaths were associated with the
storm.
Ex-ante insurance vs. ex-post disaster financing
Kunreuther and Pauly (2009) survey some of the
problems associated with ex-ante insurance coverage for large natural events: uncertainty with regard
to the magnitude of potential loses, highly correlated risk among the insured, moral hazard that leads
to excessive risk taking by the insured, and an
adverse selection of insured parties caused by
imperfect information. Their work also distinguishes
between unknown disasters (those for which the
likelihood and the distribution of probable magnitudes are at least partially known) and the unknowable (those for which no information is available).
Even though natural disasters are typically not
unknowable, these problems still clearly lead to
under-insurance. In all recent disasters, even in ones
that happened in heavily insured countries like the
United States, only a relatively small portion of
actual damages was insured. For example, hurricane
Katrina led to insurance claims totaling 46.3 billion
US dollars; while the estimated damage of the storm
was 158.2 billion US dollars.7
Policies and disasters
Perrow (2007), in a recent book on reducing catastrophic vulnerabilities in the United States, argues
that public policy should focus on the need to
‘shrink’ the targets: lower population concentration
in vulnerable (especially coastal) areas, and lower
concentration of utilities and other infrastructure in
disaster-prone locations. This advice stems from the
awareness that more ex-post assistance to damaged
communities generates a ‘Samaritan’s dilemma’, i.e.
an increase in risk-taking and a reluctance to purchase insurance when taking into account the help
that is likely to be provided should a disaster strike.6
However, apart from these ex-ante ‘shrink-the-target’ policies, many other ex-ante and ex-post policies
that can alleviate or worsen the economic impact of
disasters will necessarily be weighed before and after
any large event.
Insurance for the public sector, in order to secure the
availability of reconstruction expenditures, is also an
important policy question. There is broad consensus
on the need to design fiscal management policies to
resist the stress caused by the occurrence of disasters. Freeman et al. (2003) consider ways to create
the necessary fiscal space to deal with catastrophic
risk. Among various alternatives, they advocate
treating natural disasters as a contingent liability for
the national government (although they are skeptical about this suggestion’s practical feasibility, par7
Katrina insurance claim data are from Kunreuther and Pauly
(2009), while the figure for total damages is taken from EM-DAT.
The Congressional Budget Office estimates 70 to 130 billion US
dollars as direct damages (excluding the cost of clean-up and
repairs) for hurricanes Katrina and Rita.
6
This is similar to the moral hazard problem common in insurance
markets. Raschky and Weck-Hannemann (2007) define it as ‘charity hazard’.
31
CESifo Forum 2/2010
Focus
measures, such as purchasing market insurance, to
offset some of the costs. Healy and Malhotra (2009)
present evidence to support these conjectures even
for the transparent and fairly stable political system
of the United States. However, since governments
are typically held accountable for their response to
disasters, they have strong incentives to massively
invest in ex-post assistance.
ticularly in low-income countries). A more substantive initiative would be to implement an annual budgetary allocation to provide for natural disaster
expenditure when needed. Mexico’s FONDEN
(Fondo Nacional de Desastres Naturales) provides
this kind of fiscal provisioning against the risk of natural disasters. But these measures, while prudent,
amount to forms of self-insurance, which may be
very costly in the case of an economy with substantial credit constraints.
Of the three obstacles that deter the development
of a catastrophic risk insurance market, the one
related to market unavailability has been the most
studied. The consensus is that governments in countries that are vulnerable to natural disasters appear
to have only a limited set of options available to
insure public finances against those risks, although
progress is slowly being made. Hofman and Brukoff
(2006), Cárdenas (2008), Andersen (2007), and
Miller and Keipi (2005) survey some recent initiatives in this regard. The risk profile of catastrophe
insurance claims differs from that of other insurance products. A company providing car insurance
can easily diversify if it has many clients, since the
volume of claims would then be highly predictable.
In contrast, natural disasters are low-probability
events that can cause extremely large losses when
they occur and are thus not easily diversifiable in
the same way as car insurance. This low level of
diversification increases the cost of insurance. Its
price is very volatile and fluctuates sharply every
time there is a major catastrophic event that
depletes reserves. Primary insurers need to transfer
a considerable share of their catastrophe exposure
to large reinsurers, and this increased reliance on
reinsurers increases the cost of primary insurance,
reducing its attractiveness and scope.
Borensztein et al. (2009) argue that, in the case of
developing countries exposed to large natural disasters, insurance – or debt contracts with insurancelike features – provides an attractive alternative to
self-insurance. For example, they examine the vulnerability of Belize’s public finance to the occurrence of hurricanes and the potential impact of
insurance instruments in reducing that vulnerability.
Through numerical simulations they show that catastrophic risk insurance significantly improves
Belize’s debt sustainability.
Implementing disaster insurance in developing
countries, however, faces three types of obstacles:
paucity of markets, political resistance and inadequate institutional framework. For a number of
reasons, markets have traditionally been insufficiently developed or simply nonexistent
(more on this below). More recently, however,
advances such as the development of parametric
insurance policies have expanded the availability
of coverage for countries and households
(Cárdenas 2008).8
Political reluctance to engage in insurance purchase
derives from the fact that there is little short-run
benefit to be gained from entering into insurance
contracts. Insurance involves costs today and a possible payoff in the undetermined future, when the government may have already changed hands. In addition to these incentive problems, disasters are widely
considered as ‘acts of God’ (or natural phenomena),
and politicians are not often blamed for their occurrence. Politicians and policy-makers therefore face
very weak incentives for adopting relatively complex
Private capital markets offer some complementary
alternatives that may increase the availability of
financing options as they continue to develop. The
first capital market instrument linked to catastrophe
risk (‘cat bonds’) was introduced in 1994 as a means
for reinsurers to transfer some of their own risks to
capital markets. Since then, their success has prompted governments and international institutions to
explore their use as a mean of shielding government
budgets from the impact of natural disasters
(Andersen 2007). A catastrophe bond is a tradable
instrument that facilitates the transfer of the risk of
a catastrophic event to capital markets. In May 2006
and again in October 2009, the Mexican government
obtained earthquake and hurricane insurance by
8 Instead of basing payments on an estimate of the damage suffered, parametric insurance contracts establish the payout as a
function of the occurrence or intensity of certain natural phenomenon, as determined by a specialized agency such as the US
National Hurricane Center. In this way, the transaction costs and
uncertainty associated with insurance payments are considerably
reduced. There is no need to verify and estimate damages, and no
potential disagreement or litigation about the payouts. Moreover,
the country has immediate access to the resources when the disaster takes place.
CESifo Forum 2/2010
32
Focus
Pakko (2007) argue that this result arises because
the anti-inflationary justification for the contractionary policy will trump any desire to temporarily
expand output.
means of cat bonds and a direct purchase of coverage from international reinsurers.9
While these are encouraging developments, the private catastrophic risk market is still in its infancy.
And even if the supply side of risk financing instruments becomes fully developed, important questions
remain unanswered. For example: what is the optimal level of insurance that countries should purchase given the cost of insurance, the menu of alternative financing options (self-insurance, ex-post debt
accumulation, foreign aid, etc.), and country characteristics (access to external credit, macroeconomic
environment, institutional quality, etc.)? What is the
appropriate institutional set-up that ensures the
proper functioning of insurance schemes while minimizing moral hazard and adverse selection? What is
the appropriate role of the government vis-à-vis the
private sector in catastrophe insurance markets?
These are still open questions that warrant further
analysis.
In possibly the only empirical paper on exchange
rates and disasters, Ramcharan (2007) examines
exchange rate policy and its affect on the damage
inflicted by disasters. He finds consistent evidence
that flexible exchange rate regimes provide a cushion that ameliorates the disaster’s negative impact
on growth.
Conclusions and remaining questions
The economics of natural disasters are important. In
order to facilitate further necessary research on this
topic, we summarized the state of this literature. We
believe that large gaps in this literature remain. The
EM-DAT, the only internationally comparable and
available data on disasters, collects only limited
information on conceivably too many events.10 A
more detailed accounting of the physical destruction wrought by large disasters and their human toll
may prove to be very useful. We would especially
like to be able to distinguish among residential damage, crop devastation, infrastructure damage and
destruction of manufacturing facilities in order to
better address many of the questions that remain
unanswered.
Monetary and exchange rate policy
There has been very little research on the monetary
aspects of disaster dynamics. As far as we are aware,
even elementary questions such as, for example, the
inflationary impact of a large disaster and the aid
surge in its aftermath, have not been carefully examined. Open-economy questions, such as the impact of
disasters on exchange rates (real or nominal) or the
terms of trade have also not been examined empirically or analytically.
While the literature we reviewed examines the
short- and long-run effects of disasters and provides
detailed, if inconclusive, accounting of post-disaster
dynamics, it does not provide any description of the
channels through which disasters cause these effects.
An understanding of the channels of causality, in
both the short and the long run, will surely enable
more informed ex-post policymaking and possibly
better ex-ante preparation and mitigation.
Keen and Pakko (2007) construct a dynamic stochastic general equilibrium model calibrated for
the US economy and the impact of Katrina, and
evaluate the optimal response of monetary policy
to a Katrina-like shock. They find, intriguingly,
given public discussion and market perceptions at
the time, that optimal monetary policy design
should involve raising interest rates following a
large disaster. They show that this result holds for
both a Taylor-rule setting of interest rates, for optimal policy setting that replicates the efficient markets solution, and when the model includes nominal rigidities in both prices and wages. Keen and
We have presented some provisional evidence that
the extent of adverse impact is related to the ability
to mobilize significant funding for reconstruction.
We have also shown that poorer countries are likely
to suffer more from future disasters, but these countries are also unlikely to be able to adopt the
counter-cyclical fiscal policies that can pay for reconstruction. This constraint will make disasters’
9 The cat bond issued in 2006 (for a total of 150 million US dollars
in coverage) was the first to cover disaster risk in Latin America.
The Mexican government has followed this initiative and introduced a new cat bond issue in October 2009 sponsored by
FONDEN. This USD 290 million three year cat bond provides
cover for earthquakes on the Pacific coast (140 million US dollars),
Pacific hurricanes (100 million US dollars) and Atlantic hurricanes
(50 million US dollars). Coverage will last for three years.
10 Since the threshold used to determine what constitutes a disaster
is quite lenient, the dataset contains limited information on a large
variety of events.
33
CESifo Forum 2/2010
Focus
Coffman, M. and I. Noy (2009), Hurricane Iniki: Measuring the
Long-Term Economic Impact of a Natural Disaster Using Synthetic
Control, University of Hawaii Working Paper 09-05.
adverse consequences more severe in poorer developing countries. A better-targeted reconstruction
that is informed by the identified channels of transmission can potentially alleviate some of these
resource constraints.
Cuaresma, J. C., J. Hlouskova and M. Obersteiner (2008), “Natural
Disasters as Creative Destruction? Evidence from Developing
Countries”, Economic Inquiry 46, 214–226.
Dercon, S. (2004), “Growth and Shocks: Evidence from Rural
Ethiopia”, Journal of Development Economics 74, 309–329.
A further significant lacuna in the current state of our
knowledge is the absence of any agreement regarding the long-run effects of these disasters. Whether
these disagreements have any substantial real relevance to policy decisions can only be assessed when
the channels of transmission and propagation for
these effects become more evident.
ECLAC (2003), Handbook for Estimating the Socio-economic and
Environmental Effects of Disasters, United Nations Economic
Commission for Latin America and the Caribbean.
Eisensee, T. and D. Strömberg (2007), “News Floods, News
Droughts, and U.S. Disaster Relief”, Quarterly Journal of
Economics 122, 693–728.
Fengler, W., A. Ihsan and K. Kaiser (2008), Managing Post-Disaster
Reconstruction Finance: International Experience in Public
Financial Management, World Bank Policy Research Working
Paper 4475.
Freeman, P. K., M. Keen and M. Mani (2003), Dealing with Increased
Risk of Natural Disasters: Challenges and Options, IMF Working
Paper 03/197.
We have not reviewed the micro-development literature that has been examining the ways in which
households (typically rural households) deal with
sudden disaster events (e.g. Townsend 1994; Udry
1994; Dercon 2004). Whether these shed light on the
channels of transmission is a possibility that needs to
be further explored. Nor have we reviewed the literature on aid allocations following disasters and their
impact. This small literature was recently surveyed
by Strömberg (2007) who also provides stylized facts
on who gives relief, how much is given and who
receives it.
Hallegatte, S. (2008), “An Adaptive Regional Input-Output Model
and Its Application to the Assessment of the Economic Cost of
Katrina”, Risk Analysis 28, 779–799.
Hallegatte, S. and P. Dumas (2009), “Can Natural Disasters Have
Positive Consequences? Investigating the Role of Embodied
Technical Change”, Ecological Economics 68, 777–786.
Healy, A. and N. Malhotra (2009), “Myopic Voters and Natural
Disaster Policy”, American Political Science Review 103, 387–406.
Hochrainer, S. (2009), Assessing the Macroeconomic Impacts of
Natural Disasters – Are There Any?, World Bank Policy Research
Working Paper 4968.
Hofman, D. and P. Brukoff (2006), Insuring Public Finances against
Natural Disasters: A Survey of Options and Recent Initiatives, IMF
Working Paper 06/199.
Horwich, G. (2000), “Economic Lessons of the Kobe Earthquake”,
Economic Development and Cultural Change 48, 521–542.
References
Jaramillo, C. R. H. (2009), Do Natural Disasters Have Long-Term
Effects on Growth?, Universidad de los Andes, mimeo.
Anbarci, N., M. Escaleras and C. A. Register (2005), “Earthquake
Fatalities: The Interaction of Nature and Political Economy”,
Journal of Public Economics 89, 1907–1933.
Kahn M E. (2005), “The Death Toll from Natural Disasters: The
Role of Income, Geography, and Institutions”, Review of Economics
and Statistics 87, 271–284.
Andersen, T. J. (2007), Developing and Supporting the Use of
Disaster-Linked Financial Instruments: The Role of the IDB in Latin
America and the Caribbean, Inter-American Development Bank,
Working Group on Disaster Risk Financing.
Keen, B. and M. Pakko (2007), Monetary Policy and Natural Disasters
in a DSGE Model: How Should the Fed Have Responded to Hurricane
Katrina?, St. Louis Fed Economics Working Paper 2007-25.
Barrioneuvo, A. (2010), “Chile Braces for a Major Economic
Slowdown”, New York Times, 15 March.
Kellenberg, D. K. and A. M. Mobarak (2008), “Does Rising Income
Increase or Decrease Damage Risk from Natural Disasters?”,
Journal of Urban Economics 63, 788–802.
Besley, T. and R. Burgess (2002), “The Political Economy of
Government Responsiveness: Theory and Evidence from India”,
Quarterly Journal of Economics 117, 1415–1451.
Bluedorn, J. C. (2005), Hurricanes: Intertemporal Trade and Capital
Shocks, Nuffield College Economics Paper 2005-W22.
Kunreuther, H. and M. Pauly (2009), “Insuring against
Catastrophes”, in: Diebold, F. X., N. J. Doherty and R. J. Herring
(eds.), The Known, the Unknown and the Unknowable in Financial
Risk Management, Princeton: Princeton University Press.
Borensztein, E., E. Cavallo and P. Valenzuela (2009), “Debt
Sustainability under Catastrophic Risk: The Case for Government
Budget Insurance”, Risk Management and Insurance Review 12,
273–294.
Leiter, A. M., H. Oberhofer and P. A. Raschky (2009), “Creative
Disasters? Flooding Effects on Capital, Labor and Productivity
within European Firms”, Environmental and Resource Economics 43, 333–350.
Cárdenas, V., S. Hochrainer, R. Mechler, G. Pflug and J. LinneroothBayer (2007), “Sovereign Financial Disaster Risk Management: The
Case of Mexico”, Environmental Hazards 7, 40–53
Loayza, N., E. Olaberría, J. Rigolini and L. Christiansen (2009),
Natural Disasters and Growth-Going beyond the Averages, World
Bank Policy Research Working Paper 4980.
Cárdenas, V. (2008), Financiamiento de Riesgos Catastróficos
Naturales, IADB Research Department Working Paper 663.
Mechler, R. (2009), Disasters and Economic Welfare: Can National
Savings Help Explain Post-disaster Changes in Consumption?,
World Bank Policy Research Working Paper 4988.
Cavallo, E., A. Powell and O. Becerra (2010), “Estimating the Direct
Economic Damage of the Earthquake in Haiti”, Economic Journal
(forthcoming).
Mechler R., S. Hochrainer, A. Aaheim, Z. Kundzewicz, N. Lugeri, and
M. Moriondo (2009), “A Risk Management Approach for Assessing
Adaptation to Changing Flood and Drought Risks in Europe”, in:
Hulme, M. and H. Neufeldt (eds.), Making Climate Change Work for
Us: European Perspectives on Adaptation and Mitigation Strategies,
Cambridge: Cambridge University Press, 200-229.
Cavallo, E. and I. Noy (2009), The Economics of Natural Disasters:
A Survey, IDB Working Paper 124.
Cavallo, E., S. Galiani, I. Noy and J. Pantano (2010), Catastrophic
Natural Disasters and Economic Growth, Inter-American
Development Bank, mimeo.
CESifo Forum 2/2010
Miller, S. and K. Keipi (2005), Strategies and Financial Instruments
for Disaster Risk Management in Latin America and the
34
Focus
Caribbean, Inter-American Development Bank Document ENV145.
Noy, I. (2009), “The Macroeconomic Consequences of Disasters”,
Journal of Development Economics 88, 221–231.
Noy, I. and A. Nualsri (2007), What Do Exogenous Shocks Tell Us
about Growth Theories?, University of Hawaii Working
Paper 07–28.
Noy, I. and A. Nualsri (2008), Fiscal Storms: Public Spending and
Revenues in the Aftermath of Natural Disasters, University of
Hawaii Working Paper 08-09.
Noy, I. and T. Vu (2009), “The Economics of Natural Disasters in
Vietnam”, Journal of Asian Economics (forthcoming).
Pelling, M., A. Özerdem and S. Barakat (2002), “The
Macroeconomic Impact of Disasters”, Progress in Development
Studies 2, 283–305.
Perrow, C. (2007), The Next Catastrophe: Reducing Our
Vulnerabilities to Natural, Industrial, and Terrorist Disasters,
Princeton: Princeton University Press.
Pielke, Jr., R. A., J. Gratz, C. W. Landsea, D. Collins, M. Saunders and
R. Musulin (2008), “Normalized Hurricane Damages in the United
States: 1900-2005”, Natural Hazards Review 9, 29–42.
Plümper, T. and E. Neumayer (2009), “Famine Mortality, Rational
Political Inactivity, and International Food Aid”, World Development 37, 50–61.
Raddatz C. (2009), The Wrath of God: Macroeconomic Costs of
Natural Disasters, World Bank Policy Research Working Paper 5039.
Ramcharan, R. (2007), “Does the Exchange Rate Regime Matter
for Real Shocks? Evidence from Windstorms and Earthquakes”,
Journal of International Economics 73, 31–47.
Raschky, P. A. (2008), “Institutions and the Losses from Natural
Disasters”, Natural Hazards Earth Systems Science 8, 627–634.
Raschky, P. A. and H. Weck-Hannemann (2007), “Charity Hazard –
A Real Hazard to Natural Disaster Insurance?”, Environmental
Hazards 7, 321–329.
Rodriguez-Oreggia, E., A. de la Fuente and R. de la Torre (2009),
The Impact of Natural Disasters on Human Development and
Poverty at the Municipal Level in Mexico, mimeo.
Sadowski, N. C. and D. Sutter (2005), “Hurricane Fatalities and
Hurricane Damages: Are Safer Hurricanes More Damaging?”,
Southern Economic Journal 72, 422–432.
Selcuk, F. and E. Yeldan (2001), “On the Macroeconomic Impact of
the August 1999 Earthquake in Turkey: A First Assessment”,
Applied Economics Letters 8, 483–488.
Sen, A. (1981), Poverty and Famines: An Essay on Entitlement and
Deprivation, Oxford: Oxford University Press.
Skidmore, M. and H. Toya (2002), “Do Natural Disasters Promote
Long-run Growth?”, Economic Inquiry 40, 664–687.
Skidmore M. and H. Toya (2007), “Economic Development and the
Impacts of Natural Disasters”, Economic Letters 94, 20-25.
Strobl, E. (2008), The Economic Growth Impact of Hurricanes:
Evidence from U.S. Coastal Counties, IZA Discussion Papers 3619.
Strömberg, D. (2007), “Natural Disasters, Economic Development,
and Humanitarian Aid”, Journal of Economic Perspectives 21,
199–222.
Townsend R. (1994), “Risk and Insurance in Village India”,
Econometrica 62, 539–591.
Tschoegl, L. (2006), An Analytical Review of Selected Data Sets on
Natural Disasters and Impacts, Paper Prepared for the
UNDP/CRED Workshop on Improving Compilation of Reliable
Data on Disaster Occurrence and Impact, Bangkok, 2-4 April 2006.
Udry, C. (1994), “Risk and Saving in Northern Nigeria”, American
Economic Review 85, 1287–1300.
Vigdor, J. (2008), “The Economic Aftermath of Hurricane Katrina”,
Journal of Economic Perspectives 22, 135–154.
Yang, D. (2008), “Coping With Disaster: The Impact of Hurricanes
on International Financial Flows”, B.E. Journal of Economic
Analysis and Policy 8, 1–43.
35
CESifo Forum 2/2010
Focus
begun to tap these opportunities – indeed, only
3 percent of potential losses in developing countries
are insured compared to 45 percent in advanced
countries – and more frequent and intensive use of
insurance markets may be desirable. This article discusses available insurance modalities and a few
promising initiatives in developing and emerging
market countries, along with some key challenges for
the insurance community, donors and international
financial institutions.
MITIGATING THE IMPACT OF
NATURAL DISASTERS ON
PUBLIC FINANCES
DAVID HOFMAN*
In September 2004, the small Caribbean island of
Grenada was severely hit by hurricane Ivan. The
category 3 windstorm, afterwards nicknamed ‘Ivan
the Terrible’, caused an estimated 200 percent of
GDP in damage on the island. The disruptive
impact of the storm caused the local economy to
contract sharply, while at the same time public
spending needs soared. Faced with the overwhelming costs of the event, soon after the storm, the
Grenadian authorities saw themselves forced to
approach their creditors for a voluntary restructuring of the island’s public debts.
Preparing for disaster
Although natural disasters have taken their toll
throughout history, there are strong indications that
they have become more frequent and severe in
recent decades and that this upward trend is set to
continue in the period ahead. In part, this trend can
be explained by growing urbanization which has led
to an increasing concentration of population in vulnerable areas (see Freeman et al. 2003). For another
part, it reflects changes in weather patterns – possibly associated with the rise in global surface temperatures – which appear to have caused an increase in
the frequency and intensity of adverse weather
events such as hurricanes, floods and droughts (see
e.g. Webster et al. 2005). With more frequent and
intense natural disasters affecting increasingly
densely populated areas, their costs have risen
strongly over time.
Natural disasters (such as catastrophic hurricanes)
can have far-reaching negative effects on macroeconomic conditions in affected countries, including on
their public finances. And this is especially the case
in developing and smaller countries. Developing
countries are often unable to marshal the substantial
resources needed in the aftermath of a major disaster. Smaller countries (such as the small island states
in the Caribbean and the South Pacific) are typically
unable to achieve the geographic redistribution of
risk available to larger countries, which can subsidize
the costs associated with catastrophic events by
using revenues from unaffected regions. In these
countries, therefore, the large costs associated with
natural disasters can quickly diminish the public sector’s ability to respond effectively.
Natural disasters can put considerable pressure on
public finances. In the wake of a disaster, governments typically face a weakened revenue base while
pressures on spending are likely to soar. Such pressures could come from short-term disaster relief
operations, the need to restore key public infrastructure, or from them provision of financial support to
the private sector (for example, a government will
often be called upon – or even be required by law-to
restore damaged or destroyed housing).
Catastrophe insurance markets, however, increasingly offer opportunities for the transfer of such risks.
Thus far, developing countries have only tepidly
* International Monetary Fund. This article is an update based on
an earlier paper written with Patricia Brukoff (2006), Insuring
Public Finances Against Natural Disasters – A Survey of Options
and Recent Initiatives, IMF Working Paper WP/06/199. The views
expressed in this article are those of the author and should not be
attributed to the IMF, its Executive Board, or its management.
CESifo Forum 2/2010
To meet immediate expenditure needs, developing
disaster-prone countries often rely on ex post financing in the form of grants and loans from external
36
Focus
Table 1
Large natural catastrophes and estimated losses 1950–2008
(billion US dollars, constant 2008 prices)
1950–59
1960–69
1970–79
Number of events
20
27
47
Overall losses
53.6
93.3
161.7
Average loss
2.7
3.5
3.4
Sources: Munich Re; Guy Carpenter & Co.; author’s calculation.
donors. Relying on such flows, however, has considerable disadvantages, including because of uncertainty about financing following a disaster. It takes
considerable time before donor resources are committed and even more time before the funds are
actually made available. And there may be ‘competition’ for donor resources from other countries with
relief needs at the same time. Indeed, it is often
found that donor contributions following disasters
fall short of actual needs (see e.g. Wong et al. 2009).
Another disadvantage is that to the extent that help
comes in the form of loans, it could add to already
high public debt stocks.
1980–89
63
262.9
4.2
1990–99
91
778.3
8.6
2000–08
37
620.6
16.8
Dilemma. With predictable insurance payouts, in
contrast, countries retain incentives for fiscal provisioning and preventive structural policies.
Choosing the right insurance
Governments that seek to shield their public finances
from the impact of natural disasters by means of
insurance face a few key choices. A first choice for
governments pertains to who should be the insurance
taker and what should be insured? The inability of the
private sector to cope with the impact of a disaster is
often a key source of budgetary pressures following a
disaster. Therefore, one useful strategy involves promoting, facilitating or subsidizing the purchase of
insurance by private sector parties (for instance, property insurance for homeowners or crop insurance for
farmers) in order to limit the government’s contingent
liabilities. Alternatively, or as a complementary strategy, a government can also seek to insure itself directly
against disaster-related outlays, or budgetary pressures more broadly, in a lump-sum manner.
Providing for disasters by means of insurance, in
contrast, secures at least some of the needed
resources in advance. Such insurance is not a remote
theoretical prospect. The experience in high-income
countries, in particular the United States and Japan,
has shown that many natural perils are insurable,
and markets for disaster risk insurance are well
established there.
Given trends in catastrophe insurance pricing and
the available resources in the countries involved,
donor contributions will often be needed, ex ante, to
contribute to the premia. But such a shift from ex
post to ex ante donor financing still has important
benefits for both parties. From the perspective of the
recipient it introduces an important element of predictability into post-disaster public finance conditions since the available amount of insurance financing would be known in advance. From the perspective
of donors it helps smooth cash flow by converting ‘if
and when’ outlays into predictable insurance premia. It might also give donors greater leverage over
preventive policies (such as building codes). Last,
but not least, it reduces the perverse incentives that
recipient countries face in their dependence on postevent donor financing. Indeed, vulnerable countries
currently often have little incentive to set aside fiscal
savings or take preventive measures for natural disasters, since this might reduce donor support following an adverse event – the so-called Samaritan’s
A second key choice for governments regards the
degree to which the risk is transferred and the entity
that ultimately comes to bear the risk. The various
modalities differ crucially in the size of the pool of
risk capital among which the risk is spread. There are
several options:
• Pooling. At one end of the spectrum, countries
can pool their disaster risk with other countries –
thus creating a form of cooperative insurance.
Such a mechanism can be effective when the
number of countries sharing the risk is large
enough, and the correlation of risks between participating countries is low.
• Commercial insurance and reinsurance.
Insurance companies, however, may be better
placed to absorb risks because they typically
maintain a well-diversified portfolio of risks.
Further, second tier insurance is available
through reinsurers, who act as the insurance
37
CESifo Forum 2/2010
Focus
ed on the basis of actual and verified losses. The key
advantage of this type of trigger is that the insurance
payout is typically close to the actual loss incurred.
There are, however, also important disadvantages,
such as time-consuming claims settlement and moral
hazard issues. The use of alternative parametric
insurance triggers can alleviate some of these disadvantages, while also offering greater scope for the
standardization of contracts and thereby facilitating
the transfer of risks to capital markets.
companies of the insurers, allowing the latter to
pass on risks that exceed their absorptive capacity. In fact, because of its peculiar loss-distribution – with low payouts in most years, but sudden
spikes in disaster years – a large portion of catastrophic risk ends up with reinsurers. However,
reinsurers, too, have at times had difficulty coping with peaks in insurance claims, which is
reflected in a high volatility of reinsurance premia (for instance, insurance premiums, as measured by the ‘rate-on-line’, rose sharply following costly disasters such as hurricane Andrew in
1992 and hurricane Katrina in 2005).
• Capital markets. There where risks are testing reinsurers’ capacity, capital markets are progressively
providing risk capital that can be tapped by both
reinsurers and countries themselves through the
use of insurance-linked securities.This is an encouraging development because by allocating risks –
and potential losses – efficiently over a large pool of
investors, insurance through capital markets offers
promising prospects of reducing the premium
volatility associated with traditional reinsurance.
Parametric insurance uses objective variables that
are exogenous to the policy holder but have a strong
correlation with losses against which insurance is
desired. The payout is determined upfront and is
conditional on the chosen exogenous variable reaching a preset threshold within a certain time period.
An example of parametric insurance are so-called
weather derivatives, which link payouts to the occurrence of a certain weather event (such as wind
speeds exceeding, or precipitation falling short of,
certain pre-agreed thresholds). Parametric insurance
could be seen as essentially an informed bet against
the elements. As such, parametric insurance contracts are kindred to the options and futures contracts traded on financial markets and distinct from
traditional indemnity-based insurance.
Advances in catastrophe insurance
The possibilities for passing risk to capital markets
have been greatly enhanced by two related innovations: the use of parametric insurance triggers and
the growth of the ‘cat bond’ market.
In contrast to indemnity-based insurance, parametric
insurance tends to have a benign incentives structure.
Since payout and actual damage are not directly linked,
moral hazard is limited and the insured party retains
incentives for prevention and mitigation of risks.
Parametric insurance – keeping it simple
Another key advantage of parametric insurance contracts is their relative simplicity and transparency. The
use of an exogenous variable
greatly reduces the information
Figure 1
asymmetries associated with traditional insurance and eliminates
CATASTROPHE INSURANCE RATE-ON-LINE: 1990 TO 2009
the need for an assessment or
1990 = 100
450
verification of actual damage.
400
Consequently, transaction costs
350
are relatively low. A related advantage is the potential speed of
300
payout, which, in contrast to
250
indemnity-based insurance, can
200
be a matter of weeks or even days
150
after the contract is triggered.
Traditionally, insurance has relied on indemnitybased triggers where insurance payouts are calibrat-
100
50
0
1990
1992
1994
1996
1998
2000
2002
2004
Source: Guy Carpenter & Co.
CESifo Forum 2/2010
38
2006
2008
Since parametric insurance uses
objective and often publicly available information, it also allows for contract standardization,
Focus
thereby facilitating risk transfer
to international capital markets.
Indeed, as more sophisticated
systems (including satellite
imagery) become available to
monitor and measure natural
events, parametric insurance contracts have the potential to become increasingly palatable to
international capital markets.
Moreover, such technological advances also increasingly facilitate
the reliable monitoring of events
in developing countries, thereby
expanding their possibilities to
successfully tap international
insurance and capital markets.
Figure 3
STRUCTURAL OVERVIEW OF A CAT BOND ISSUANCE
A. Transaction
Principal
Premium
Special Purpose
Sponsor
Sponsor
Vehicle (SPV)
Investors
Coupon
Coupon
B. Possible end positions
Insurance
Coverage
Sponsor
Sponsor
Principal
Special Purpose
Vehicle (SPV)
(if catastrophe
occurs)
Investors
(if no
no catastrophe
catastrophe
(if
occurs)
Source: Adapted from Chacko et al. (2004)
(2004).
issuance has suffered from the global financial crisis, the market has been recovering swiftly over the
past year.
The main inherent disadvantage of parametric insurance triggers is the so-called basis risk: since there is
no relation (at least ex post) between the predetermined payout and actual damage, the insurance
claim may either exceed or undershoot the actual
loss. Refinements in loss modeling, however, can
potentially reduce basis risk.
The typical cat bond issue involves the establishment,
by the ‘sponsor’ (usually a reinsurance company but
conceivably another entity), of a Special Purpose
Vehicle (SPV). The task of this SPV is to issue the
bond and to invest the capital in low-risk securities
Cat bonds – tapping a wider market
(such as treasuries). The returns on these investments
are paid to the holders of the bonds, together with a
A key example of an innovative instrument which
premium that is paid by the sponsor (see Figure 3,
emergence was facilitated by the use of parametric
panel A). If the bonds mature without a prespecified
insurance triggers is the catastrophe (or ‘cat’) bond.
event (i.e. a narrowly defined type of catastrophe)
Cat bonds have been an important vehicle for the
having taken place, the principal is repaid to the
transfer of catastrophe risks to capital markets.
investors, similar to regular bonds (panel B).
Indeed, the market for cat bonds has grown rapidly
However, in the event that the prespecified catastrosince its inception in the second half of 1990s, and
phe does occur within the life time of the bond,
while – like many other financial instruments –
investors agree to forfeit part or
all of their claims, and the SPV
Figure 2
will pay out to the sponsor
instead. The catastrophe risk is
CATASTROPHE BOND ISSUANCE: 1997 TO 2009
Volume
thus transferred to the investors.
million US dollars
Number of issues
40
8 000
35
7 000
30
6 000
25
5 000
20
4 000
15
3 000
10
2 000
5
1 000
0
0
1997
1999
2001
2003
2005
2007
Source: Guy Carpenter & Co.
39
2009
Because assets and liabilities
related to the bond issue are
allocated with the SPV, cat
bonds function as a pure insurance arrangement for the sponsor, and are not debt creating.
The key advantage of cat bonds
is that it allows for the break up
and transfer of risks to a large
group of investors in cases
where insurance with a single
counter party might not be
CESifo Forum 2/2010
Focus
against specific outlays, governments can seek
general, lump-sum support that is conditional on
a certain disaster taking place. Such funds could
then be spent at the government’s discretion.
Schemes of this type have been gaining some
popularity in recent years. For instance, the
World Bank has implemented a scheme along
these lines in the Caribbean from May 2007. This
‘Caribbean Catastrophe Risk Insurance Facility
(CCRIF)’, is the first multi-country risk pool in
the world and helps insure 16 islands in the
region – including Grenada, which was mentioned above – to insure against the risk of earthquakes and hurricanes, using parametric insurance triggers. The pool is funded by resources
from participating governments and contributions from donors, while risks that exceed the
capacity of the pool are being transferred to
reinsurance markets. This two-tier structure
allows the facility to cope with large losses and
also provides participating governments with
insurance coverage at about half the price they
would have paid if they had purchased insurance
individually (Wong et al. 2009). The Caribbean
facility has so far proven successful and it has
made several payouts over the past 3 years. Most
recently, a payout was made to Haiti, which
received 8 million US dollars from the fund
within two weeks after a devastating earthquake
hit the island in early 2010. Based on the positive
experiences, the CCRIF is currently considering
the possibility of insuring more frequent events
and widening its coverage to include flood coverage and agricultural damage, while the World
Bank is preparing a similar initiative for Pacific
island countries.
available or be more expensive. From the perspective of the investor, cat bonds yield above-market
rates (since a premium is paid on top of the lowrisk/risk-free return), while offering a unique
opportunity for portfolio diversification as a catastrophe risks tend to be uncorrelated with trends in
stock or bond markets.
Key initiatives in low and middle-income countries
In recent years, there have been several promising
initiatives in low- and middle-income countries,
some of which have benefitted from the recent innovations in insurance. The initiatives can be divided
into three broad categories:
• Schemes aimed at limiting government contingent
liabilities. These schemes target the private sector
so as to reduce the need for government support
following disasters. A good example is the Turkish
Catastrophe Insurance Pool (TCIP), which is supported by the World Bank and pools and reinsures risks from a compulsory earthquake insurance scheme for private home-owners. Similarly,
there have been World Bank and International
Finance Corporation (IFC) supported projects
that helped offer drought insurance to individual
farmers in several low-income countries, including India and Malawi.
• Schemes to provide resources for disaster relief
and reconstruction. With these schemes, the government seeks to secure resources to cover relief
operations in the event of a catastrophe. An
example is the 2006 World Food Program (WFP)
project in Ethiopia that used a weather derivative to ensure resources in the case of a catastrophic drought. In this case, the insurance
money was designed to be spent by the government and the WFP was responsible to relieve the
plight of affected farmers, while donors contribute to the premia. Another example is
FONDEN in Mexico. This fund started as a
means of earmarking resources for future disaster relief, to be spent by local governments on an
as-needed basis. In May 2006, the fund got on
more secure financial footing when Mexico
became the first middle-income country to issue
a cat bond to secure sufficient funds in the event
of a major earthquake, with a verifiable parametric trigger.
• Schemes to provide lump sum support to the government budget. Instead of purchasing insurance
CESifo Forum 2/2010
Amid these encouraging initiatives, the key area
where exploring is still in its early stages, is the
transfer of risk to capital markets. Thus far, only
Mexico has significant experience with tapping the
international capital market by means of cat
bonds. Two promising new World Bank projects,
however, are likely to enable a broader range of
low and middle income countries to tap a range
of financial market counterparties and capital
markets.
• Catastrophe bond issuance platform. Building on
Mexico’s experience, in October 2009, the World
Bank launched the ‘MultiCat program’, a catastrophe bond issuance platform that will make it
easier for governments and public entities in low
40
Focus
ricane Katrina in 2005 – have been raising doubts
about the way forward. Indeed, the insurance industry has been paying increasing attention to climate
change and its implications for their risk modeling
and risk management. The increasing risk of natural
disasters, or persistent uncertainty with respect to
the effects of climate change, may have an adverse
effect on catastrophe insurance availability and premia going forward.
income countries to access the cat bond market.
Under the platform, the World Bank will act as
arranger for the transactions and all bonds
issued will carry the common MultiCat brand
name and benefit from a common legal structure
and documentation. This standardization reduces the set up costs to the issuing countries,
and also makes the bonds more palatable to
investors.
• Weather derivative intermediation. In 2008, the
World Bank has also begun to offer intermediation services to low-income countries that want
to use weather derivatives. Here, the World
Bank intermediates the risk of weather-based
catastrophes by entering into mirroring transactions with the client country and a financial
market counterpart. In the event of a severe
weather event, the country would receive a payout from the World Bank, the total value of
which would be based on a parametric index
used as an estimate of the financial impact. The
payout would be funded with the payout that
the World Bank would receive from the financial market counterpart in the mirroring transaction. Malawi has been the first country to
use this new facility, purchasing insurance cover
against a drought-related shortfall in maize
production, with Swiss Re as the insurance
provider. Malawi was sponsored by Britain
(DFID) to cover the costs of the insurance premium.
A second source of uncertainty lies with the appetite
for catastrophe risk in international capital markets.
Up to the global financial crisis that broke in the fall
of 2008, issuers have had relatively few problems in
selling the innovative and relatively risky cat bonds
to international investors seeking risk diversification. But the success of these new (and relatively low
volume) instruments was spurred by favorable global liquidity conditions and a quest for yield on the
part of investors, which led to a gradual decline in
risk premia. Although early indications of recovery
in the cat bond market are encouraging, it remains to
be seen whether a similar favorable environment
will prevail in the years ahead.
These issues aside, affordability of catastrophe insurance for developing countries will remain an issue
even under more favorable scenarios. Indeed, in light
of the frequent high cost and volatility in insurance
premia, the viability of catastrophe insurance mechanisms for developing countries may crucially depend
on the contribution of donors, particularly in the lowincome context. Mobilizing further, and continued,
donor support for disaster insurance schemes is therefore another challenge. While donor involvement so
far is encouraging, it is uncertain whether there is a
willingness to increasingly engage in structural support arrangements at the expense of post-disaster
relief. The latter remains the norm and may offer
greater benefits in terms of public recognition and in
satisfying the urge to show support after a catastrophe
has taken place. Thus, further developing sustainable
models for collaboration among donors and recipients in disaster insurance schemes remains key.
Weathering storms on the horizon
Transferring risks to international capital markets
has substantial benefits because it greatly expands
the pool of insurance capital available to developing
countries, and significant progress has been made in
recent years. Nonetheless, there remain a number of
uncertainties associated with the insurance of natural disaster risk. Importantly, even though there are
well-established markets for insuring certain catastrophe risks, it cannot be taken for granted that all
natural disaster risks can be insured in the market at
an affordable cost. Specifically, the catastrophe
insurance market faces two sources of uncertainty:
the first is climate change and its possible effect on
the frequency and intensity of natural disasters.
While the insurance industry has coped so far, the
record insurance losses in recent years – including
high losses from a multitude of events in 2008 and
the record-breaking USD 45 billion losses from hur-
The potential benefits of a change from ex post to ex
ante insurance financing are considerable. While natural disasters are likely to remain a painful fact of
life, such a shift would at least help reduce the second round fiscal effects, thereby limiting economic
disruption and facilitating faster recovery, while also
providing better incentives for the adoption of preventive policies.
41
CESifo Forum 2/2010
Focus
References
Freeman, P., M. Keen and M. Mani (2003), Dealing with Increased
Risk of Disasters: Challenges and Options, IMF Working Paper
WP/03/197.
Heller, P. and M. Mani (2002), “Adapting to Climate Change”,
Finance and Development 39, 29–31.
Rasmussen, T. (2006), “Natural Disasters and Their Macroeconomic
Implications”, in: Ratna Sahay et al. (eds.), The Caribbean – From
Vulnerability to Sustained Growth, Washington DC: IMF, 181–203.
Webster, P. J., G. J. Holland, J. A. Curry and H. R. Chang (2005),
“Changes in Tropical Cyclone Number, Duration, and Intensity in a
Warming Environment”, Science 309, 1844–1846.
Wong, Y. C., A. Lemus and N. Wagner (2009), Insuring Against
Natural Disasters in the Caribbean, IMF Country Report 09/176,
Washington DC: IMF.
CESifo Forum 2/2010
42
Focus
caused by natural disasters for many countries for
years 1972 through 2005. Extensive information on
disasters that has been underutilized by economists
is available from the Office of US Foreign Disaster
Assistance/Center for Research on the Epidemiology of Disasters (OFDA/CRED). We use
these data to estimate the relationship between fiscal decentralization and the effects of natural disasters, while controlling for a range of other factors
found to be important determinants of disasterinduced human fatalities.
NATURAL DISASTER IMPACTS
AND FISCAL
DECENTRALIZATION
HIDEKI TOYA* AND
MARK SKIDMORE **
Introduction
It is generally understood that as a country develops, it devotes greater resources to safety, including
implementing precautionary measures designed to
reduce the impacts of natural disasters. The recent
onslaught of hurricanes/typhoons and earthquakes
in the Caribbean and Asia along with the accompanying devastating human and economic impacts has
spurred interest in the factors that determine the
patterns and types of resulting losses. Previous
research (Anbarci, Escalaras and Register 2005;
Kahn 2005; Toya and Skidmore 2007; Kellenberg
and Mobarak 2008) demonstrates that there is a
distinguishable and predictable pattern between
losses from natural disaster events and economic
development. Paralleling efforts to spur development around the world has been a growing interest
among policymakers and economists in fiscal
decentralization.1 While the existing research on
the impacts of decentralization are generally positive in terms of public service delivery, as pointed
out by Bardham (2002) existing studies are ‘largely
descriptive, not analytical, and often suggest correlations rather than causal processes’. Because government plays such vital roles in both the preparation for and response to disaster events, and
because naturally occurring disaster events are not
systematically related to levels of development
(Kahn 2005), natural disasters may provide an
opportunity to evaluate the effectiveness of governmental structure in protecting human life. In this
study, we merge fiscal, economic, demographic and
geographic data with information on total deaths
To preview the main finding, our analysis shows that
while controlling for a variety of factors, nations
with governments that are more decentralized experience fewer disaster-related deaths. However, we
also document important interrelationships
between decentralization, educational attainment
and disaster impacts. Places with more decentralized
government systems also tend to have higher average educational attainment. The analysis suggests
that the one route by which decentralization
reduces disaster-induced deaths is through human
capital. Generally, our findings suggest that decentralized governments are more effective in disaster
preparations and/or responses relative to more centralized governmental systems.
The following section provides a review of the literature and a discussion of fiscal decentralization and
disaster mitigation. This section also offers a theoretical discussion of the effects of fiscal decentralization
on disaster mitigation decisions. In sections three
and four we present the empirical framework and
analysis, respectively. Section five concludes.
Literature review and theoretical discussion
We review two relevant strands of literature: the
research on fiscal decentralization and the economics of natural disasters. In these contexts, we further
limit our review to those studies that examine the
performance of fiscal decentralization and research
on the role of economic development in mitigating
disaster-related fatalities.
* Nagoya City University.
** Michigan State University.
1 See Bardham (2002) for a review of the extensive literature on fiscal federalism in developing countries.
43
CESifo Forum 2/2010
Focus
Fiscal decentralization
Another related line of research has focused on the
role of decentralization in promoting economic
growth. The limited number of existing studies on
this topic provides mixed evidence. Davoodi and
Zou (1998) use a panel data set of 46 countries over
the 1970-1989 period to examine the relationship
between fiscal decentralization and growth, reporting evidence of a negative relationship between
decentralization and growth for developing countries. They attribute this finding to the potential inefficient use of resources at the local level in developing countries. Xie, Zou and Davoodi (1999) look at
again at this same issue only in the context of the
United States. This analysis suggests that existing
spending shares of state and local governments are
consistent with growth maximization, and that further efforts at decentralization in the United States
might be counterproductive. Akai and Sakata (2002)
use data for the 50 US states over the 1992 through
1996 period to determine the relationship between
several measures of decentralization and growth. In
contrast to previous work, they report some evidence
that decentralization may encourage economic
growth. Thornton (2007) uses data from OECD
countries to reexamine this issue, finding decentralization has no statistically significant impact on
growth. In terms of theory, Brueckner (2006) develops an endogenous growth model to more carefully
explore the connection between decentralization and
economic growth. The model offers a formal analysis
to show that more units of government allows for differing levels of public goods across jurisdictions in
order to better meet differing community demands
for public services. This creates an incentive to
increase savings and investment in human capital,
and this in turn generates a higher rate of economic
growth. Of special relevance to the present study is
the recent work of Arze del Granado, MartinezVasquez and McNab (2005), which provides empirical support for Brueckner’s conjecture by showing
that decentralization may increase public education
spending. This line of research is relevant to the issue
of the role of decentralization in protecting human
life for two reasons. First, it demonstrates that there
is an important interconnection between decentralization and economic well-being. This suggests that if
we are to isolate the effect of decentralization on
preserving human life, we will want to control for
measures of economic development. Second, there
may important interactions between decentralization
and educational attainment that require consideration in assessing the determinants of disaster-induced
fatalities.
Of the substantial literature on fiscal decentralization there is a strand that focuses on assessing the
effectiveness of fiscal decentralization efforts in
transition and developing economies. As discussed
in Bardham (2002), even though decentralization
efforts are occurring in a number of countries
around the world, quantitative evidence of effectiveness is limited. Studies that do exist utilize several methodologies, but generally attempt to evaluate government service delivery in a ‘beforeafter’ framework. For example, Santos (1998) evaluated the decentralization initiative in Bolivia,
finding that access to basic sanitation services
(water and sewage) and utilization of elementary
and secondary schools increased two-fold following decentralization. Alderman (1998) utilized
household survey data to evaluate a social assistance program in Albania that was decentralized in
1995. He finds evidence of modest gains in efficiency and cost-effectiveness as a result of decentralization. Azfar, Kähnkönen and Meagher (2000)
surveyed households and government officials in
the Philippines to ascertain the matching of public
investment priorities between government officials
and local residents. They find that the stated priorities of municipal authorities more closely matched
those of local residents than did provincial authorities, suggesting that decentralization may improve
public investment decision-making. The 1994
World Development Report on Infrastructure
cited cases of cost savings as well as quality
improvement in public infrastructure projects following the transfer of management responsibility
to local authorities. The study, which included data
from 42 countries, cited numerous cases in which
decentralized governments were more effective in
providing infrastructure such as roads and water
supply at a lower cost.
These and a number of other studies suggest that
decentralized governmental systems provide public services more efficiently and at a lower cost
than more centralized systems. However, Bardham (2002) asserts that many of the studies are
unable to identify causal processes. Further, in our
review of this literature we found no studies that
utilized cross-country data to evaluate the effectiveness of decentralization. The lack of crosscountry analysis is largely due the unavailability of
comparable data on costs and effectiveness of government activity.
CESifo Forum 2/2010
44
Focus
designed to reduce the risk of the most hazardous
factors in the environment.
In the context of the even more specific literature
on disaster management and response, Wildasin
(2008) points out that the particular institutional
structures within fiscal federalism may create
incentives for local governments to limit financial
and policy preparations. In the United States, for
example, much of the incidence of local disasters is
shifted to the rest of society through intergovernmental transfers. Wildasin argues that it may be
necessary for central government to implement
new forms of federal control of subnational governments, including requirements for the creation
of ‘rainy day’ funds targeted at disaster management/recovery. Effective public-sector ex ante
avoidance measures as well as ex post disaster response requires coordination and sharing of financial costs between national and subnational governments. Subnational governments are perceived to
have a comparative advantage over national governments in the management of land use, economic
development, safety and other regionally-based
policies that affect disaster risk. On the other hand,
there is a role for national governments in setting
certain disaster-management policies. This suggests
that there may be an optimal mix of responsibility
between national and subnational governments in
disaster management activity, and more generally
in the devolution of public responsibilities.
The decision as to which of the existing hazards
should be mitigated depends on the marginal benefit and marginal cost of hazard reduction for each of
the potential hazards. For example, in the United
States reducing malaria, measles and small pox were
high priority issues in the early to mid twentieth
century because they were prime killers at the time.
As these diseases were mitigated via vaccines and
other means, other lower priority hazards came to
the forefront: cancer and heart disease became high
priority research issues much later. The implementation of seat-belt laws, child seats and other
improvements in automobile safety also occurred
much later. This sequence of hazard reducing policy
implementation is closely related the level of economic development. As income rises, it becomes
possible to reduce risks, and of course the most dangerous hazards ought to be mitigated first. We can
place risk from natural events within this continuum. The implementation of warning systems, building codes, emergency response plans, etc. will only
occur when income levels are high enough to support them and when other more hazardous factors
have first been addressed.
While both developed and developing countries
have some degree of disaster protection initiated by
the public sector, the degree to which economic
agents benefit from and are able to comply with and
employ their own established safety standards
depends on the level of income. For example,
whether or not new construction complies with code
depends on the costs of compliance relative to
income. In addition, at some threshold level of
income private disaster protection emerges
(Horwich 2000) – e.g. emergency and risk management departments in commercial and other enterprises, private disaster consultants, disaster property
insurances, including self-insurance through private
saving (Skidmore 2001).
Economics of natural disasters
A critical underlying factor in any economy’s
response to disaster events is its level of wealth.
Horwich (2000) argues that increased income translates to a general increase in the level of safety.
Wildavsky (1988) interprets the degree of safety
enjoyed by citizens of a country as a natural product
of a growing market economy. Wildavsky broadly
defines as protection against hazardous things and
circumstances (e.g. less dangerous machinery,
improved construction quality, more reliable automobile braking and steering mechanisms, and more
reliable means of transportation and communication). He describes a learning-process by which individual buyers weigh the cost of each technically feasible increment of safety against the expected benefit. In this framework, since demand for safety rises
with income, a nation’s per capita income is a good
initial indication of its degree of safety. An increase
in income provides not only leads to improvements
in general safety, but also additional protection
against natural disasters. As a society becomes
more developed it sequentially implements policies
Tol and Leek (1999) and Burton et al. (1993) also
discuss the potential for reduced vulnerability as
income increases. Burton et al. (1993) show a modest inverse relationship between deaths due to natural disasters and income for twenty countries for
years 1973 and 1986. As noted by Tol and Leek
(1999), there is probably a rapid transition between
relatively vulnerable and invulnerable that occurs
somewhere in the modernization process. According
45
CESifo Forum 2/2010
Focus
ing resources when and where they were most needed. In contrast, government officials suffered from
paralysis immediately following the quake, taking
days to mobilize and provide meaningful assistance.
Free market economists have used the term ‘decentralization’ as a synonym for privatization (Bardham
2002), arguing for the benefits of reducing the power
of a strong centralized government.
to Albala-Bertrand (1999), the people most affected
by direct disaster events are primarily those who
have weaker economic and political bases. While
disasters occur in both industrialized and developing countries, about 95 percent of the deaths occur
in the developing world (Alexander 1993). This discussion suggests that income and wealth are highly
correlated with the number of deaths caused by natural disasters. As an illustration, the United States
seems to have made such a transition during the
20th century. The annual average number of deaths
caused by hurricanes on the Atlantic coast during
the 1900–1940 period was 327. However, over the
1972–2005 there was only an average of about
58 deaths annually, including the 1,319 deaths
caused by hurricane Katrina.
In recent years governments have been criticized
for lack of preparation and inability in response to
extreme large magnitude disaster events. For example, in the aftermath of hurricane Katrina both state
and local officials in Louisiana were criticized for
not enhancing the flood control infrastructure,
which ultimately led to massive flooding. Public
responsibilities regarding safety, land use and economic development decisions, etc. play important
roles in limiting the impacts of natural catastrophes.
The first level of assistance in response to a disaster
event will come from police and fire authorities.
Such services account for a significant portion of
public service expenditures regardless of fiscal
structure. Within a decentralized governmental
system, public safety services are typically provided
by more autonomous subnational authorities. Importantly, the ex ante public sector preparations
(establishing and enforcing land use and building
code regulations, maintaining infrastructure, etc.) as
well as ex post response (public safety) is crucial in
terms of protecting human life. It would then seem
that data on deaths from natural disaster events
provides a good test case for examining the role of
government structure on public service delivery as
it relates to protecting the populace from the
potentially devastating impacts of catastrophic
events. Further, given evidence suggests that decentralized fiscal systems increase educational attainment, and education enables citizens to better prepare for and respond emergency circumstances.
In a recent study that utilizes disaster data from
OFDA/CRED, Kahn (2005) shows that income and
institutional quality are important determinants of
human casualties from natural disasters. Of special
note, Kahn (2005) also shows that, while the probability of disaster occurrence is not related to the level
of development, the number of deaths, injured and
homeless are reduced as income rises. His work also
suggests that more democratic countries experience
fewer human losses than do less democratic countries. Using data similar to that of Kahn (2005),
Anbarci and Escaleras (2005) examine the relationship between earthquake fatalities and income
inequality, finding that countries with greater
inequality experience greater losses. Toya and
Skidmore (2007) extend the research on the development-disaster relationship by examining additional factors such as human capital accumulation and
the degree of openness, finding that both greater
human capital and openness reduce losses from natural disaster events. Toya and Skidmore (2007) suggest that distinct from the private disaster-incomesafety relationship is the existence of an underlying
social/economic fabric that increases safety for all of
society. Even more recently, Kellenberg and
Mobarak (2008) demonstrate that there may be
important nonlinearities between the level of development and disaster impacts.
Wildasin (2008) identifies another important consideration: intergovernmental transfer mechanisms
may weaken the incentives for subnational governments to engage in ex ante disaster preparations.
Under the current US federal government emergency management policy, assistance has effectively
shifted much of the burden of disaster losses (particularly floods) to the rest of society. While such a policy relieves the financial stress on the affected
region, it also reduces the incentive for subnational
governments to invest in costly but effective disaster
avoidance activities. Wildasin (2008) suggests anoth-
As highlighted by Horwich (2000), disaster mitigation efforts (ex ante and ex post) are sometimes most
effectively implemented by the private sector. For
example, Horwich (2000) found that the most
responsive organization following the 1995 earthquake in Kobe was the Japanese mafia. This marketoriented group was especially effective at distribut-
CESifo Forum 2/2010
46
Focus
er approach to strengthen subnational incentives: mandate
subnational governments to
build disaster reserve funds.
Figure 1
FISCAL DECENTRALIZATION AND NATURAL DISASTER LOSSES
Number of killed/population for 1990–2005
0.9
0.8
This discussion outlines two
0.7
0.6
opposing forces in decentraliza0.5
tion with regard to disaster man0.4
agement and response. First, the
0.3
literature on the effectiveness of
0.2
decentralization efforts in the
0.1
0.0
developing world suggests that
-0.1
subnational governments may
0
10
20
30
40
50
60
provide many public services
Sub-national government expenditures/total government
expenditures for 1990–2000, average
more efficiently, including eduSource: Authors' calculation.
cation, and may be more responsive to heterogeneous local
association between disaster-related fatalities and
needs. An opposing view is the certain institutional
decentralization. While this slope coefficient is genstructures within fiscal federalism may weaken the
erated without controlling for other factors that
incentives for subnational units of government in
may affect the degree to which natural disasters
terms of ex ante preparation as well as ex post recovlead to death, the figure motivates a more carefully
ery from disaster events.
examine of the issue.
We add to both of these lines of research by examining the relationships between fatalities induced by
natural disaster events, education and decentralization while controlling for a number of other variables that characterize the level of economic development. Our analysis adds new insight to the
research the effectiveness of fiscal decentralization
as well as to the research on disaster mitigation. We
acknowledge that our empirical analysis provides a
more general evaluation: clearly, more focused country-specific analyses such as that of Wilsdasin (2008),
Chernick and Haughwout (2006) and Horwich
(2000) also yield important insights.
In order to rigorously evaluate the impact of decentralization on mitigating disaster losses, we control for
other factors found to be important in previous studies
(Kahn 2005; Anbarci, Escalaras and Register 2005;
Toya and Skidmore 2007) such as income, human
capital, openness and size of government. We examine
these issues by including per capita GDP, average
years of secondary and higher schooling completed,
openness ((exports + imports)/GDP), government size
(government expenditure/GDP) into our empirical
analysis. We also control for other factors such as a
time indicator variables, population, land area, OECD
dummy, disaster type that determine human fatalities
induced by catastrophic events.
We present some initial evidence regarding the
relationship between disasters and economic
growth in Figure 1. This figure shows the simple linear relationship between the natural logarithm of
the number of annual disaster-induced deaths and
the ratio of subnational to total government expenditures for 61 countries over the 1990–2005 period.
The vertical axis represents the natural logarithm
of natural disaster-induced deaths. Along the horizontal axis is the ratio of subnational to total government spending. The disaster data illustrated
in Figure 1 represents current and detailed information on natural disaster fatalities from the
Center for Research on the Epidemiology of
Disasters (CRED) (EMDAT 2004). The figure indicates a clear negative and statistically significant
To provide a more formal framework for our empirical analysis, consider the following Cobb-Douglas
production function which maps the relationship
between the production of safety that society enjoys
and the factors that determine safety:
(1)
Si = A ji K ái Lâi
where S measures the quantity of ‘safety’ produced
by society i, K is capital and L represents labor and
is equal to ψhN where h is average schooling years
and N is population. Following Topel (1999), we
specify human capital as an exponential function of
schooling in the production function. This implies
47
CESifo Forum 2/2010
Focus
lation as a control variable. From this model we
expect higher levels of GDP per capita and educational attainment to reduce death tolls, and higher
levels of population to increase death tolls. The
degree of trade openness is expected to reduce
deaths as well (Toya and Skidmore 2007). However, greater fertility and inequality are expected
to increase disaster-induced fatalities. We test
these notions in the empirical analysis which is
presented next.
that schooling will enter the ‘safety’ production function linearly.2 Last, Aji is a productivity parameter
specific to disaster type j (e.g. climatic or geologic).
Dividing both sides by population N yields:
á
â
S
K
N
(2) = A jt e h N á + â 1
N t
N t N t
Simplifying equation (2) results in:
(3)
S
á
h + -1
= A jt k t (et ) N t
N
t
Empirical analysis
where k is equal to physical capital per capita. To
convert the notion of ‘safety’ into a more tangible
measure, deaths (D) resulting from natural disasters,
consider the following relationship:
(4) D = f(S)N =
Data on natural disasters come from the
OFDA/CRED International Data Base (2006), and
macroeconomic data are available from several
sources (Barro and Lee 1996; International Financial Statistics website; Heston, Summers and
Aten 2002; World Development Indicators 2003).
Government fiscal data are taken from the World
Bank website. The OFDA/CRED database is a
result of collaboration between the Office of US
Foreign Disaster Assistance and the Center for
Research on the Epidemiology of Disasters. Efforts
to establish better preparedness for and the prevention of disasters have been a primary concern
for donor agencies, implementing agencies and
affected countries. For this reason, demand for
complete and verified data on disasters and their
human impact, by country and type of disasters has
been growing. The OFDA/CRED initiative to
develop a validated database on disaster impacts is
a response to this need.
N
S
where f is the probability of disaster induced death,
f’<0
_ and simplifying further yields:
Substituting D for N
S
(5)
h + -á
Dt = A -1
Nt
jt k t et
Finally, taking the natural logarithm of both sides
and combining lnA and lnψ together into Z, an overall ‘technology’ parameter, produces an equation
that is the basis for our empirical analysis:
(6)
ln (D )t = Z jt áln (k )t ht + (á + â )lln(N)t
Using this merged data set we conduct empirical
analyses to determine the relationship between
decentralization and disaster-induced fatalities,
while controlling for a range of other factors.
OFDA/CRED uses specific criteria in classifying a
natural disaster (ten or more people killed, 100 or
more people were affected/injured/homeless, significant damages were incurred, a declaration of a
state of emergency and/or an appeal for international assistance was made (http://www.cred.be/
emdat). Although the data set provides information on a number of natural disaster types, we
restrict our analysis to earthquakes, floods, slides,
volcanic eruptions and extreme winds.3 Summary
statistics and data sources are in presented in the
Appendix (Tables A1 and A2).
The productivity parameter, Zj, is then a function
of decentralization, government size (government
expenditure/GDP), as well as factors such as openness ((exports + imports)/GDP), fertility and a
measure of inequality. We also control for other
factors such as land area, OECD dummy, time
indicator variables and disaster type that determine human fatalities induced by catastrophic
events. Given that GDP is highly correlated with
k, we use GDP per capita as a proxy for (k).
Consistent with the model, we also include popu2
For example, consider a standard Cobb-Douglas production function:
y=Akαh1-α, where y is per capita GDP, A is the level of technology, k is
per capita capital stock, and h is per capita human capital.Transforming
this function into log form yields: lny = lnA + αlnk + (1-α)lnh. With a
Mincer-type model, h equals Ψes with s = average years of schooling.
Taking the log yields: lnyt = (lnAt) + α(lnkt) + (1-α)(lnht). Substituting
in for h yields: lnyt = lnAt + α(lnkt) + (1-α)(lnΨ + st).
CESifo Forum 2/2010
3
Hurricanes, tornadoes, typhoons, etc. are included in this
category.
48
Focus
where deathsjit is the total number of deaths5 caused by
natural disaster type j (hurricane, earthquake, flood,
etc.) in country i during period t; decentjit is defined as
sub-national own source expenditures/total government expenditures; viit is the ratio of national intergovernmental transfers to subnational expenditures; yjit
represents a vector of j variables that may determine
the deaths caused by the natural disaster (e.g. natural
logarithm of per capita GDP in real US dollars, a measure of human capital (years of secondary and higher
education schooling), openness ((exports+imports)/
GDP), population, land area, OECD dummy, a series
of indicator variables characterizing the type of disaster); and t represents a series of time indicator variables. In addition to controlling for various aspects of
development, it is critical to control for the size of government to isolate the impact of government structure.
We therefore include the ratio of government spending
to GDP as an additional control.
An average 193 deaths resulted from each recorded
disaster events over the 1970–2005 period.4 In our
sample, the most common types of disasters were
floods and extreme wind, accounting for 39 and
38 percent of the total, respectively. Seven percent of
the total resulted from slides, 12 percent were earthquakes, and volcanic eruptions accounted for just
three percent of the total.
It is also important to provide a careful definition of
our key independent variable, the subnational share of
total government expenditures. First, expenditure data
include intergovernmental transfers in addition to
functional expenditures. Thus, caution is warranted in
making comparisons across government tiers. Second,
these data only provide a proxy for expenditure autonomy because a large portion of subnational expenditures may be mandated by the central government.
Despite these limitations, this is probably the best and
most utilized measure of decentralization available. It
includes the following expenditure categories: general
public services; defense; public order and safety; education; health; social security and welfare; housing and
community amenities; recreational, cultural, and religious affairs and services; fuel and energy; agriculture,
forestry, fishing, and hunting; mining and mineral
resources, manufacturing and construction; transportation and communication; other economic affairs and
services; and other expenditures. Expenditures related
to disaster preparation and management occurs in a
number of functional categories. Clearly, the public
order and safety category is critical, but other functional areas like health, transportation and communication, and economic affairs are likely to also play
important roles. We also include a measure of vertical
imbalance, which is the ratio of intergovernmental
transfers from national government to subnational
government expenditures. Our analysis estimates
effects of the government structure in terms of spending at the subnational level while controlling for the
degree subnational reliance on national authorities for
financial resources. In some specifications, we also
interact decentralization with educational attainment
to determine how human capital alters government
productivity in providing safety.
As shown in Figure 2, roughly one-third of our
observed disaster events record zero deaths. Our
dependent variable is therefore truncated at zero:
(8) ln( deaths jit ) = max(0, ln( deaths jit ))
We therefore use a Tobit random effects specification to properly treat the censored variable within a
panel data framework.6
One last potentially important econometric issue
warrants our attention: it is possible that the likelihood of a disaster event being recorded depends, in
part, on the level of development. However, in a
careful analysis Kahn (2005) demonstrates that the
probability of a disaster event occurring and being
recorded is not dependent on the level of development: with the exception of floods, high and low
income countries are equally likely to experience a
naturally occurring disaster event. Relying on
Kahn’s result, we move directly to an analysis of the
determinants of disaster-related fatalities.
Results
We estimate a series of regressions to determine the
relationship between disaster-induced fatalities and
decentralization. Our basic regression is characterized by the following equation:
(7)
Table 1 presents a series of regressions using data for
all natural disasters recorded for all countries over
4 The average is pulled up by the catastrophic tsunami that
occurred in Asia in 2004. Omitting this event reduces the average
number of deaths to 124.
5 Since we use the natural logarithm of deaths, to avoid arithmetic
error we use ln(deaths+1) as our dependent variable.
6 The Tobit model within the fixed effects framework could potentially be used, but this econometric approach generates biased
parameter estimates.
ln(deaths jit ) = 1 ( decentit ) + 2 ( viit ) + n ( y jit )
n ( y jit ) + tt + e jit
49
CESifo Forum 2/2010
Focus
in reducing disaster-induced
deaths. One possible interpretaFREQUENCY DISTRUBUTION OF DISASTER-INDUCED FATALITIES
tion is that a route by which
Frequency
decentralization leads to fewer
1 000
deaths is through providing a
more educated population that
800
is better able to prepare for and
600
respond to catastrophic events.
Another possibility is a more
400
highly educated population enhances the effectiveness of a
200
decentralized government structure. Consider also the coeffi0
0
5
10
cients on government size and
In(deaths+1)
the degree of vertical imbalance.
Source: Authors' calculation.
The coefficients on government
size and vertical imbalance are
the 1970–2005 period. In column 1, we present a base
never significant in the regressions.
regression in which measures of decentralization are
excluded from the analysis. Column 2 presents a
Before turning to additional regressions in Table 2,
regression that includes a primary measure of decenconsider the coefficients on the other control varitralization (decent), but excludes educational attainables. The coefficient on GDP per capita is negative,
ment. The column 3 regression includes both educabut only statistically significant in three of the five
tional attainment and decentralization measures
regressions. As expected, greater trade openness
together simultaneously. A key purpose in presentreduces disaster-induced fatalities. Openness is
ing these first three regressions is to demonstrate
thought to enhance commodity distribution networks
important interrelationships between decentralizaso that international assistance can be allocated to
tion, education and death
tolls from disasters. Column 1
Table 1
results show negative relationFiscal decentralization and natural disaster losses
ship between educational attainDependent variable: Log (1+Number of killed)
ment and death tolls, although
1
2
3
the coefficient in only significant
Log (GDP per capita)
– 0.474
– 0.441
– 0.515
at the 90 percent level of
(– 1.905) (– 2.468) (– 2.249)
Secondary
and
higher
schooling
years
– 0.392
– 0.119
confidence. In column 2, we see
(–
2.757)
(–
0.938)
that the coefficient on decentralLog (Openness)
– 0.499
– 0.647
– 0.688
ization is negative and high(– 2.600) (– 4.188) (– 4.202)
ly significant. Now consider
Fertility
0.006
0.020
0.012
the regression presented in co(0.060)
(0.237)
(0.121)
lumn 3. Here, we include both
Gini coefficients
0.002
0.008
0.004
(0.157)
(0.928)
(0.431)
educational attainment and
Log (Size of government)
0.324
0.091
– 0.002
decentralization together in a
(1.087)
(0.421) (– 0.006)
single regression. The coefficient
Local Gov total / Total Gov total
– 2.057
– 1.837
on educational attainment is
(– 2.967) (– 2.324)
smaller than in column 1 and it is
Vertical imbalance
– 0.171
– 0.300
now insignificant. Also, the coef(– 0.483) (– 0.787)
No. of countries
59
69
55
ficient on decentralization is
No. of observations
2761
2712
2581
smaller than in column 2, but its
Log-likelihood
– 4956.1 – 4858.2 – 4632.9
statistical significance is mainNotes: Numbers in parentheses are z-values. Other independent variables
tained. These three regressions
not reported here are Constant, Log (Population), Log (Area), OECD,
demonstrate an important relaEarthquake, Flood, Volcano, Wind, Wave and a series of year indicator
variables.
tionship between decentralizaSource: Authors’ calculation.
tion and educational attainment
Figure 2
CESifo Forum 2/2010
50
Focus
areas in need more effectively. In
addition, countries with greater
openness may have access to
technologies and practices that
improve safety. Countries with
higher rates of fertility and
greater income disparities tend
to have more disaster-induced
deaths, although the coefficients
on these variables are only sometimes significant. These findings
are generally consistent with
Kahn (2005), and Toya and
Skidmore (2007).
In Table 2, we present additional
regressions to examine robustness of our key finding to the
inclusion of government ‘quality’ measures, and to further
explore the decentralizationeducational attainment relationship. In column 1 we repeat the
estimates presented in column 3
of Table 1, except we include a
measure of the degree of political freedom and a measure of
civil liberty. Note that our core
result remains: the coefficient on
the overall measure of decentralization maintains its statistical significance. Civil liberty and
political rights are not significant determinants of disaster deaths.
Table 2
Fiscal decentralization, human capital and natural disaster losses
Dependent variable: Log (1+Number of killed)
1
2
3
– 0.590
– 0.420
– 0.497
(– 2.941) (– 2.322) (– 2.645)
Secondary and higher schooling years – 0.075
0.184
(– 1.043)
(1.498)
Log (Openness)
– 0.712
– 0.589
– 0.581
(– 4.987) (– 4.820) (– 4.747)
Fertility
0.093
0.111
0.092
(1.171)
(1.517)
(1.239)
Gini coefficients
0.011
0.013
0.011
(1.399)
(1.720)
(1.503)
Log (Size of government)
– 0.129
– 0.308
– 0.192
(– 0.530) (– 1.553) (– 0.900)
Local Gov total / Total Gov total
– 1.575
(– 2.661)
(Local Gov total / Total Gov total)
– 0.357
– 0.620
Secondary and higher
(– 3.283) (– 2.996)
schooling years
Vertical imbalance
– 0.276
– 0.133
– 0.194
(– 0.941) (– 0.476) (– 0.686)
Political right
0.057
(0.862)
Civil liberty
– 0.037
(– 0.420)
No. of countries
54
55
55
No. of observations
2527
2581
2581
Log-likelihood
– 4518.3 – 4479.6 – 4475.3
Notes: Numbers in parentheses are z-values. Other independent variables
not reported here are Constant, Log (Population), Log (Area), OECD,
Earthquake, Flood, Volcano, Wind, Wave and a series of year indicator
variables.
Source: Authors’ calculation.
Log (GDP per capita)
having an educated work force from which to hire
capable public employees.
In columns 2 and 3, we examine whether decentralization enhances the degree to which education
reduces safety by including an interaction between
educational attainment and decentralization.
Consider column 2, which repeats the column 2
regression in Table 1 except that the decentralization measure is now replaced with the interaction
variable. The coefficient on the interaction variable
is negative and highly significant. Once we include
educational attainment as in column 3, the coefficient on the interaction term is smaller but it maintains its statistical significance. One interpretation
of this finding is that the productivity of an educated population in terms of protecting life is
enhanced by greater autonomy at the local level.
An alternative interpretation is that the ability of
local government to operate effectively and thus
produce a safer environment is greatly enhanced by
How many lives are likely to be saved as a result of
decentralization? From column 2 of Table 1, an
increase in subnational government expenditures
relative to total government expenditures of 0.2
reduces lives lost by about 30 percent. Thus, if a
very centralized government with a small ratio of
subnational to national expenditures of, say, 0.1
were to decentralize such that this ratio increased
to about the average (0.3), the number of disasterinduced fatalities would fall by 40 percent. If we
control for educational attainment as in column 3,
direct decentralization effect indicates that fatalities would fall by 36 percent. One should, however,
be careful not to extend such calculations to hyperbole: countries that are already decentralized are
not likely to save more lives as a result of further
decentralization.
51
CESifo Forum 2/2010
Focus
Robustness
Previous research that has sought to assess the
effectiveness of fiscal decentralization is limited in
terms of identifying causal relationships, and more
systematic analysis has been hampered by the lack
of consistent cross-country data on government
costs and effectiveness (Bardham 2002). Despite the
tragic nature of catastrophic events, they in fact provide a good opportunity to assess the role of government structure in limiting the fatalities resulting
from such events: death tolls provide a clear and
consistent time varying cross-country measure of
effectiveness.
We have also estimated these regressions using the
pooled Tobit7 econometric technique.8 Results from
this alternative specification yields qualitatively similar results. We also estimated a series of regression
using the fixed effects estimation procedure.
However, a word of caution in interpreting these
findings is in order. The parameter estimates from
the fixed effects estimates are generate from the
within country variation in the independent variables. It therefore seems prudent to identify which
countries experienced significant changes in the
ratio of subnational to national spending over the
period of analysis. While many countries experience
relatively minor changes in the degree of decentralization, we identified eight countries (Argentina,
Bolivia, Brazil, Italy, Mexico, Spain, Peru and South
Africa) that had both a significant number of recorded disaster events as well as significant changes over
time in fiscal federalism. The fixed effects parameter
estimate on the decentralization variable is largely
driven by the experiences of these eight countries.
Thus, one should interpret the fixed effects results
with this in mind. The coefficient on our key decentralization variable is again negative and highly significant: countries that have become more decentralized over time have experienced fewer disasterinduced deaths.9
Our analysis provides evidence that more decentralized countries as measured by the ratio of subnational
expenditures to national expenditures experience
fewer deaths. Our findings also suggest that one route
by which a decentralized system reduces disasterinduced deaths is through educational attainment.This
finding is consistent with the work of Brueckner
(2006), and Arze del Granado, Martinez-Vasquez and
McNab (2005) who provide arguments and evidence
for the notion that decentralized systems increase
human capital. More generally, these findings are
robust to the inclusion (or exclusion) of a variety of
other factors that have been found to be important
determinants of disaster-related deaths in previous
studies. The results are also robust to alternative
econometric methods. The research presented here
increases our understanding of the role of government
in disaster management and response. Disaster death
toll data has enabled us to conduct the first time-varying cross-country evaluation of the effectiveness of
decentralization efforts, and therefore this research
contributes to both the literatures on the fiscal decentralization and the economics of natural disasters.
Conclusion
Private entities, governments and not-for-profit organizations engage in a variety of actions to reduce the
impacts of natural disasters. For example, in areas
where seismic activity is present building codes (and
compliance) are likely to be more stringent. In hurricane prone areas, certain measures may be undertaken to protect life and property (forecasting, warning
systems, planning, building codes, etc.). Further, the
public sector preparations and response to natural
disasters are critical. Lack of proper preparation and
delayed and/or ineffective response may result in lives
lost. The primary objective of this study is to determine whether government fiscal structure is important in disaster-induced fatality prevention.
References
Akai, N. and M. Sakata (2002), “Fiscal Decentralization Contributes
of Evidence from State-Level Cross-Section Data for the United
States”, Journal of Urban Economics 52, 93–108.
Albala-Bertrand, J. (1993), Political Economy of Large Natural
Disasters, New York: Oxford University Press.
Alderman, H. (1998), Social Assistance in Albania: Decentralization
and Targeted Transfers, LSMS Working Paper 134, World Bank.
Alexander, D. (1993), Natural Disasters, New York: Chapman and
Hall.
Anbarci, N., M. Escalaras and C. A. Register (2005), “Earthquake
Fatalities: The Interaction of Nature and Political Economy”,
Journal of Public Economics 89, 1907–1933.
7 A likelihood ratio test indicates that the panel-level variance component is important, and therefore the random effects estimator is
more appropriate than that pooled Tobit estimator. Nevertheless,
we examine the pooled Tobit estimator.
8 The number of recorded fatalities can be thought of count data
with over-dispersion. The negative binomial is therefore potentially an appropriate alternative estimation technique.
9 These additional estimates are available from the authors upon
request.
CESifo Forum 2/2010
Arzaghi, M. and V. Henderson (2005), “Why Countries Are Fiscally
Decentralizing”, Journal of Public Economics 89, 1157–1189.
Arze del Granado, F., J. Martinez-Vasquez, and R. McNab (2005),
Decentralization and the Composition of Public Expenditures,
52
Focus
Working Paper 05-10, International Studies Program, Georgia State
University.
Azfar, O., S. Kähnkönen and P. Meagher (2000), Conditions for
Effective Decentralized Governance: A Synthesis of Research
Findings, IRIS Center Working Paper, University of Maryland.
Barankay, I. and B Lockwood (2007), “Decentralization and the
Productive Efficiency of Government: Evidence from Swiss
Cantons”, Journal of Public Economics 91, 1197–1218.
Barro, R. and J. Lee (1996), “International Measures of Schooling
Years and Schooling Quality”, American Economic Review 86,
218–223.
Bardham, P. (2002), “Decentralization of Governance and
Development”, Journal of Economic Perspectives 16, 185–205.
Brueckner, J. (2006), “Fiscal Federalism and Economic Growth”,
Journal of Public Economics 90, 2107–2120.
Burton, K., R. Kates and G. White (1993), The Environment as
Hazard, 2nd edition, New York: Guilford Press.
Chernick, H. and A. F. Haughwout (2006), Economic Resilience,
Fiscal Resilience, and Federalism: Evidence from 9–11, mimeo.
Davoodi, H. and H. Zou (1998), “Fiscal Decentralization and
Economic Growth: A Cross-Country Study”, Journal of Urban
Economics 43, 244–257.
EMDAT (2004), The OFDA/CRED International Disaster
Database, Unversite Catholicque de Louvain, Brussels,
www.md.ucl.ac.be/cred.
Heston, A., R. Summers and B. Aten (2002), Penn World Table
Version 6.1, Center for International Comparisons at the University
of Pennsylvania (CICUP), http://pwt.econ.upenn.edu/.
Horwich, G. (2000), “Economic Lessons from the Kobe
Earthquake”, Economic Development and Cultural Change 48,
521–542.
International Monetary Fund (IMF), International Financial
Statistics Online, http://ifs.apdi.net/imf/.
Appendix: Tables A1 and A2
International Monetary Fund (IMF), Government Finance Statistics
Annual Yearbook, various years, Washington DC.
Kahn, M. (2005), “The Death Toll from Natural Disasters: The Role
of Income, Geography, and Institutions”, Review of Economics and
Statistics 87, 271–284.
Kellenberg, D. and A. Mobarak (2008), “Does Rising Income
Increase or Decrease Damage Risk from Natural Disasters?”,
Journal of Urban Economics 63, 788–802.
Santos, Boaventura de Sousa (1998), “Participatory Budgeting in
Porto Alegre: Toward a Redistributive Democracy”, Politics and
Society 26, 461–510.
Skidmore, M. (2001), “Risk, Natural Disasters, and Household
Savings in a Life Cycle Model”, Japan and the World Economy 13,
15–34.
Thornton, J. (2006), “Fiscal Decentralization and Economic Growth
Reconsidered”, Journal of Urban Economics 61, 64–70.
Tol, R. and F Leek (1999), “Economic Analysis of Natural
Disasters”, in: Downing, T., A. Olsthoorn and R. Tol (eds.), Climate
Change and Risk, London: Routledge, 308–327.
Toya, H. and M. Skidmore (2007), “Economic Development and the
Effects of Natural Disasters”, Economics Letters 94, 20–25.
Wildavsky, A. (1988), Searching for Safety, New Brunswick, NJ:
Transaction Books.
Wildasin, D. (2008), “Disaster Policies: Some Implications for Public
Finance in the U.S. Federation”, Public Finance Review 36, 497–518.
World Bank (2003), World Development Indicators 2003,
http://go.worldbank.org/BDEXK5OE00.
Xie, D., H. Zaou and H. Davoodi (1999), “Fiscal Decentralization
and Economic Growth in the United States”, Journal of Urban
Economics 45, 228–239.
53
CESifo Forum 2/2010
Focus
Appendix
Table A1
Summary of statistics variables used in the analysis
Mean
2.248
Standard
deviation
1.985
Number
of observations
2761
8.940
2.742
1.003
1.822
2761
2761
3.601
2.792
37.65
-1.218
0.678
1.250
9.446
0.417
2761
2761
2761
2761
Local Gov total / Total Gov total
Vertical imbalance
(Local Gov total / Total Gov total)
Secondary and higher
schooling years
0.315
0.405
1.031
0.169
0.184
0.998
2712
2712
2581
Log(Population)
Log(Area)
OECD
Earthquake
Slides
Volcano
Wave
Wind
Flood
Political right
Civil liberty
11.49
14.32
0.376
0.106
0.071
0.024
0.005
0.408
0.386
2.676
2.916
1.671
1.711
0.484
0.309
0.257
0.154
0.068
0.492
0.487
2.076
1.857
2761
2761
2761
2761
2761
2761
2761
2761
2761
2527
2527
Log (1+Number of killed)
Log (GDP per capita)
Secondary and higher
schooling years
Log (Openness)
Fertility
Gini coefficients
Log (Size of government)
Source: Authors’ calculation.
CESifo Forum 2/2010
54
Focus
Table A2
Definitions and sources of variables
Variables
Number of killed
GDP per capita
Secondary and higher
schooling years
Openness
Fertility
Gini coefficients
Size of government
Local Gov total / Total Gov total
Vertical imbalance
Population
Area
OECD
Wind
Flood
Earthquake
Slides
Volcano
Wave
Political right
Civil liberty
Definition
The number of persons confirmed as dead and persons missing
and presumed dead
Real GDP per capita
Years of secondary and higher schooling in the total population
aged 15 and over
Ratio of exports plus imports to GDP
Total fertility rate
Gini coefficients
The ratio of total government expenditures to GDP
The ratio of total sub-national government expenditures to total
government expenditures
The ratio of intergovernmental transfers to sub-national
expenditures
Logarithm of population
Logarithm of land area
Dummy for OECD countries
Dummy for wind
Dummy for flood
Dummy for earthquake
Dummy for slides
Dummy for volcano eruption
Dummy for wave
Political right (range from 1(good) – 7(bad))
Civil Liberty (range from 1(good) – 7(bad))
Source
EMDAT
HSA
BL
HSA
WDI
WIID
GDN
WB_FDI
WB_FDI
HSA
WDI
EMDAT
EMDAT
EMDAT
EMDAT
EMDAT
EMDAT
FH
FH
Sources: BL = Barro and Lee (1996); EMDAT = EMDAT (2004); FH = Freedom in the World from
http://www.freedomhouse.org/template.cfm?page=15; GDN = Global Development Network Growth Database
from http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/0,,contentMDK:20701055~
pagePK:64214825~piPK:64214943~theSitePK:469382,00.html; HAS = Heston, Summer and Aten (2002);
WB_FDI = Fiscal Decentralization Indicators from http://www1.worldbank.org/publicsector/decentralization/
fiscalindicators.htm; WDI = World Development Indicators 2006 from http://data.worldbank.org/datacatalog/world-development-indicators/wdi-2006; WIID = World Income Inequality Database from http://www.
wider.unu.edu/research/Database/en_GB/wiid/.
55
CESifo Forum 2/2010
Focus
Albala-Bertrand (2007) claimed that economic
impact of a disaster, which causes localized damages and losses on capital and activities, may not
affect negatively the macro-economy in both short
and longer term. This appears to contradict with
some empirical observations, such as the 1999 ChiChi earthquake in Taiwan which caused a hike in
price of computer memory chips in the United
States and other countries, and the 2005 hurricane
Katrina which led to the increase in oil price
domestically and internationally. These observations indicate that while the degree of damages and
losses is much severer in the areas hit by such a natural hazard, the impacts of the event appear to
spread over many other areas and nations. In this
regard, the propagation process of disaster impact
in a global sense is examined in this paper using the
empirical case of the 2004 Indian Ocean earthquake and tsunami.
GLOBALIZATION AND
LOCALIZATION OF DISASTER
IMPACTS: AN EMPIRICAL
EXAMINATION
YASUHIDE OKUYAMA*
Introduction
The damages and losses brought by disasters, such
as earthquakes, floods, hurricanes, cyclones and so
on, can potentially have significant and intense
impacts on a nation’s economy. However, despite
the importance of assessing the economic impacts of
damages and losses in the aftermath of such events,
estimating the impacts is rather challenging. The
consequences associated with the event will have
many other aspects including damages on demand
and supply sides, for example, since the event may
affect a wide range of economic activities in many
different ways. The difficulties with impact analysis
of disasters are, therefore, (1) disentangling the consequences stemming directly and indirectly from the
event, (2) deriving possibly different assessments at
each spatial level – cities, region, or nation –
(Hewings and Mahidhara 1996), and (3) evaluating
the reaction of households which are poorly understood (West and Lenze 1994). Data availability for
the impact assessment is another important issue.
West and Lenze (1994) claim that sophisticated economic impact models requiring precise numerical
input have to be reconciled with imperfect measurements of the damages. They proposed a systematic
way to estimate the impacts from the available data;
however, “impact assessment of unscheduled events
is an inexact science” (Hewings and Mahidhara
1996, 216).
In the following section, Albala-Bertrand’s ‘Globalization and Localization: An Economic Approach
(2007)’ is reviewed and analyzed. The third section
defines and describes terminology associated with
economic impact assessment of natural disaster.
Analysis of empirical case study based on the 2004
Indian Ocean earthquake and tsunami is carried out
and discussed. The final section concludes the paper
with some remarks on future directions for this line
of research.
Review of globalization and localization
Albala-Bertrand has been actively studying about
the economic consequences of natural disasters,
especially in the developing country’s context. In his
studies, he often claims that economic impacts of
natural disasters are rather minor in a macroeconomic sense, even with a catastrophic one. For example, he claimed that the indirect effects of disaster
are “more a possibility than a reality” (AlbalaBertrand 1993, 104). He also argued that in a long
run the negative impacts from damages made by a
disaster and the positive impacts from recovery and
reconstruction may potentially cancel out and then
* University of Kitakyushu and Tokyo Institute of Technology. This
research was supported by the JSPS Grant-in-Aid for Scientific
Research (C) [21510199] and partly by the Global Facility for
Disaster Reduction and Recovery (GFDRR) and the World Bank.
The author would like to express gratitude to Apurva Sanghi and
Sebnem Sahin for their support, encouragement and insights, and
to Hirokazu Tatano of the Kyoto University for his helpful comments.
CESifo Forum 2/2010
56
Focus
disequilibrium between negative and positive
impacts in various economic spaces and agents, is
necessary and essential on the way to display the
macroeconomic impacts.
the estimation of the total impacts often ends up
deriving insignificant values (Albala-Bertrand 1993).
While his arguments were based on some empirical
evidence of the past disasters (in 1960s and 1970s for
his 1993 publication), with the recent progress on
globalization and increased interdependency
between economies, economic impact of disasters
should be examined with the augmented complexity
of recent economy.
Economic impacts of disasters: concept and
definition
In order to discuss the economic impacts of disasters,
we need to clarify the terminology first, since the use
of similar words has created some confusion in many
disaster literatures. According to Okuyama and
Chang (2004, 2), “hazard is the occurrence of the
physical event per se, and disaster is its consequence”. In this context, while the occurrence of hazards cannot be prevented, the extent and intensity of
a disaster can be managed. Hence, the measuring the
extent and intensity of economic consequences (disaster) caused by a hazard is necessary to evaluate
and determine the countermeasures against hazards
and is central to understand how the consequences
of a hazard become a disaster.
In his recent publication (Albala-Bertrand 2007),
Albala-Bertrand analyzed the effect of globalization
on disaster impacts from an economic perspective.
His main conclusions are:
1. Disaster impacts, such as casualties and economic losses, will be economically localized and thus
are unlikely to influence negatively the national
economy, even in a long run.
2. Positive features of globalization, like access to
larger markets and suppliers, etc., may lead to
even more localization of disaster impacts, while
negative features of globalization, which are fast
efficiency and productivity improvements
through privatization and deregulations and lead
to thinner and weaker social fabric against emergency situations, make localized disaster impact
much more condensed into the local community
than before.
3. The synchronization of business cycle caused by
globalization, especially with the US economy,
may regulate the financial capability for disaster
response of the world, especially when the leading economies are under recession.
In terms of ‘disaster’ economic impacts, many comparable terms, such as damages and losses which
are further differentiated between direct and indirect losses, have been employed interchangeably
without making any distinction or definition of
them, and have led to further perplexity. Oftentimes, the direct loss refers to the damages on stock
like buildings, roads, houses, etc., while the indirect
loss implies the loss of flow due to business disruptions caused by stock damages. And then, the total
loss is often calculated by adding these direct and
indirect losses. However, in economics term, stock
and flow are two different things and summing
these up leads to potential double counting (Rose
2004). Also, in the above way, the distinction
between flow losses caused directly by the stock
loss and flow losses caused via inter-industry linkage (often referred as ‘ripple’ effect) cannot be
made, and this distinction is vital to illustrate the
extent of disaster impact.
These conclusions reflect his above arguments on
disaster impact, while he acknowledges the uncertainty of disaster consequences, shown in 2 and 3,
may increase due to globalization.
In a macroeconomic or an aggregated sense, his
arguments appear plausible for not such significant
total impacts. However, the economic structure can
affect the extent and significance of disaster impacts
in different parts of an economy, geographically
and/or socially, and the distribution and volume of
negative and positive impacts may differ over space
and for different sectors. In this line, AlbalaBertrand’s claim, which urges for future studies to
start classifying disaster impacts over localities, is
imperative. This may contradict his claim of negligible total impact of disasters in an aggregate sense,
but the disaggregation of disaster impacts, showing
Consequently, the clear definition of disaster
impact should be made. Okuyama and Sahin
(2009) proposed the following terminology for disaster economic impacts: damages are by economics definition the damages on stocks, which include
physical and human capitals; losses are business
interruptions, such as production and/or consumption, caused by damages and can be considered as
57
CESifo Forum 2/2010
Focus
first-order losses; higher-order effects, which take
into account the system-wide impact based on
first-order losses through inter-industry relationships; and total impacts are the total of flow
impacts, adding losses (first-order losses) and
higher-order effects. Rose (2004) further suggested that listing both damages and losses, but not
adding them together, is appropriate for showing
the different aspects of economic impact. In the
following sections, these terms are used for the
analysis.
The 2004 Indian Ocean disaster
The December 2004 Indian Ocean disaster was
caused by an earthquake, and the earthquake generated a tsunami, carrying many million tons of water
in a series of very large waves that traversed the
Indian Ocean in a matter of hours. These waves hit
beaches, flooding low-lying lands coastal areas. The
destruction was widespread: the most seriously
affected areas were Banda Aceh, Indonesia, as well
as in tourism resorts in Thailand, Sri Lanka, and the
Maldives. Many small and medium sized rural villages located along the beachside in the five countries were also wiped out (ADPC 2005).
What we are going to estimate in this paper is the
economic intensity of a natural hazard on flow,
while a comprehensive assessment of a natural disaster requires to include both negative impact of a
natural hazard and positive effects of recovery and
reconstruction activities. More concretely, the
results shown in the following section are only the
negative impact of a natural hazard over a year,
without any restoration, recovery, or reconstruction. This appears to be a very unlikely scenario, but
this serves as the worst-case scenario (do-nothingscenario)1 and also provides the extent to which
recovery and reconstruction need to be done. In
addition, those restoration, recovery and reconstruction strategies will be decided based on the
total impacts of a disaster and the distribution of
them; thus, the estimation of negative impact only
becomes a basis of decision making and is well
worth doing.
According to the preliminary assessment of damages
and losses, total of 281,900 persons died as a result of
the earthquake and tsunami; 189,500 persons were
injured, physically and psychologically, and required
immediate or medium-term treatment; and, 1.2 million persons became homeless and even a year after
the tsunami many were still housed in temporary
camps, a sizable fraction of which still requires shelter, food and health services. The total economic
effects of this event were estimated as USD 5.6 billion of damages and 4.3 billion of losses over five
countries – Indonesia, India, Sri Lanka, Maldives,
and Thailand (ADPC 2005). In this paper, the total
impacts of this event are estimated and analyzed for
Indonesia and Thailand using the 2000 Asian
International IO Table, since other three countries
(India, Maldives and Sri Lanka) was not included in
the IO Table.
Empirical examination: 2004 Indian Ocean
earthquake and tsunami
Here are some nation-specific information on the
damage and loss for Indonesia and Thailand. The
total damage and loss in Indonesia were estimated
as 2,664 and 1,136 million US dollars, respectively
(ADPC 2005). The housing sector had the largest
damage with 1,398 million US dollars (52 percent
of total damage). The transport sector had the second largest damage, 409 million US dollars. The
productive sector, especially agriculture and industry (manufacturing), also had some sizable damages. On the other hand, the losses were concentrated on these productive sectors, 550 million US
dollars for agriculture and 280 million US dollars
for industry, and together, they had about 73 percent of total loss.
This section examines how economic impacts of a
local disaster can (or cannot) spread over internationally, employing the 2004 Indian Ocean earthquake and tsunami as the case study. While this event
was a multi-country incident, involving at least five
countries (India, Indonesia, Maldives, Sri Lanka and
Thailand), the damaged areas in each country were
relatively limited geographically. Thus, using AlbalaBertrand’s term, this was a localized event for each
country. The economic impacts of this event are evaluated using the 2000 Asian International InputOutput Table for analyzing whether or not any sizable economic impacts were propagated over other
countries, i.e. globally.
The total damages and losses in Thailand were estimated to reach 509 and 1,690 million US dollars,
respectively. The damages were concentrated on
1
There would potentially be some worse scenarios than this, if the
recovery and reconstruction activities were misguided to create
further negative influence.
CESifo Forum 2/2010
58
Focus
Solving this relationship for x yields:
tourism with 376 million US dollars (74 percent of
the total damage), resulted from the washed out
resorts and hotels on the beaches. Other noticeable
damages were on agriculture. The losses were also
mostly on tourism with 1,470 million US dollars
(87 percent of the total loss), and agriculture and
industry had some losses around 100 million US dollars each.
x = (I - A ) f
-1
(I-A)-1 is the Leontief inverse matrix. For the impact analysis, the impact of changes in final demand
can produce the changes in output in the following
manner:
x = (I - A ) f
-1
Methodology
x A C x f y = V 0 y + g x B (I + CKVB ) BCK f y = KVB
K g (1)
j
B=(I-A)-1 is the Leontief inverse matrix;
BC is a matrix of production induced by endogenous
consumption;
VB is a matrix of endogenous income earned from
production;
L=VBC is a matrix of expenditures from endogenous income; and
K=(I-L)-1 is a matrix of the Miyazawa interrelational income multipliers.
In this paper, the IO model used is transformed to
the Miyazawa’s extended IO framework for the
analysis of impact on income generation.
(2)
In the matrix notation, (2) becomes:
x = Ax + f
(7)
where:
where xi is the output of sector i, xij is intermediate
demand from sectors j to i, and fi is the final demand
for sector i. Direct input coefficient, aij, is calculated
by aij = xij /xj, and equation (1) can be transformed as
follows:
xi = aij x j + fi
(6)
where x is a vector of output, y is a vector of total
income for some r-fold division of income groups, A
is a block matrix of direct input coefficients, V is a
matrix of value-added ratios for r-fold income
groups, C is a corresponding matrix of consumption
coefficients, f is a vector of final demands except
households consumption, and g is a vector of exogenous income for r-fold income groups. Solving this
system yields:
Input-output (IO) framework was developed by
Wassily Leontief in the late 1920s and early 1930s.
The structure of IO mimics the double-entry style of
bookkeeping scheme. For the production side, the
output is determined as the sum of intermediate
demand and final demand as follows:
j
(5)
Miyazawa’s (1976) extended input-output analysis
intends to analyze the structure of income distribution by endogenizing consumption demands in the
standard Leontief model. In some sense,
Miyazawa’s system is considered the most parsimonious in terms of the way it extends the familiar
input-output formulation. Miyazawa considered the
following system:
There is a wide range of methodologies adopted to
estimate the higher-order effects, and thus the total
impacts of disasters (further detailed discussion of
methodologies for impact estimation can be found at
Rose (2004), Okuyama (2007) and Greenberg et al.
(2007)). Input-Output (IO) model has been the most
widely used methodology for disaster impact estimate for the recent decades (for example, Cochrane
1997; Gordon and Richardson 1996; Rose et al. 1997;
Okuyama et al. 1999; Hallegatte 2008). The popularity of IO models for disaster related research is
based mainly on the ability to reflect the economic
interdependencies within an economy in detail for
deriving higher-order effects, and partly on its simplicity. On the other hand, this simplicity of the IO
model creates a set of weaknesses, including its linearity, and rigid structure with respect to input and
import substitutions as well as a lack of explicit
resource constraints, and responses to price changes
(Rose 2004).
xi = xij + fi
(4)
(3)
59
CESifo Forum 2/2010
Focus
In this decomposition, M1 captures intraregional (or
domestic) effects, M2 contains interregional (or
international) spillover (a.k.a. open-loop) effects,
and M3 records interregional (or international)
feedback (a.k.a. closed-loop) effects. This multiplicative decomposition can be transformed into an additive decomposition (Stone 1985), through isolating
the net effects as follows:
In order to analyze the interregional (or international) spillovers of a particular impact, the method of
multiplier decomposition2 is employed. Suppose we
have a two-region (or two-nation) system, consisting
of regions r and s. The above equation (6) can be
rewritten as follows:3
xr A rr
s sr
x = A
y r V rr
s sr
y V
A rs
Crr
A
ss
sr
V
rs
0
V
ss
0
C
Crs xr f r Css xs f s +
0 y r g r 0 y s gs (8)
M 3 M 2 M1 = I + (M1 - I ) + (M 2 - I )M1 +
(M 3 - I )M 2M1
We can isolate the intraregional and interregional
elements of the coefficient matrix in the following
manner:
A rr
sr
A
V rr
sr
V
A rs
A ss
V rs
V ss
Crr
Csr
0
0
Crs Css =
0 0 A rr
0
= rr
V
0
0
A ss
0
V ss
Crr
0
0
0
0 0
Css A sr
+
0 0
0 V sr
In the above additive form, (M1-I) indicates the net
intraregional effects, (M2-I)M1 shows the net interregional spillover effects, and (M3-I)M2M1 captures the
net interregional feedback effects. Analyzing these
decomposed multiplier may reveal to what extent
the higher-order effects can propagate intraregionally and interregionally.
(9)
A rs
0
V rs
0
0
Csr
0
0
The empirical model used in this paper is the 2000
Asian International Input-Output Table published
by the Institute of Developing Economies (IDE),
the Japan External Trade Organization (JETRO).
This table includes nine countries and one region
(Indonesia, Malaysia, Philippines, Singapore,
Thailand, China, Taiwan, Korea, Japan, and the
United States) with seven industrial sectors (agriculture, mining, manufacturing, utility, construction,
trade and transport and services). This is the only
officially available data source for this type of international input-output table, while the data year 2000,
is a bit earlier than the year when the event occurred
(the end of 2004), implying that some of the international trade relationships as well as the domestic
economic structures may have changed between
these years, especially with China. In addition, since
the original international input-output table is a multiregional table, in which the international final
demand transactions and wage transactions are not
separated from the domestic final demand and wage
values, when transformed to the Miyazawa structure,
the consumption coefficient matrix C and the value
added coefficient matrix V are converted to blockdiagonal matrices, having zeros in interregional
blocks.
Crs 0 0 0 Then, the multiplier matrix in equation (7) can be
decomposed in the following multiplicative form:
B (I + CKVB ) BCK = M 3M 2M1
KVB
K (10)
where:
M1 = I - A
-1
-1
2
; M 2 = I + A* ; and M 3 = I - (A* ) ;
while
A
ª A rr
«
«0
« V rr
«
¬0
Crr
0
A
ss
0
V
0
0
ss
0
0 º
»
Css »
and A* = I - A
0 »
»
0 ¼
A - A .
-1
2 The detailed illustration of multiplier decomposition is articulated in Millar and Blair (2009).
3 For more precise formulation based on the Isard’s fully specified
interregional model, the final demand vector in this formula should
have the interregional components and thus should become a
matrix. Because the damages brought by a hazard will not have
interregional damages, this formulation, a multiregional model, is
sufficient in this setting. After all, damages caused by a hazard are
indeed very much localized.
CESifo Forum 2/2010
(11)
The sectors in the original model are aggregated as
much as possible to fit with the data of damages and
losses in order to maintain the detail and reliability
of the input data. The IO model is a demand driven
60
Focus
Table 1
111
166
186
167
Damages
1,389
409
68
27
132
1,136
9
18
550
280
Losses
39
148
0
3
89
Data
2,664
Analysis
Sectors in model
Agriculture
Mining
Manufacturing
Utilities
Construction
Trade & transport
Services
HH income
decrease
The economic impacts of the 2004 Indian
Ocean earthquake and tsunami for Indonesia
and Thailand, and other Asian countries are calculated and evaluated in this subsection. Table 1
shows the input data, damages and losses, and
the results, output impact and income impact in
Indonesia. The derived total impacts amount to
2,386 million US dollars (0.93 percent of 2004
GDP) for output and 1,219 million for income.
The most significant output impact falls on
manufacturing with 814 million US dollars
(with 280 million of output decrease as loss),
followed by agriculture with 672 million. The
sectors with large impact tend to be accompanied with large losses, while the other sectors
with small or no losses, such as mining, utilities,
and construction, have limited higher-order
effects. This may lead to the relatively small
impact multiplier of 2.10.
Converted
Output decrease
550
0
280
3
0
148
116
39
1,136
Demand decrease
410
0
158
3
0
113
80
1,219
61
Calculated
Output impact
Income impact
672
69
814
30
20
370
412
1,219
2,386
The derived impact for Thailand on output and
income are 3,205 million US dollars (1.99 percent of 2004 GDP) and 1,240 million, respectively, as seen in Table 2. The total impacts fall mostly on services (including tourism industry) with
1,535 million (48 percent of the total output
impact). Meanwhile, manufacturing has a sizable
impact of 872 million US dollars (27 percent of
the total output impact), indicating that
Thailand’s domestic industries are highly interwoven and interdependent so that the total
impacts spread across the sectors. However, the
calculated impact multiplier is 1.90 – a relatively
low value. This implies that while the tourism
industry is one of the major industries in
Economic impacts of 2004 Indian Ocean earthquake and tsunami – Indonesia (in 2007 million US dollars)
Sector
Infrastructure
Housing
Transport
Electricity
Water & sanitation
Urban & municipal
Water resources
Social
Health & nutrition
Education
Production
Agriculture
Industry
Service
Tourism
Total
Source: Author’s calculation.
model so that the input to model should be the
form of changes in final demand, and then
changes in output will be derived. Therefore,
losses (decreased output level) are converted to
final demand change in each sector, using Miller
and Blair’s (2009) method – dividing the changes
in output (output loss) by the diagonal term of
the Leontief inverse matrix for IO model. Then,
the derived changes in final demand model are
multiplied with Leontief inverse matrix to calculate impact by sector. Because of extension to
the Miyazawa framework, the model can yield
both the output impact (higher-order effects)
and the impact on income generation (income
impact) as the results.
CESifo Forum 2/2010
Focus
Table 2
Damages
22
7
4
1
15
3
Losses
0
9
10
3
Sectors in model
Agriculture
Mining
Manufacturing
Utilities
Construction
Trade & transport
Services
HH income
decrease
Converted
Output decrease
102
0
93
13
0
9
1,473
0
1,690
Demand decrease
89
0
58
10
0
7
946
1,240
62
Calculated
Output impact
Income impact
228
33
872
132
3
401
1,535
1,240
3,205
So, does this mean that this type of localized but
multi-country disaster can have any global economic impact? At this stage of the analysis the
answer would be probably ‘No’, since the high-
Economic impacts of 2004 Indian Ocean earthquake and tsunami – Thailand (in 2007 million US dollars)
9
102
93
Data
75
1,470
1,690
Table 3 indicates the impacts for these countries.
Except those directly affected countries
Indonesia and Thailand, Japan receives the
largest total impacts (thus the largest higherorder effects, since there are no first-order losses
in Japan) in this system, with 428 million US dollars. The United States has the second largest
total impacts of 306 million. China follows these
two countries and has USD 156 million of the
total impacts. Among the sectors, manufacturing
has the most significant impact in total
(2,307 million US dollars) and for each country
in this system. This also is an evidence of increasing interdependence among manufacturing firms
through international trades a la vertical specialization. Compared to the total impacts in
Indonesia and Thailand and to their own GDPs,
these impacts in the other countries can be considered as negligible. At the same time, for the
system as a whole, the aggregated total impacts
reach 6,761 million US dollars with the impact
multiplier of 2.39, and these numbers are noticeably larger than the above two countries’. For the
multi-country disaster case such as this Indian
Ocean earthquake and tsunami, this type of
international analysis is useful to capture the
comprehensive picture of the impacts.
CESifo Forum 2/2010
376
509
As seen above, the impacts of the event appear
not so large within the two countries (0.93 percent of GDP in Indonesia and 1.99 percent of
GDP in Thailand). With increased economic
interdependency between countries through
international trades, this simultaneous damages
and losses in multiple neighboring countries
may bring the higher-order effects to other surrounding countries or even globally. As
described in the previous subsection, the model
used for this particular case (2000 Asian
International IO Table) includes the above two
countries, seven other Asian countries, and the
United States so that the impacts to those countries can be estimated.
Sector
Infrastructure
Housing
Transport
Electricity
Water & sanitation
Urban & municipal
Water resources
Social
Health & nutrition
Education
Production
Agriculture
Industry
Service
Tourism
Total
Source: Author’s calculation.
Thailand, the losses are concentrated on one
industry (tourism) and thus the total impacts are
somehow limited and not widely spread to the
entire economy.
Focus
Table 3
Malaysia
2
5
36
1
0
5
9
58
22
Philippines
1
0
7
1
0
2
2
14
5
Singapore
0
0
33
1
0
7
9
50
12
14
156
39
China
19
7
96
6
1
14
15
69
24
Taiwan
2
0
42
1
1
9
19
90
26
Korea
3
0
59
2
0
7
110
428
154
Japan
8
1
230
11
4
64
114
306
143
USA
13
4
120
7
2
47
2,239
6,761
2,855
Total
948
118
2,307
192
30
926
63
Thailand
228
33
872
132
3
401
1,535
3,205
1,240
Now, the decomposition of total impact to intraregional (domestic: Indonesia, and Thailand) effects,
interregional (international) spillover effects and
interregional (international) feedback effects, based
on the multiplier decomposition technique described
in the previous subsection, is summarized in Table 4.
With the gross multipliers, the domestic effect in
both Indonesia and Thailand becomes around
81 percent of the total impact, while international
spillover effect and international feedback effect are
estimated to be around 18 and 0.4 percent, respectively. When using the net multipliers, subtracting initial demand decrease, these figures change to 77,
23 and 0.5 percent, respectively. Since the initial
demand decrease only falls on to the countries
where the event occurred, the net percentages
become larger than the gross numbers. In either way,
the non-domestic effects, summing international
spillover effect and international feedback effect, are
quite significant, being the range around one fifth of
the total impact. This is not a negligible share.
Indeed, looking at the actual values, the sum of international spillover effect and international feedback
effect and the initial demand decrease are nearly
equal, 1.85 and 1.92 billion US dollars, respectively.
This is a striking result indicating that the size of initial demand decrease, which is converted from the
value of losses, is transmitted to and is replicated in
the rest of system. At the same time, the distribution
of non-domestic effects across sectors is quite different from the one for the initial demand decrease.
While service sector in Thailand is the largest
demand decrease, followed by agriculture sector and
manufacturing sector in Indonesia, the manufacturing sector in the rest of the system is the largest for
the non-domestic effects, and services sector and
trade and transport sector are the second and third.
This difference in the distribution may come from
the two factors: the one is the distribution of losses
which determines the origin of impact path and the
path itself – forward and backward linkages of inter-
Spatial distribution of total impacts of 2004 Indian Ocean earthquake and tsunami (in 2007 million US dollars)
Sectors in model
Indonesia
Output Agriculture
672
impact Mining
69
Manufacturing
814
Utilities
30
Construction
20
Trade &
370
transport
Services
412
Total
2,388
Income impact
1,219
Source: Author’s calculation.
er-order effects to other surrounding countries are
very small in value. As described in the previous
subsection, the derived economic impacts do not
include the positive impacts from recovery and
reconstruction activities in the respective countries.
If included, the economic impacts to other surrounding countries may become much smaller than
the values in Table 3. Thus, we need to further analyze the derived total impact in detail through the
decomposition into domestic and non-domestic
(international) effects.
CESifo Forum 2/2010
Focus
Table 4
Indonesia
Thailand
Rest of
system
Total
Agriculture
Mining
Manufacturing
Utilities
Construction
Trade &
transport
Services
HH income
Agriculture
Mining
Manufacturing
Utilities
Construction
Trade &
transport
Services
HH income
Agriculture
Mining
Manufacturing
Utilities
Construction
Trade &
transport
Services
HH income
Source: Author’s calculation.
Total
impact
671
68
812
30
20
370
407
1,202
227
32
859
131
3
397
Domestic
effect
667
65
794
30
19
364
4
13
1
1
9
1
0
3
International
spillover effect
4
2
16
0
0
5
0.69
2.63
0.26
0.07
2.60
0.14
0.00
0.65
International
feedback effect
0.65
0.99
2.38
0.06
0.04
0.76
0.15
0.53
100
100
100
100
100
100
1.07
1.28
0.58
2.25
1.33
0.49
0.34
0.83
Gross %
0.63
4.33
2.24
1.45
1.20
1.53
0.40
0.53
100
100
100
100
100
100
1.33
1.33
0.94
2.25
1.42
0.53
0.34
0.84
Net %
1.62
4.33
2.77
1.59
1.20
2.20
0.23
16.89
9.05
no initial loss
282.96
253.78
no initial loss
144.61
0.43
40.08
0.26
no initial loss
5.36
4.98
no initial loss
2.78
Non-domestic
effect/Initial
demand decrease
(%)
0.84
no initial loss
8.43
3.43
no initial loss
4.71
Decomposition of total impact to domestic and international effects (in 2007 million US dollars)
Demand
decrease
410
0
158
3
0
113
411
1,218
228
33
871
132
3
401
0.48
1.34
0.58
0.30
7.59
0.41
0.10
2.15
Non-domestic effects
Output
decreases
(loss)
550
0
280
3
0
148
80
39
89
0
58
10
0
7
2
5
45
21
602
32
8
171
Decomposition of total impact
116
39
102
0
93
13
0
9
1,532
1,233
0
0
0
0
0
0
36.52
1301.86
1,535
1,239
45
21
610
33
8
173
100
100
23.42
946
0
0
0
0
0
0
0
100
100
18.85
1,473
0
0
0
0
0
0
0
4.68
6.35
36
0.37
0.45
0
0
1,912
370
501
1,814
18.49
22.96
0
0
2,826
0
0
7,963
81.15
76.58
375
508
9,813
Gross %
Net %
64
CESifo Forum 2/2010
Focus
ment above, has made such complexity and interdependency of economic activities in the world. More
thorough analysis using two (or more) different time
points with the same disaster situation, which may
reveal how the progress of globalization affects the
propagation of disaster impact, is called for. If proper development and domestic policy and appropriate
recovery and reconstruction strategies were not
practiced, these higher-order effects would spread
over globally.
industry relationship; and the other is the international trade relationships which lead to the extent of
international spillover effect and feedback effect. As
globalization is reaching every corner of the world,
these domestic and international inter-industry relationships have become so intertwined, and thus they
demand a careful inspection.
The effects on household income deserve some discussion. The initial household income decrease
amounts to just 39 million US dollars in Indonesia
due to housing damages, whereas the total impact
on household income is estimated as almost 3 billion US dollars, more than 75 times larger than the
initial one. The total impact on household income
includes wage decrease due to the decreased output,
caused not only by the damages and through interindustry production relationships but also by the
consumption decrease due to such wage declines,
leading to further ripple effect on the wage-consumption relationship. It is also remarkable that the
interregional spillover effect on household income
becomes around 500 million US dollars and this
effect creates further repercussions in the rest of
system, bringing further USD 6 million of household income decrease.
At the same time, all of the derived results above
include only negative impact of losses without having positive impact of recovery and reconstruction
activities; thus, they are after all potential total
impact. Perhaps, in the real event, the propagation of
negative impact to other countries was very minimal,
if at all, because the negative impact was cancelled
out with (or the propagation was avoided by) the
positive impact of recovery and reconstruction activities. However, this fact does not undermine the
results in this paper, since without such recovery and
reconstruction activities the results of this paper
would have been unfolded and materialized. And,
some parts of recovery and reconstruction activities
were aided and financed by the international organizations and international community. This implies
that if proper development and domestic policy to
install countermeasures for such hazards and appropriate recovery and reconstruction strategies were
not practiced, these higher-order effects would have
spread over the world. In other words, the localized
risk of natural hazard will be shared with, or extend
to global community. This is the true issue of globalization and localization of hazard risk.
Summary and conclusions
In this paper, the total impacts of the 2004 Indian
Ocean earthquake and tsunami were estimated
using the 2000 Asian International Input-Output
Table. The results show that the higher-order effects
and total impacts of disasters are significant and
complex domestically. The spread of higher-order
effects to other surrounding countries do exist, while
the value per se is relatively small compared to the
localized higher-order effects and to their respective
size of the economy. However, this does not mean
that there is little global or international ripple effect
of the disaster impact. Rather, with the multiplier
decomposition technique, the non-domestic effect
including international spillover effect and international feedback effect consists of around 20 percent
of total impacts, in both gross and net analyses. The
results also display that the distribution of higherorder effect across sectors is quite different from
that of initial losses, especially for the non-domestic
ones. This is caused by the increased complexity of
domestic economic structure as well as of international trade. The effect of globalization, especially
the first half of 2 and 3 in the Albala-Bertrand argu-
The data for damages and losses used as input for estimation in this paper are based mostly on the ECLAC
methodology (UN ECLAC 2003). While the accuracy
of these data is the key for the precision of the estimated results and in this regard the data collection
methodology needs to be streamlined further
(Greenberg et al. 2007), this ECLAC methodology
can standardize the assessment of damages and losses
of a disaster, and this standardization not only enables
inter-disaster comparison but also encourages the discussion of mitigation, preparedness against disasters
and vulnerability analysis of economies based on the
common framework. An important next step would
be to make the estimation methodology of higherorder effects a part of a standardized methodology –
such as the ECLAC methodology – evaluating a more
accurate measure of disaster impacts.
65
CESifo Forum 2/2010
Focus
References
Albala-Bertrand, J. M. (1993), The Political Economy of Large
Natural Disasters: With Special Reference to Developing Countries,
Oxford: Clarendon Press.
Albala-Bertrand, J. M. (2007), “Globalization and Localization: An
Economic Approach”, in: Rodriguez, H., E. L. Quarantelli, and R.
R. Dynes (eds.), Handbook of Disaster Research, New York: Springer, 147–167.
Asian Disaster Preparedness Center (ADPC, 2005), Regional
Analysis of Socio-Economic Impacts of the December 2004
Earthquake and Indian Ocean Tsunami, Bangkok.
Cochrane, H. C. (1997), “Forecasting the Economic Impact of a
Midwest Earthquake”, in: Jones, B. G. (ed.), Economic Consequences of Earthquakes: Preparing for the Unexpected, Buffalo, NY:
National Center for Earthquake Engineering Research, 223–247.
Gordon, P. and H. W. Richardson (1996), The Business Interruption
Effects of the Northridge Earthquake, Lusk Center Research
Institute, University of Southern California, Los Angeles.
Greenberg, M. R., M. Lahr and N. Mantell, (2007), “Understanding
the Economic Costs and Benefits of Catastrophes and Their
Aftermath: A Review and Suggestions for the U.S. Federal
Government”, Risk Analysis 27, 83–96.
Hallegatte, S. (2008), “An Adaptive Regional Input-Output Model
and Its Application to the Assessment of the Economic Cost of
Katrina”, Risk Analysis 28, 779–799.
Hewings, G. J. D. and R. Mahidhara (1996), “Economic Impacts:
Lost Income, Ripple Effects, and Recovery”, in: Changnon, S. A.
(ed.), The Great Flood of 1993: Causes, Impacts, and Responses,
Boulder, CO: Westview Press, 205–217.
Miller, R. E. and P. D. Blair (2009), Input-Output Analysis:
Foundations and Extensions, Second Edition, New York: Cambridge
University Press.
Miyazawa, K. (1976), Input-output Analysis and the Structure of
Income Distribution, New York: Springer.
Okuyama, Y. (2007), “Economic Modeling for Disaster Impact
Analysis: Past, Present, and Future”, Economic Systems Research 19, 115–124.
Okuyama, Y. and S. Sahin (2009), Impact Estimation of Disasters: A
Global Aggregate for 1960 to 2007, World Bank Policy Research
Working Paper 4963.
Okuyama, Y., G. J. D. Hewings and M. Sonis (1999), “Economic
Impacts of an Unscheduled, Disruptive Event: A Miyazawa
Multiplier Analysis”, in: Hewings, G. J. D., M. Sonis, M. Madden and
Y. Kimura (eds.), Understanding and Interpreting Economic
Structure, Berlin: Springer, 113–144.
Rose, A. (2004), “Economic Principles, Issues, and Research
Priorities in Hazard Loss Estimation”, in: Okuyama, Y. and S. E.
Chang (eds.), Modeling Spatial and Economic Impacts of Disasters,
New York: Springer, 13–36.
Rose, A., J. Benavides, S. E. Chang, P. Szczesniak and D. Lim (1997),
“The Regional Economic Impact of an Earthquake: Direct and
Indirect Effects of Electricity Lifeline Disruptions”, Journal of
Regional Science 37, 437–458.
Stone, S. R. (1985), “The Disaggregation of the Household Sector in
the National Account”, in: Payatte, G. and J. I. Round (eds.), Social
Accounting Matrices: A Basis for Planning, Washington DC: The
World Bank Group, 145–185.
United Nations Economic Commission for Latin America and the
Caribbean (UN ECLAC, 2003), Handbook for Estimating the
Socio-Economic and Environmental Effects of Disasters, Santiago,
Chile.
West, C. T. and D. G. Lenze (1994), “Modeling the Regional Impact
of Natural Disaster and Recovery: A General Framework and an
Application to Hurricane Andrew”, International Regional Science
Review 17, 121–150.
CESifo Forum 2/2010
66
Focus
A HURRICANE HITS HAWAII:
A TALE OF VULNERABILITY
TO NATURAL DISASTERS
ties to these disasters. We then assess the impact of
Iniki on the island of Kauai, and finish by describing
the current state of knowledge on the connection
between climate change and storms.
MAKENA COFFMAN AND
ILAN NOY*
The economics of natural disasters
The economy of the State of Hawaii, with its unique
set of vulnerabilities, will most likely be impacted
differently by large disaster events than significantly
bigger countries. Storms, unlike geo-physical disasters such as earthquakes and volcanic eruptions, are
fairly expected events with some advance warning,
even if their ferocity can be unpredictable. This level
of certainty, together with the resources that can be
mobilized both before and after a storm hits, also differentiate Hawaii’s likely experience with that of, for
example, Japan’s experience with the surprise of the
Kobe earthquake or hurricane Nargis’ impact on
Myanmar where little ex ante preparation occurred.
(Several papers identify the importance of institutional characteristics and the type of disaster in
determining the post-disaster outcomes in terms of
mortality, direct damages, and long-term economic
cost – see e.g. Skidmore and Toya 2002; Kahn 2004;
Anbarci et al. 2005; Noy and Nualsri 2007; Noy 2009;
Cavallo, Galiani, Noy and Pantano 2010.)
Introduction
Hurricane Iniki, which hit the island of Kauai on
11 September 1992, was the strongest hurricane that
hit the Hawaiian Islands in recorded history, and the
one that wrought the most damage, estimated at
USD 7.4 billion (in 2008 USD) though fortunately
mortality was low. Today, with the evident warming
of the planet and the changes in the patterns of climatic events that are predicted to accompany such
warming, the importance of understanding the longterm economic impact of natural disasters cannot be
overstated.
Research in both the social and natural sciences has
been devoted to increasing our ability to predict disasters and prepare for them; but curiously, there are
few analyses of their ex-post economic impact. We
were surprised to realize at the onset of this project,
for example, that no comprehensive or even cursory
attempt has been made to account for the long-term
impact of hurricane Iniki on the economy of the
island of Kauai; nor, as far as we know for any other
natural disaster that has previously impacted the
Hawaiian islands.
Two research projects have investigated the impact
of hurricanes on economies in the Caribbean. Given
similarities between island economies, these studies
are likely to provide more insight into the case of
Hawaii. Rasmussen (2004) conducts a tabulation of
the data for the island members of the Eastern
Caribbean Currency Union (ECCU). He finds that:
“among these […] the median number of affected
persons amounted to 9 percent of the country’s population and the median value of damage was equivalent to 14 percent of the country’s annual GDP”
(Rasmussen 2004, 7).1
Here, we would like to focus on Iniki’s impact on the
economy of Kauai. Observing Iniki’s impact is easier
both because of the availability of ex post, long-term
data and because of the existence of an obvious control group (the island of Maui). Because Hawaii’s
experience with disasters is not in any way unique,
we first describe the current state of the literature
that examines the economics of disasters in small
island (mostly nation-) states, and discuss a recent
attempt by the United Nations to assess vulnerabili-
1 Though, as Rasmussen (2004) points out, some events can be significantly worse. “For example, in 1979 hurricane David hit
Dominica killing 42 people, damaging 95 percent [of GDP] and
completely destroying 12 percent of buildings, damaging or
destroying the entire banana crop and 75 percent of the country’s
forests, rendering virtually the entire population homeless, and
leading to the temporary exodus of about a quarter of the population” (Rasmussen 2004, 7).
* University of Hawaii-Manoa.
67
CESifo Forum 2/2010
Focus
Because Hawaii is a State and not a country,
Baritto’s set of indicators do not necessarily translate to the case of Hawaii. Nonetheless, the general
themes of his vulnerability assessment remain:
(1) diversification of export products and destinations, (2) levels of wealth, (3) net food import ratio,
(4) manufacturing and services value-added proportion, and (5) characteristics of imports. We briefly
discuss each in turn.
Heger et al. (2008) also focus on Caribbean islands
(not limiting themselves to ECCU countries only).
Their results also do not agree with the earlier largely optimistic research which concluded that disasters
are typically followed by a period of higher growth.
They find that as growth collapses, the fiscal and
trade deficits both deteriorate and the island
economies of the region find it difficult to rebound
from the short-run impact of the disaster. They relate
this deepening recession to the reliance of island
economies in the region on very few sectors.
Regions with a large export sector will generally be
less vulnerable to disasters because the demand markets for products will presumably be unaffected. At
the same time, an economy with a less diversified set
of export products is generally more vulnerable to
external shocks. Hawaii has a very small manufacturing export base, with 2.4 percent of the value of
Hawaii’s USD 90.2 billion economy, roughly 2.2 billion US dollars. By far the largest export sector in the
state is tourism. Although tourism activities take
place within the State, goods and services are
‘exported’ to out-of-State consumers. Tourism, with
its heavy reliance on domestic infrastructure and
public perceptions, is different than other export sectors in terms of its vulnerability to disasters. Tourism
activity typically decreases significantly following
disasters, as was the case after hurricane Iniki (see
next section). Hawaii’s tourism sector accounts for
22 percent of Hawaii’s state gross product or seven
times the value of all other exported goods. As is
often the case, this sector is not even diversified in
terms of destinations. Nearly 70 percent of visitor
expenditures are from the continental United States.
Japanese visitors comprise 15.5 percent of spending
and Canadian visitors only 5 percent.
Vulnerabilities to disasters
Regions are vulnerable to disasters for many reasons.
Besides traditionally conceived geographical characteristics such as proximity to coastal areas, volcanoes
or tectonic fault lines, there is a growing understanding of economic conditions that make regions vulnerable as well. Economic circumstances may lead to
resiliency in the aftermath of a disaster or, alternatively, exacerbate its impacts. Vulnerability has been
defined in many ways. The United Nations
International Strategy for Disaster Reductions
describes it as the conditions determined by physical,
social, economic and environmental factors or
processes, which increase the susceptibility of a community to the impacts of hazards. Resiliency, the
antonym of vulnerability, is described as the capacity
of a system, community or society potentially
exposed to hazards to adapt by resisting or changing
in order to reach and maintain an acceptable level of
functioning and structure (United Nations 2010).
Vulnerability (or conversely resiliency) to hazards
can be categorized both ex-ante and ex-post. Ex-ante
vulnerability is described by indicators such as frequency and magnitude of disaster events, susceptibility of the economy to be impacted by external
shocks, and existence of disaster management plans
such as adequate shelters. Ex-post vulnerability is
characterized by the ability to recover from the
aftermath of a disaster. Relevant factors include success in the deployment of disaster management
plans, access to inflows of outside aid, and the ability
of different economic sectors to rebound. For the
purposes of this note, and as a useful example, we
briefly discuss the economic characteristics of vulnerability in Hawaii.
Per capita income in Hawaii in 2007 was 39,239 US
dollars, which is relatively high for an island economy. High income better enables reconstruction activity. Nonetheless, in terms of distribution, nearly
11 percent of the population is considered to be living under the poverty line. The poor are generally
much more vulnerable during disasters, as the events
following hurricane Katrina amply and painfully
demonstrate.
Baritto (2008) also suggests that regions with high
levels of food imports are more vulnerable to disasters, particularly depending on the infrastructure
required to bring in food. Over 85 percent of the
food consumed in Hawaii is imported through one
primary port (Leung and Loke 2008). This suggests
Baritto (2008) assesses which factors can make an
economy sensitive to the impact of external shocks.
CESifo Forum 2/2010
68
Focus
that Hawaii’s food ‘infrastructure’ may be quite vulnerable in
the case of disaster.
Figure 1
UNEMPLOYMENT RATE IN KAUAI AND MAUI 1990 TO 2008
not seasonally adjusted
20
%
Kauai’s hurricane Iniki
15
Kauai County
Hurricane Iniki hit land on the
south shore of Kauai on the
10
afternoon hours of 11 September 1992: for a more detailed
5
description of Iniki’s trajectory
Maui County
and consequences, see Coff0
man and Noy (2009). CRED1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
EMDAT, the most comprehenSource: University of Hawaii Economic Research Organization (UHERO) database.
sive and readily available international data source on natural
Japanese economy was suffering from the aftermath
disasters, estimates that 4 people were killed, 25,000
of its real estate and stock market bubbles. However,
were affected, and there was USD 7.4 billion (2008
immediately after the hurricane, unemployment on
USD) destruction of infrastructure and property.
Kauai shot up to 17 percent. Maui, our control
According to the National Oceanic and Atmosgroup, also experiences a rise in unemployment as a
pheric Administration (NOAA), 14,350 homes
result of the Japanese financial crisis, but on Maui
were damaged or destroyed on Kauai, and electric
unemployment peaked at only 9 percent. Clearly, the
power and telephone service were lost throughout
drastic rise in unemployment was primarily due to
the island and only 20 percent of the power had
Iniki.
been restored four weeks later. Crop damage was
likewise extensive.
Maybe more striking is that it took Kauai seven
years for its labor market to recover to its previous
Iniki was not the only hurricane to hit Kauai in the
pre-Iniki unemployment rate of 7 percent (by that
past 50 years but the direct destruction that Iniki
time, Maui’s unemployment was less than 5 percent).
wrought was unprecedented. These direct impacts,
This pattern of a recovery that takes seven to eight
however, do not necessarily represent the longer-run
years, while evident in several other statistics, is also
indirect economic effects of the hurricane.
misleading because there was a striking influx of resDistinguishing these long-term economic effects is,
ident out-migration. Figure 2 presents population
however, not easy since Kauai, like the other
figures for Kauai and Maui.
Hawaiian Islands, was hit at about the same time by
a prolonged and painful recession in Japan. The similarities
Figure 2
between Kauai and Maui
(another Hawaiian island that
RESIDENT POPULATION OF KAUAI AND MAUI 1970 TO 2008
in thousands
was not hit by the hurricane),
Maui County
Kauai County
150
70
enable us to identify more definitively the hurricane’s economic
126
58
effects.
The massive destruction of property and infrastructure resulted
in a dramatic rise in unemployment as is shown in Figure 1.
Unemployment was already
inching up from a low of around
3 percent in 1990 to 7 percent
just before the hurricane as the
102
46
78
34
54
22
30
10
1970
1975
1980
1985
1990
1995
2000
2005
Source: University of Hawaii Economic Research Organization (UHERO) database.
69
CESifo Forum 2/2010
Focus
Figure 3
TOURIST ARRIVALS FOR KAUAI AND MAUI 1970 TO 2008
seasonally adjusted, in thousands
Maui County
Kauai County
700
400
600
300
tant in spite of the events of
11 September 2001, and visitor
arrival numbers only reached
the pre-Japanese bubble numbers of 1990 in 2006.
A reason why Kauai’s experience with Iniki could have been
much worse is evident once one
500
200
examines the amount of funds
the county received from state
400
100
and federal sources (see Figure 4). Clearly, the spike in
300
0
funds associated with Iniki
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
enabled a quicker recovery
Source: University of Hawaii Economic Research Organization (UHERO) database.
than otherwise would have
been the case. However, even
with this massive increase in
Both islands seem to follow a similar population tratransfers to the state government (about 450 miljectory starting in 1970, with both experiencing a
lion US dollars), the economy of Kauai only recovvery constant rate of population increase. However,
ered after nearly a decade, and by some measures
Kauai’s population trajectory shifts in 1992. For sevit had never recovered.
eral years after that (about seven to eight) the population is constant, but the rate of population increase
after that is slower than before. Not only does Kauai
Climate change and natural disasters
not recover the population it ‘lost’ as a result of the
hurricane, it has yet to return to its previous growth
The Intergovernmental Panel on Climate Change
rate. In that sense, Kauai seems to have permanent(IPCC 2007) states that: ‘warming of the climate sysly ‘lost’ about 10 percent of its population. By comtem is unequivocal, as is now evident from observaparing Kauai and Maui’s population trajectories, it
tions of increases in global average air and ocean
again becomes apparent that the changes in Kauai
temperatures, widespread melting of snow and ice
were the result of Iniki since both were exposed to
and rising global average sea level’. Average global
the same external economic conditions beffeting the
surface warming is projected to increase by between
islands during the 1990s.
1.8 and 4 degrees Celsius depending on the success
of emissions mitigation strategies by 2100 (IPCC
Figure 3 examines the tourism sector in both islands
2007). The 2007 IPCC report predicts sea levels to
by looking at tourist arrival data for the years surrounding the hurricane. The
tourism cycle seems to be highly
Figure 4
correlated between the two
islands, except for the seven to
TRANSFER PAYMENTS FOR KAUAI AND MAUI 1969 TO 2007
in million US dollars
eight years following Iniki.
Maui County
Kauai County
800
600
Immediately after the hurricane, tourist arrivals drop by 70
percent, with a temporary spike
600
450
up in tourist arrivals in Maui
when previously booked trips
400
300
were re-directed to Maui facilities. Kauai’s tourism based
200
150
economy only recovered almost
a decade later when tourist
arrivals climbed back up to pre0
0
1969
1973
1977
1981
1985
1989
1993
1997
2001
2005
Iniki numbers in 2000. This
number remained fairly consSource: University of Hawaii Economic Research Organization (UHERO) database.
CESifo Forum 2/2010
70
Focus
dependence on ocean ecosystems, and limited fresh
water availability), but also on the economic conditions of islands (tourism dependence, narrow export
markets, large import base and limited agricultural
production). As the case of Iniki shows, high dependence on tourism made Kauai particularly vulnerable in the aftermath of the storm. It was difficult for
Kauai’s economy to rebound quickly as it immediately lost its primary export and income generator:
visitors. Kauai did not regain its previous levels of
visitor arrivals until 2000, eight years after Iniki hit.
That resulted in out-migration of Kauai residents
from which the island has never fully recovered.
rise between 0.18 and 0.59 meters by 2100. Current
predictions of global sea level rise are considerably
more drastic, however, as increased information on
glacial melting has become available since 2005. In
addition, the absorption of carbon in the ocean has
lead to increased acidity and has resulted in calcification of coral reefs. Coral bleaching leads to
destruction of surrounding ecosystems, both harming fisheries and deteriorating reef systems that protect coastal areas from storm surges.
There is limited understanding on how global warming will affect hurricanes (or cyclones). There are five
necessary conditions for hurricane formation:
(1) ocean water temperature greater than 26 degrees
Celsius (80 degrees Fahrenheit) to a depth of about
fifty meters, (2) an unstable atmosphere (i.e. thunderstorm activity), (3) high relative humidity in the
middle troposphere, (4) a pre-existing disturbance
with cyclonic circulation, and (5) little to no change
in the wind speed or direction so that warm air is
concentrated over one area (Businger 2009). Several
studies posit that, as global sea surface temperatures
rise, there will either be more or more intense hurricanes (see e.g. Webster et al. 2005).
References
Anbarci, N., M. Escaleras and C. A. Register (2005), “Earthquake
Fatalities: The Interaction of Nature and Political Economy”,
Journal of Public Economics 89, 1907–1933.
Benson, C. and J. C. Edward (2004), Understanding the Economic
and Financial Impacts of Natural Disasters, World Bank Disaster
Risk Management Series 4.
Businger, S. (2009), Hurricanes in Hawaii, Poster Developed for the
Hurricanes and Extreme Weather Phenomena Symposium,
http://www.soest.hawaii.edu/MET/Faculty/businger/poster/hurricane.
Cavallo, E., S. Galiani, I. Noy and J. Pantano (2009), Disasters and
Institutions, mimeo.
Coffman, M. and I. Noy (2009), Hurricane Iniki: Measuring the
Long-term Economic Impact of a Natural Disaster Using Synthetic
Control, University of Hawaii Economics Working Paper 09-05.
The science, however, is not entirely conclusive.
IPCC (2007) states that: ‘there is observational evidence of an increase in intense tropical cyclone
activity in the North Atlantic since about 1970, with
limited evidence of increases elsewhere. There is no
clear trend in the annual numbers of tropical
cyclones. It is difficult to ascertain longer-term trends
in cyclone activity, particularly prior to 1970’ (IPCC
2007). Elsner et al. (2008) suggested that warming
temperatures allowed for already strong storms to
get even stronger; while there may not necessarily be
an increased occurrence of storms, there will be an
increased occurrence of strong storms. Islands are
extremely vulnerable to the expected (and often
already occurring) effects of climate change. Sea
level rise will exacerbate inundation, storm surge,
erosion and other coastal hazards (IPCC Working
Group II 2007).
Cuaresma, J. C., J. Hlouskova and M. Obersteiner (2008), “Natural
Disasters as Creative Destruction? Evidence from Developing
Countries”, Economic Inquiry 46, 214–226.
Elsner, J., J. Kossin and T. Jagger (2008), “The Increasing Intensity of
the Strongest Tropical Cyclones”, Nature 455, 92–95.
Heger, M., A. Julca and O. Paddison (2008), Analysing the Impact of
Natural Hazards in Small Economies: The Caribbean Case,
UNU/WIDER Research Paper 2008/25.
Intergovernmental Panel on Climate Change (IPCC, 2007), Climate
Change: The Physical Science Basis, Geneva: IPCC.
IPCC Working Group II (2007), “Summary for Policymakers”, in:
Parry, M. L., O. F. Canziani, J. P. Palutikof, P. J. van der Linden and C.
E. Hanson (eds.), Climate Change 2007: Impacts, Adaptation and
Vulnerability”, Contribution of Working Group II to the Fourth
Assessment Report of the Intergovernmental Panel on Climate
Change, Cambridge: Cambridge University Press, 7–22.
Kahn, M. E. (2004), “The Death Toll from Natural Disasters: The
Role of Income, Geography, and Institutions”, Review of Economics
and Statistics 87, 271–284.
Landsea, C. W., B. A. Harper, K. Hoarau and J. A. Knaff (2006), “Can
We Detect Trends in Extreme Tropical Cyclones?” Science 313,
452–454.
Leung, P. and M. Loke (2008), Economic Impacts of Increasing
Hawaii’s Food Self-Sufficiency, mimeo.
Noy, I. (2009), “The Macroeconomic Consequences of Disasters”,
Journal of Development Economics 88, 221–231.
Conclusion
Noy, I. and A. Nualsri (2007), What Do Exogenous Shocks Tell Us
about Growth Theories?, University of Hawaii Working Paper 07-28.
Islands are vulnerable to effects of climate change
with very high confidence (IPCC Working Group II
2007). This assessment is based not, however, only on
the physical implications of climate change on island
ecosystems (proportionately larger coastal areas,
Perrow, C. (2007), The Next Catastrophe: Reducing Our
Vulnerabilities to Natural, Industrial, and Terrorist Disasters,
Princeton: Princeton University Press.
Raschky, P. A. (2008), “Institutions and the Losses from Natural
Disasters”, Natural Hazards Earth Systems Science 8, 627–634.
71
CESifo Forum 2/2010
Focus
Rasmussen, T. N. (2004), Macroeconomic Implications of Natural
Disasters in the Caribbean, IMF Working Paper WP/04/224.
Skidmore, M. and H. Toya (2002), “Do Natural Disasters Promote
Long-Run Growth?”, Economic Inquiry 40, 664–687.
United Nations (2010), Terminology of Disaster Risk Reduction,
http://www.unisdr.org/eng/library/lib-terminology-eng%20home.
htm.
Webster, P. J., G. J. Holland, J. A. Curry and H. R. Chang (2005),
“Changes in Tropical Cyclone Number, Duration, and Intensity”,
Science 309, 1844–1846.
CESifo Forum 2/2010
72
Focus
(National Research Council 1999). 9/11, a manmade
disaster, was of similar magnitude.
SHORT-RUN ECONOMIC
IMPACTS OF HURRICANE
KATRINA (AND RITA)
Covering the first year after the disaster, several
studies on its economic impacts were completed.
However, most of this research was from governmental reports mainly focusing on the direct losses
or on speculations about future impacts on the area.
Louisiana received federal reimbursements for losses of about 105 billion US dollars (Kent 2006).
Nordhaus (2006) based an analysis on the economic
impacts from US hurricanes since 1950 and came up
with an estimate of 81 billion US dollars for hurricane Katrina.
PETER GORDON*,
JAMES E. MOORE II*,
JIYOUNG PARK** AND
HARRY W. RICHARDSON*
A tropical depression formed over the southeastern
Bahamas on 23 August 2005, moved toward the Gulf
of Mexico, and strengthened to category 5 on the
Saffir-Simpson Hurricane Scale over the central
Gulf of Mexico (NCDC 2005). When hurricane
Katrina made landfall on the Louisiana coast with
category 3 intensity on 29 August 2005, 130mph of
sustained winds breached the levees of New
Orleans and caused substantial inundation. A flood
following the storm, devastated the Crescent City,
and the disaster was recorded as the costliest natural disaster ever in US history, resulting in an 80 percent flood in the City of New Orleans and over
1,800 casualties (Louisiana Geographic Information
Center 2005).
However, the total (direct, indirect and induced)
economic losses were higher than these estimates, in
part because of the interdependence between economic sectoral activity and household consumption.
Park et al. (2006a) estimated the direct and indirect
economic losses because of the inoperability of the
Port of New Orleans in the seven months after the
hurricane as 62.1 billion US dollars. Of course, the
sectors that rely heavily on waterborne commerce
were more severely affected, although all major economic sectors were negatively impacted during the
storm and recovery period.
The repercussions of hurricane Katrina (and hurricane Rita that happened soon after) continue until
today and beyond into the future. However, most of
the efforts now are focusing on housing provision,
social reconstruction and community development.
Of course, these have an economic impact, but this
paper primarily focuses on the business interruption
impacts, soon after the disaster.
It is useful to examine other economic losses in the
region. Several oil and gas refineries were shut down
for more than a week. 115 offshore oil platforms
were missing, sunk, or went adrift. One-half of
1.3 million evacuees from the New Orleans metropolitan area were not able to return in the first
month after the storm, and many key workers were
away for much longer (Katz et al. 2006).
Prior to hurricane Katrina, the three costliest natural disasters in terms of dollar magnitude of damages
recorded in the United States were the drought in
1988 with estimated losses of over 39 billion US dollars, hurricane Andrew in 1992 which cost 30 billion
US dollars and the Northridge earthquake in 1994
which resulted in over 44 billion US dollars
The Energy Information Administration (EIA
2006a) released a report analyzing historical impacts
of tropical cyclones on Gulf of Mexico crude oil and
natural gas production over the period 1960 through
2005, and refinery operations over the past 20 years.
The analysis showed that tropical storms and hurricanes in the Gulf area typically cause seasonal disruption of shut-in production of 1.4 percent for
crude oil and 1.3 percent for natural gas compared to
* University of Southern California.
** State University of New York-Buffalo.
73
CESifo Forum 2/2010
Focus
normal annual production from
Figure 1
wells on the Outer Continental
Shelf (OCS). However, these
averages are skewed upwards by
the 19 percent of oil production
and 18 percent of natural gas
production that was shut in during 2005. Also, the Government
Accountability Office (GAO
2006) released a report addressing the factors causing natural
gas price increases, influences on
consumers according to the
higher prices, and the adequacy
of roles of federal government
agencies played in ensuring natural gas prices competitive. In
September 2005, natural gas
spot prices increased to over 15
US dollars per million BTUs,
which is roughly twice as high as
the average price in July of that
US national market because crude oil production as
year. The skyrocketed price resulted from a substanwell as petroleum products in the area accounted for
tial portion of domestic supply disruption and excesnearly three-fifths of the total US output in this secsive demand because of colder weather than expecttor in 2004 (EIA 2006b).
ed (ibid). In research of economic losses because of
the employment changes, a report of Bureau of
Not surprisingly, the Gulf of Mexico offshore instalLabor Statistics (BLS 2006) presented the impacts of
lations have a significant place in the US oil and gas
Katrina on employment in the Gulf coast area by
industries such that the domestic gasoline price escaexamining over-the-year changes. Employment in
lated significantly right after the two storms. Figure 2
the most severely affected parish in Louisiana was
indicates the fact that the effects of the hurricanes
down by nearly 40 percent in September 2005 comwere not confined to the area. The total volume of
pared to a year before. Colgan and Adkins (2006)
production for the entire US shows an abrupt drop
discussed the proportion of employment and wages
in September 2005. The flow parallels the Gulf coast
of the affected industries defining them as ‘ocean
flow while the rest of the US shows a relatively
industries’. Including oil and gas exploration as well
as marine transportation and
related goods and services, the
Figure 2
ocean economy of the region
encompassing Florida, Alabama,
PETROLEUM PRODUCTION PRE- AND POST-KATRINA AND RITA
Mississippi, Louisiana and Texas
Million barrels per day
20
employed 291,830 people in
Katrina
Rita
wage and salary jobs paying
15
nearly 7.7 billion US dollars for
the wages in 2004.
10
CESifo Forum 2/2010
5
Sources: EIA; authors' calculation.
74
Sep 06
Jul 06
May 06
Mar 06
Jan 06
Nov 05
Sep 05
Jul 05
May 05
Jan 05
Others
Mar 05
Nov 04
PADD III
Sep 04
Jul 04
May 04
Mar 04
US Total
0
Jan 04
Figure 1 shows the severe impacts of two hurricanes (Katrina
and Rita) on oil and gas production platforms on the coast of
the Gulf of Mexico, where the
inoperability of gas and oil
industries severely affected the
Focus
multi-regional input-output model for the 50 states
and the District of Columbia. Both models provide
results for 47 industrial sectors (labeled the USC
Sectors). NIEMO has a supply-side as well as a
demand-side capability. In applications to hypothetical or actual port closures, for example, the loss of
exports is best modeled via the demand-side
NIEMO, whereas the loss of imports is modeled via
the supply-side NIEMO.
steady trend. In other words, the rapid decrease in
petroleum production of the United States from
September 2005 mainly resulted from reduced production in the Gulf coast.
Crude oil industries in the Gulf region are closely
related to port activity. The analysis conducted by
Park et al. (2006a) addressed disruptions of port
activity, including oil industries. This study, however, focuses on the oil-refinery industries of
the Gulf of Mexico by subtracting foreign and
domestic exports from the total output of oil
refineries.
This type of model is most useful for short-term
impact analysis because buyers and sellers can be
expected to eventually make substitutions in light
of the price changes that follow longer-term major
disruptions. Omitting these effects is a well-known
limitation of the IO approach. Here, we describe
how to use post-event information on concurrent
demand and value-added changes to identify the
technological (production function) changes that
occur after a major disruption. We compare these
results to the estimates from the baseline NIEMO
to show the detailed impacts of substitutions and
adaptations.
The Gulf of Mexico region is defined as PADD III
(Texas Inland, Texas Gulf Coast, Louisiana Gulf
Coast, North Louisiana-Arkansas and New
Mexico) shown in Figure 3. This requires spatial
aggregation by modifying our National Interstate
Economic Output Model (NIEMO), to 47 regions
from the original 52 (including the rest-of-theworld) regions. The next section illustrates our
approach to estimating direct impacts required as
input data for the 47-region NIEMO, and applying
what we call the Flexible National Interstate
Economic Model (FlexNIEMO).
As seen in Table 1, Louisiana experienced an economic decline (by 1.5 percent) in the years 2004 to
2005. According to BEA’s data, except for Louisiana
and Alaska, all the other states grew in terms of
Gross State Product. Also, the mining sector including oil and gas production was the most negatively
impacted component of GDP.
Input-output models have been applied to the problem of economic impact estimation for many years.
In recent years, our group has developed and applied
IO models that include substantial spatial disaggregation. Most decision makers are interested in local
effects and our models can estimate these. Our
National Interstate Economic Model (NIEMO) is a
Methodologies
Figure 3
PETROLEUM ADMINISTRATION FOR DEFENSE DISTRICT MAPS
The Holt-Winters time-series
approach was used to estimate
normal economic trends, if the
hurricanes had not occurred.
These estimates provided the
direct impacts necessary for
input data into the demand-side
NIEMO. We then used the
FlexNIEMO to construct monthto-month supply-side versions of
NIEMO.
The Holt-Winters approach to
estimating the normal economic
status using times-series methodology is described in several
recent articles (Park et al. 2006a;
Park et al. 2006b; Richardson et
al. 2007; Gordon et al. 2007). The
75
CESifo Forum 2/2010
Focus
Table 1
Contributions to Percent Change in Real GSP, 2004–2005
State and region
US
Southeast
Louisiana
Agriculture, forestry, fishing and hunting
– 0.05
– 0.01
– 0.05
Mining
– 0.04
– 0.12
– 1.66
Utilities
0.01
0.00
– 0.07
Construction
0.13
0.27
– 0.07
Durable goods manufacturing
0.40
0.37
– 0.04
Non-durable goods manufacturing
0.08
0.08
0.48
Wholesale trade
0.07
0.13
– 0.04
Retail trade
0.20
0.32
0.03
Transportation and ware-housing
0.11
0.11
0.02
Information
0.34
0.40
0.14
Finance and insurance
0.54
0.54
0.22
Real estate, rental and leasing
0.32
0.67
– 0.41
Professional and technical services
0.48
0.47
– 0.13
Management of companies
0.01
0.04
0.05
Administrative and waste services
0.21
0.32
0.19
Educational services
0.01
0.01
– 0.02
Health care and social assistance
0.33
0.34
– 0.08
Arts, entertainment, and recreation
0.02
0.03
– 0.05
Accommodations and food services
0.13
0.17
0.00
Other services
0.06
0.06
– 0.06
Government
0.18
0.41
0.07
Total
3.60
4.60
– 1.50
Notes: Real GSP is adjusted based on 2000 dollars. The Southeast region includes Alabama, Arkansas, Florida,
Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia and West Virginia.
Source: US Bureau of Economic Analysis.
in the IO model. The model aggregated 52 regions to
47 regions, because the Gulf of Mexico corresponds
to six states, and treated the Gulf of Mexico as one
region. Therefore, the newly defined NIEMO has
(47x47)x(47x47) different coefficients for each
month (August 2005 to September 2006) after hurricane Katrina for 47 regions and 47 sectors.
approach allows the estimated coefficients to change
gradually over time, based on data for previous periods and exponentially declining weights. Based on
the estimated coefficients, the forecast oil-refinery
industry values are obtained at the end of 2005
(August to December) and for first three quarters of
2006. Direct impacts are calculated from the difference between the actual and predicted production of
oil-refinery activities.
Results
Second, FlexNIEMO was used to construct monthly
versions of the supply-side NIEMO. The approach
developed by Park et al. (2007b) allows the fixed
coefficients in the input-output world to be continuously modified, reflecting previous economic events
and interindustry substitutions. Because oil-refinery
products are important to supporting the economy
in the Gulf of Mexico and the United States, the supply-side NIEMO approach is helpful. One problem
is how to adjust the supply-side model to reflect
demand-side adjustments during the recovery period. The analysis combines the demand-driven
NIEMO described in Park et al. (2007a) with the
supply-side NIEMO in Park (2006). This solution
overcomes some of the major shortcomings inherent
CESifo Forum 2/2010
Figure 4 shows the 13 months of forecasts using the
Holt-Winters method, which is adjusted monthly.
The R-Square is 87 percent and Theil’s U statistic
which summarizes the forecasting accuracy show
0.071. Because the U statistic is close to 0 and the U
of no predictive power is 1 (Theil 1966; Maddala
1977), the forecasts are statistically acceptable.
The analysis here has concentrated on the total business interruption impacts of the hurricanes on the
dominant sector (oil refining) by using a multiregional input-output model (NIEMO). These
amounted to 8.28 billion US dollars in the first year
after the hurricanes; even in September 2006 actual
76
Focus
Figure 4
CHANGES OF ACTUAL AND ESTIMATED OIL-REFINERY VALUES:
PADD III
25 000
of input-output structural decomposition analysis (see Rose
and Casler 1996).
Million US dollars per 1 000 barrels
Recent developments
20 000
Prior to the Gulf oil spill of April
2010, New Orleans was finally
beginning to recover. This paper
does not deal with the most
10 000
Estimated values
recent events, but this New York
Actual values
Times quotation (7 May 2010)
5 000
offers a good summary: “since
2004
2005
2006
the Saints won the upon the
Sources: EIA; authors' calculation.
backdrop of Mardi Gras, followed by the landslide election
output remained below estimated output from a
of a popular new mayor, New Orleans had been, by
forecasting model. Using the other variant of the
all accounts, getting its groove back. Five years
model (FlexNIEMO) which allows for input substiremoved from hurricane Katrina, the tangible signs
tutions in response to changes in relative prices, the
of a real recovery are everywhere: in rebuilt homes
total impacts fall to 4.85 billion US dollars. As for the
and refurbished parks, in old restaurants come back
state-by-state impacts, not surprisingly, most
to life and in new businesses thriving”. The conseoccurred in the Gulf states – more than 92 percent.
quences of the oil spill are still unclear, especially in
Similarly, because oil refining has few interrelationterms of its impact on seafood and tourism, but
ships with other sectors, almost all the sectoral
appear to get worse day by day. Only 20 percent of
impacts (98.04 percent) are restricted to the oil secthe seafood Americans eat is domestic, but most of it
tor. The primary policy implication from the analysis
comes from either Alaska or Louisiana. The shrimp
is that the business interruption costs from the hurriindustry alone, which produced 90 million pounds in
canes (and from the more recent Gulf oil spill) pro2008, brings in 1.3 billion US dollars a year.
vide an upper threshold on how much policymakers
might pay to prevent and/or mitigate similar events.
The difference in the results from an application of
Other issues
NIEMO and an application of FlexNIEMO are dramatic. The original model estimates an overall multiAlthough this paper primarily focuses on the shortplier of 1.83 while the new results indicate a much
term (typically 2005–2006) economic impacts of the
smaller multiplier, 1.07.
hurricanes, especially as they affected the dominant
industry of oil refining, there are some other policy
Impact modeling using widely available input-output
issues that merit attention, even if briefly.
approaches routinely includes the caveat about the
fixed technologies assumption and how that overOne important item is the reconstruction of the levstates the estimated results. We have adapted a new
ees in New Orleans by the US Army Corps of
and operational multiregional input-output model of
Engineers. Presumably because of budget conthe US NIEMO, to analyze substitutions and have
straints, they are rebuilding only to category 3 hurriconsidered their scale and scope for the case of oil
cane standards. This makes little sense because a catand gas refinery losses in the Gulf of Mexico followegory 5 hurricane is possible, even likely at some
ing hurricanes Katrina and Rita in late 2005. The
time. A related problem is the lack of back-up elecresults suggest that a detailed study of substitutabilitricity generators for the pumping stations. The failty is useful because overstated impacts from the
ure of these was a primary factor in the severity of
application of conventional IO are substantial.
the flooding.
NIEMO generates millions of multipliers that
remain to be explored at the individual sector level,
Yet another problem is the defects in the insurance
by month, sector and region. This is in the tradition
system (Kunreuther and Michel-Kerjan, 2008). The
15 000
77
CESifo Forum 2/2010
Focus
2006, hotel and restaurant employment was about
70 percent of the pre-Katrina level. Nevertheless,
the comeback was quite surprising. The explanation was that the high-lying French Quarter
escaped serious flood damage, although some
hotels and restaurants were damaged by wind and
activities were impeded in the short run by power
outages and other inconveniences. Attendance in
the last weekend of the 2006 Mardi Gras was
70 percent of the 2005 level (about 700,000). In
2007, it was about 100,000 more. By 2010, however,
all three major festivals (Mardi Gras, the Essence
and Jazz Festivals) achieved record attendances.
As suggested above, it is unclear whether these
performances will be repeated in 2011 because the
short-term future of seafood production (that
plays such an important role in New Orleans
tourism) is in doubt as a result of the new disaster,
the Gulf oil spill of April 2010.
scale of hurricane Katrina made the defects in catastrophe insurance obvious. There needs to be a
major shift to risk-based premiums, an incentives
scheme to encourage firms and households to invest
in mitigation measures, and to deal with equity issues
via some kind of subsidy program but not by subsidizing insurance premiums.
The housing issue which has still not been resolved is
critical because it has affected the lives of so many
people. More than 200,000 structures were damaged,
most of them because of severe flooding resulting
from the levees. Of these structures, well more than
one-half were housing units, about evenly split
between owner-occupied rental housing (approximately 67,000 of each). According to the US
Department of Housing and Urban Development
(HUD), 71.5 percent of the occupied units in
Orleans Parish (the city) were damaged and
41.9 percent were severely damaged or destroyed. In
the five most impacted counties 305,000 units were
damaged, i.e. 65.1 percent of the occupied housing
stock, and 22 percent were severely damaged or
destroyed. In New Orleans itself only 3 high-lying
neighborhoods out of 14 avoided severe damage to
rental housing and not a single neighborhood
(Bostic and Molaison 2008).
Conclusions
This paper’s primary focus has been the economic
impacts of hurricane Katrina (and, to a lesser extent,
Rita) in the first year aftermath. From then on, the
emphasis was on social and economic reconstruction
and recovery. Of course, these had economic
impacts. A common argument in both natural and
manmade disaster discussions is that such disasters
are over time a ‘wash’ because the positive subsequent recovery impacts more or less balance out the
negative initial disaster impacts. However, the problem with this approach is that it neglects the opportunity costs of the resources used in the recovery
efforts. The implication is that including recovery
activities among the favorable economic impacts
associated with a disaster is misleading, if not downright wrong.
Housing damage was the primary factor explaining
the population loss of the City of New Orleans from
458,000 prior to Katrina to a post-Katrina low of
137,000 four months later. After that, it began to
recover but slowly and still remains below its peak.
The population loss was not confined to the city
alone. Population declined in other parishes outside
the city, in one case – St. Bernard Parish – even more
than in New Orleans itself (by 80 percent in the first
six months compared with 60 percent in New
Orleans itself – see Bostic and Molaison 2008). Not
surprisingly, the situation stimulated a surge in
repair, reconstruction and new construction and an
inundation of workers, many of them Latino. The
associated demand of non-resident workers for
rental accommodation made the housing problem
even worse.
References
Bostic, R. W. and D. Molaison (2008), “Hurricane Katrina and
Housing: Devastation, Possibilities and Prospects”, in: H. W.
Richardson, P. Gordon and J. E. Moore II (eds.), Natural Disaster
Analysis after Hurricane Katrina, Cheltenham: Edward Elgar,
253–278.
Bureau of Labor Statistics (BLS, 2006), “The Labor Market Impact
of Hurricane Katrina: An Overview”, Monthly Labor Review 129/8,
3–10, http://www.bls.gov/opub/mlr/2006/08/art1abs.htm.
As for tourism (the second most important sector), Mardi Gras made a partial recovery in 2006.
Some of the parades were cancelled (six fewer
parades in Orleans Parish, with an average of
three fewer floats on each parade) and hotel occupancy rates were about 25 percent below the festival rates in 2005 (Deloughery 2008). In March
CESifo Forum 2/2010
Colgan, C. S. and J. Adkins (2006), “Hurricane Damage to the Ocean
Economy in the U.S. Gulf Region in 2005”, Monthly Labor Review
129/8, 76–78, http://www.bls.gov/opub/mlr/2006/08/art7abs.htm.
Deloughery, K. (2008), “Is New Orleans Ready to Celebrate after
Katrina? Evidence from Mardi Gras and the Tourism Industry”, in:
H. W. Richardson, P. Gordon and J. E. Moore II (eds.), Natural
Disaster Analysis after Hurricane Katrina, Cheltenham: Edward
Elgar, 134–146.
78
Focus
Parks”, in: Richardson, H. W., P. Gordon and J. E. Moore II (eds.),
The Economic Costs and Consequences of Terrorism, Cheltenham:
Edward Elgar, 235–253.
Energy Information Administration (EIA, 2006a), The Impact of
Tropical Cyclones on Gulf of Mexico Crude Oil and Natural Gas
Production,
http://tonto.eia.doe.gov/FTPROOT/features/hurricanes.pdf.
Theil, H. (1966), Applied Economic Forecasting, Amsterdam: North
Holland.
Energy Information Administration (EIA, 2006b), Weekly
Petroleum Status Report, http://www.eia.doe.gov/oil_gas/petroleum/data_publications/weekly_petroleum_status_report/wpsr.html.
Gordon, P., H. W. Richardson, J. E. Moore, II and J. Y. Park (2007),
“The Economic Impacts of a Terrorist Attack on the U.S.
Commercial Aviation System”, Risk Analysis 27, 505–512.
Government Accountability Office (GAO, 2006), Natural Gas:
Factors Affecting Prices and Potential Impacts on Consumers (GAO06-420T), http://www.gao.gov/new.items/d06420t.pdf.
Katz, B., M. Fellowes and M. Mabanta (2006), Katrina Index:
Tracking Variables of Post-Katrina Reconstruction, Washington DC:
The Brookings Institution.
Kent, J. D. (2006), 2005 Louisiana Hurricane Impact Atlas Volume 1,
Louisiana Geographic Information Center, http://lagic.lsu.edu/lgisc/
publications/2005/LGISC-PUB-20051116-00_2005_HURRICANE_ATLAS.pdf.
Knabb, R. D., D. P. Brown and J. R. Rhom (2006), Tropical Cyclone
Report Hurricane Rita: 18–26 September 2005, National Hurricane
Center, http://www.nhc.noaa.gov/pdf/TCR-AL182005_Rita.pdf.
Kunreuther, H. C. and E. O. Michel-Kerjan (2008), Comprehensive
Disaster Insurance: Will It Help in a Post-Katrina World?”, in: H. W.
Richardson, P. Gordon and J. E. Moore II (eds.), Natural Disaster
Analysis after Hurricane Katrina, Cheltenham: Edward Elgar, 8–33
Louisiana Geographic Information Center, (2005), Louisiana
Hurricane Impact Atlas, Vol. 1, http://lagic.lsu.edu/images/hurricanes/2005_LAGIC_HURRICANE_ATLAS.pdf.
Maddala, G. S. (1977), Econometrics, New York: McGraw-Hill.
National Climate Data Center (NCDC, 2005), Climate of 2005:
Summary of Hurricane Katrina, http://www.ncdc.noaa.gov/oa/climate/research/2005/katrina.html#top.
National Hurricane Center of National Weather Service (2007),
Tropical Weather Summary – 2005 Web Final, http://www.nhc.noaa.
gov/archive/2005/tws/MIATWSAT_nov_final.shtml.
National Research Council (1999), The Impacts of Natural
Disasters: A Framework for Loss Estimation, http://www.nap.edu/
catalog/6425.htmlb.
Nordhaus, W. D. (2006), The Economics of Hurricanes in the United
States, NBER Working Paper 12813.
Park, J. Y. (2006), The Economic Impacts of a Dirty-Bomb Attack on
the Los Angeles and Long Beach Port: Applying Supply-Driven
NIEMO, Paper Presented at 17th Annual Meeting of the
Association of Collegiate Schools of Planning, Fort Worth, TX,
9–12 November.
Park, J. Y., P. Gordon, S. J. Kim, Y. K. Kim, J. E. Moore II and H. W.
Richardson (2006a), Estimating the State-by-State Economic
Impacts of Hurricane Katrina, Paper Presented at CREATE
Symposium: Economic and Risk Assessment of Hurricane Katrina,
University of Southern California, CA, 18–19 August.
Park, J. Y., C. K. Park and S. J. Nam (2006b), The State-by-State
Effects of Mad Cow Disease Using a New MRIO model, Paper
Presented at 2006 American Agricultural Economic Association
(AAEA) Annual Meeting, Long Beach, CA, 23–26 July.
Park, J. Y., P. Gordon, J. E. Moore II, H. W. Richardson and L. Wang
(2007a), “Simulating the State-by-State Effects of Terrorist Attacks
on Three Major U.S. Ports: Applying NIEMO (National Interstate
Economic Model)”, in: Richardson, H. W., P. Gordon and J. E.
Moore II (eds.), The Economic Costs and Consequences of
Terrorism, Cheltenham: Edward Elgar, 208–234.
Park, J. Y., P. Gordon, J. E. Moore II and H. W. Richardson (2007b),
Constructing a New Resilient National Interstate Economic Model,
Paper Presented at the 46th Annual Meeting of the Western
Regional Science Association, Newport Beach, CA, 21–24 February.
Rose, A. and S. Casler (1996), “Input-Output Structural
Decomposition Analysis: A Critical Appraisal”, Economic Systems
Research 8, 33–62.
Richardson, H. W., P. Gordon, J. E. Moore, II, S. J. Kim, J. Y. Park and
Q. Pan (2007), “Tourism and Terrorism: The National and
Interregional Economic Impacts of Attacks on Major U.S. Theme
79
CESifo Forum 2/2010
Focus
it is a measure of the labor market and not of the
overall economy, although the two are certainly
related (Ewing, Kruse and Thompson 2003 and
2004; Skidmore and Toya 2002). Thus, one shortcoming to using employment or unemployment
rate data to infer overall economic impact is that it
does not account for payroll or wages, thus treating
all jobs equally and inherently treating the consumption functions of various institutions and
households as homogeneous. In this respect, the
labor market approach, while informative, does not
capture the typical multiplier effects that one might
find estimated in input-output regional economic
models.
MEASURING THE REGIONAL
ECONOMIC RESPONSE TO
HURRICANE KATRINA
BRADLEY T. EWING*,
JAMIE B. KRUSE** AND
MARK A. THOMPSON*
Introduction
Catastrophic events such as natural disasters have
an enormous impact on regional economies. In fact,
a vast literature on the economic impact of hurricanes, tornadoes and other disasters exists (Ewing,
Kruse and Sutter 2007). A strand of this research
utilizes regional labor market and output measures
as indicators of regional well-being and recovery. In
particular, many time series econometric studies
examine the economic impacts by comparing preand post-disaster periods, primarily using variations
of event study techniques and intervention analysis.
As such, previous researchers have generally
focused on obtaining post-event periods that were
as long as possible in order to increase the number
of post-disaster data points. One drawback of the
event study line of research is that it does not provide ‘real time’ assessments of the impact of major
disasters. Another issue deals with the choice of the
appropriate performance measure to capture the
economic condition of the region. In this paper, we
illustrate how the use of regional-level (i.e. US
state) coincident indexes may be used to infer the
extent and magnitude of a natural disaster on a local
economy.
On the other hand, gross domestic product (GDP)
does measure the performance of the overall economy but is released on a quarterly basis at the
national (US) level and only annually at state and
regional levels (as defined by the US Bureau of
Economic Analysis and the US Bureau of Labor
Statistics). Quarterly and annual frequencies can
be problematic when estimating the economic
impact of disasters since differences in the immediate and longer-term effects have been found to
exist (Ewing, Kruse and Thompson 2009).
Furthermore, GDP metrics at the state and regional level often lag two to three years, further compounding the measurement problem. The lack of
timely information could lead policymakers into
making poor or suboptimal resource allocation
decisions.
One approach to resolve these issues combines several regional economic indicators into a composite
index that can be tracked through the business cycle.
The composite index offers the potential of better
and, possibly, more complete information on the
condition of the economy than individual economic
indicators and would be more current than GDP
estimates. Consequently, a properly constructed
composite index more closely captures elements of
the overall economic well-being of a region than
using a single labor market measure and, simultaneously, is timelier than using a single output measure
such as gross product.
A number of time series studies have examined the
impact of tornadoes and hurricanes using monthly
labor market data (Ewing and Kruse 2001 and
2002; Ewing, Kruse and Thompson 2003, 2004,
2005a, 2005b and 2009). While employment-related
data is timely and regionally defined, by definition
* Texas Tech University.
** National Oceanic and Atmospheric Administration, Silver
Springs, Maryland.
CESifo Forum 2/2010
80
Focus
al. 2007). Standard macroeconomic theory highlights
how the losses in physical and human capital place a
drag on long-term growth and potential output of
the region. However, there is an inflow of financial
assistance from insurance to federal, state and local
aid. Additionally, following a disaster, the rebuilding
of infrastructure, networks, and the installation and
implementation of new technology may lead to higher growth over time. Moreover, the economic
improvement may spread to connecting regions.
Ewing, Kruse and Thompson (2005b) document this
effect following the 1999 Oklahoma City tornado
outbreak. They attribute much of this economic
change to improvements to the supply chain.
Recently, Crone and Clayton-Matthews (2005)
developed a consistent set of US state-level coincident indexes that are now produced by the Federal
Reserve Bank of Philadelphia. Thompson (2009)
provided an example of how such a state-level
index could be used to measure the effect of a disaster and the subsequent regional economic
response. In particular, Thompson (2009) focused
on Louisiana’s economy and hurricane Katrina,
which made its first landfall in southern Florida as
a category one storm, moved into the Gulf of
Mexico where it intensified, and made its second
landfall 29 August 2005 near the LouisianaMississippi border as a category four storm. This
paper updates and expands the work of Thompson
(2009) and examines what, if any, effect Katrina had
on other states along the gulf coast.
The approach of this paper is to examine the behavior of the Louisiana state coincident index to determine the change in economic activity immediately
following hurricane Katrina. The recovery process is
followed through time to capture a sense of the
longer-term or ongoing effects of the hurricane on
the state. Additionally, the corresponding coincident
indexes of the other Gulf Coast states are tracked to
see the impact of Katrina and to document the
extent to which the impacts, both immediate and
ongoing, may have played out in the economic activities of those states.
Hurricane Katrina and the economy
The Federal Response to Hurricane Katrina:
Lessons Learned (2006) reports that the economic
cost of hurricane Katrina was nearly 100 billion US
dollars making it the most costly US disaster in
terms of economic losses. These losses resulted from
a major disruption in economic activity and physical
damage to infrastructure and homes. While previous
studies provided information as to the economic
impact on New Orleans (Guimaraes et al. 1993; West
and Lenze 1994), the shear size and magnitude of
Katrina on such a major city required policymakers
to make immediate decisions with respect to recovery. For this type of situation, timely and relevant
information is required.
Index composition
Certainly, a number of factors help to determine the
impact of a hurricane on a regional economy including the severity of the storm and its atmospheric
characteristics, the built environment and the area’s
economic structure. Ewing, Kruse and Sutter (2007)
provide a thorough review of the economic research
on hurricanes and describe various approaches to
modeling disasters. They conclude that valuable information can be derived from a variety of models;
however, certain decisions must be made in a short
time frame and thus sources of timely information is
necessary for efficient resource allocation.
Regional policymakers need accurate and timely
information on the current state of the economy for
planning purposes. According to Crone (2006), state
and regional markets may not follow national trends
and cycles. Moreover, natural disasters are regionally-specific events placing even greater importance
on accurate models of the regional economy for
planning purposes. While hurricane Katrina was a
major storm, its impact is still relatively regional to
the gulf coast states of the United States. As previously mentioned, such comprehensive measures of
the economy are often inadequate for timely regional information as to the impact of the storm. Due to
these shortcomings, a composite index of several
regional economic indicators may provide a better
and timelier measure of the regional business cycle
and its response to a disaster.
The short-term disruption from a hurricane may
result in out-migration from the affected region to
another area and the resulting loss in human capital
may hinder future growth and recovery (Landry et
The approach that Crone and Clayton-Matthews
take is based on the Stock and Watson (1989)
dynamic single-factor model. The dynamic singlefactor model is a variation of principal component or
81
CESifo Forum 2/2010
Focus
highlights the severity of hurricane Katrina relative
to past recessions.
factor analysis, where the resulting index represents
the underlying state of the economy. The structure of
the model is as follows:
(1)
xt = + ( L)st + µ t
(2)
D( L) µ t = t
(3)
( L)s t = + t
Table 1 provides a comparison between Louisiana’s
economic activity and that of the United States during the past five national recessions as noted by the
National Bureau of Economic Research (NBER). In
addition, note that Louisiana’s economy did not
experience a decline in economic activity as the
national economy entered two recessions (i.e. the
1980 and the 1990–91 recessions). With respect to the
1981 recession, Louisiana’s economy lagged the
national economy by seven months going into the
recession and about six months coming out of the
recession. While both the Louisiana and national
economy entered the 2001 recession at the same
time, the Louisiana economy lagged the US economy by six months in coming out of the recession. The
NBER indicated that the latest recession started in
December 2007; however, Louisiana peaked in
January 2009 and has declined in economic activity
every month thereafter (to the end of the sample
period). Crone (2006) points outs that most states do
not always follow the national business cycle –
Louisiana does not simply mirror the behavior of the
national economy.
where xt is the logarithm of the observed variable in
period t, st is the logarithm of the state variable to be
estimated (i.e. the common factor), and L represents
the lag operator.
The idiosyncratic components in the measurement
equations from (1) follow an autoregressive process
and are uncorrelated with one another. In particular
to developing a consistent set of state-level indexes,
Crone and Clayton-Matthews (2005) use employment, unemployment, hours worked in manufacturing, and real wages as the measurement variables in
equation (1). The underlying state of the economy is
represented by the state variable st and the final
index sets the estimated state variable to 100 (for a
given date). Crone and Clayton-Matthews (2005)
retrend each index to the respective state GDP trend
to allow for comparisons across states.
Since states may not necessarily follow the national
economy, their response to natural disasters may differ as well. Table 1 illustrates the timing of hurricane
Katrina. The Louisiana economy declined for three
months following the hurricane. Using the index, we
can measure how long it took for the state economy
to recover to its pre-hurricane level. Following hurricane Katrina, the Louisiana index dropped to
Each month, the Federal Reserve Bank of
Philadelphia produces the 50 state coincident indexes. It is also worth noting that the indexes are re-estimated each month with the revised data. Below, we
use the state coincident index to examine how
Louisiana’s economy responded before and after
hurricane Katrina. In addition,
we examine how other gulf coast
Figure 1
states responded.
ECONOMIC GROWTH RATES IN LOUISIANA AND THE US
Louisiana’s state coincident
index and hurricane Katrina
Louisiana growth rates
US growth rates
8
6
Figure 1 illustrates the growth in
economic activity for Louisiana
and the United States using the
index produced by the Federal
Reserve Bank of Philadelphia.
From the figure, it is clear that
the state and national economy
do not always move together
further emphasizing the need to
have regional-specific measures
of the economy. The figure also
CESifo Forum 2/2010
4
2
0
-2
-4
-6
-8
1980
1985
1990
1995
2000
2005
Notes: The grey vertical line shows the hurricane Katrina date. The yellow areas represent NBER
recession dates.
Sources: Federal Reserve Bank of Philadelphia; NBER.
82
2009
Focus
Table 1
Peak to trough dates for Louisiana state coincident index
Peak to trough
Peak to trough
Peak to trough
Peak to trough
Peak to trough
State coincident index
–
2/1982 – 5/1983
–
3/2001 – 5/2002
1/2009 –
Official US recession
1/1980 – 7/1980
7/1981 – 11/1982
7/1990 – 3/1991
3/2001 – 11/2001
12/2007 –
Hurricane
Katrina to trough
8/2005 – 11/2005
Trough to pre-Katrina
11/2005 – 1/2007
Sources: Federal Reserve Bank of Philadelphia; NBER.
8/2005
coincident indexes may exhibit different behaviors following the hurricane, indicative of underlying structural changes in the economy. The monthly index values
may be used to track how gulf coast states co-move
before and after hurricane Katrina. One way to measure the co-movement between states is to compute
the proportion of time that the two states spend in the
same business cycle phase. The simple non-parametric
statistic known as ‘concordance’ may be used for this
exercise and is calculated as follows:
120.91 in September from 126.36 in August 2005. The
index continued to decline until its trough in
November 2005. The magnitude of the decline
equates to a 6.7 percent drop in overall economic
activity or the equivalent of an index reading from
June 1997 (November 2005 index = 117.84). Alternatively, Katrina destroyed approximately eight
years of economic progress in Louisiana. Over the
next 14 months, the index reverts back to its preKatrina level of 126.62. From trough to pre-hurricane level, it took 14 months to recover.
T
(4)
Hurricane Katrina and the other gulf coast state
coincident indexes
Ci, j =
(( S
t=1
i,t
S j,t ) + (1 Si,t ) (1 S j,t )
)
T
where T is the sample size and Si,t (Sj,t) is a binary
Figure 2 shows the state-level coincident indexes for
indicator series where the value is one when the
each of the gulf coast states: Florida, Alabama,
respective i (j) state index is expanding and zero
Mississippi, Louisiana and Texas. Using the state
when contracting.
coincident indexes to examine how the other gulf
Table 2 reports the degree of concordance between
coast states responded to hurricane Katrina, it is
Louisiana and the other gulf coast states before and
seen that only Mississippi shows signs of having been
after hurricane Katrina. Prior to Katrina, there was a
impacted by the hurricane in terms of state economrelatively high degree of co-movement among
ic activity. According to the state coincident index,
Mississippi’s economic activity
dropped 1.11 percent following
Figure 2
the hurricane. However, by the
COINCIDENT INDEX FOR GULF COAST STATES
end of 2005, Mississippi was
190
back to pre-hurricane index levFlorida
els. None of the other gulf coast
180
Texas
Alabama
states experienced significant
170
Mississippi
drops in economic activity folLouisiana
160
lowing the hurricane.
150
With the exception of Mississippi,
there does not appear to be a significant drop in economic activity
for the other gulf coast states.
However, Katrina may have altered the structure of the supply
chain and/or economic landscape
in other ways. If so, then the state
140
130
120
110
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Note: The grey vertical line shows the hurricane Katrina date.
Source: Federal Reserve Bank of Philadelphia.
83
CESifo Forum 2/2010
Focus
Table 2
Degree of concordance (Cij)
j-States
Pre-Katrina
Post-Katrina
Mississippi
0.76
0.59
Alabama
0.78
0.51
Florida
0.95
0.38
Texas
0.84
0.74
Note: The degree of concordance is the percentage of time that the two states, Louisiana and state j (i.e. j =
Mississippi, Alabama, Florida and Texas), are in the same phase.
Source: Authors’ calculation.
es in tax revenues during the recovery process in
Louisiana (Eaton 2006). A state revenue forecaster
and subsequent policymakers may be interested in
knowing if this increase in tax revenues is a onetime windfall or a longer-term gain due to the
implications on fiscal budgeting and planning. Use
of regional/state coincident indexes may shed light
on this question in a more timely fashion given the
use of monthly vs. yearly data. In the case of
Louisiana, the index rebounded to its pre-Katrina
level and thus the observed surge in tax revenues
were probably due to the purchase of replacement
goods as opposed to a permanent increase in consumer spending.
Louisiana and the other gulf coast states. Even though
Florida is not a border state with Louisiana, they tended to move together most of the time (i.e. 95 percent
of the time). However, the degree of co-movement
among the gulf coast states (with Louisiana) has
declined considerably following the hurricane. Most
notably, Florida and Louisiana are in the same business cycle phase just 38 percent of the time.
The change in co-movements between Louisiana
and the other gulf coast states is consistent with a
major shift in economic relationships. In fact, the
decline in co-movements ranges from 12 to 60 percent. In particular, this finding is in line with the earlier work on disasters altering supply chains. Thus, at
least in the relatively short term, the destruction of
the economic landscape resulting from hurricane
Katrina appears to have isolated the state of
Louisiana from the other gulf coast states. It remains
to be seen if a re-building of the region will increase
the linkages to pre-storm levels.
Economic models for assessing the impact and
response of a region to a catastrophic event should
be both timely and able to capture the broad economic condition of the affected area. However, these
two simple goals are not always met as broad measures of economic activity (e.g. GDP) are available
only at annual frequencies whereas more timely (i.e.
monthly) measures of economic performance are
often narrowly defined (e.g. unemployment).
Concluding remarks
This research demonstrated the use of the Federal
Reserve Bank of Philadelphia’s state coincident
indexes for modeling economic activity in the gulf
coast states for the case of hurricane Katrina. The
approach provides a broader and timelier estimate
as to the overall economic disruption and subsequent response to a major hurricane than the use of
state GDP or single-variable labor market indicators. The findings are encouraging and suggest that
regional composite indexes may be used to complement standard measures of economic impact and
recovery.
This paper examined the use of state level composite
indexes for providing timely information as to the
economic response of a region to natural disasters. It
is shown how the state index can capture the economic condition of a region and subsequently be
used to measure the extent and magnitude of a catastrophic event on economic activity. In particular, the
economic activity of Louisiana and the gulf coast
states were seen to have changed following hurricane Katrina. Moreover, the results are consistent
with the devastation of Katrina altering the underlying structure and/or supply chain of the region.
Policymakers can make more informed decisions
with objective or quantifiable data to facilitate a
more efficient recovery process. One such example
may be the media and news story about the increas-
CESifo Forum 2/2010
References
Crone, T. (2006), “What a New Set of Indexes Tells Us about State
and National Business Cycles”, Federal Reserve Bank of
Philadelphia Business Review, First Quarter, 11–24.
84
Focus
Crone, T. and A. Clayton-Matthews, A. (2005), “Consistent
Economic Indexes for the 50 States”, Review of Economics and
Statistics 87, 593–603.
Eaton, L. (2006), “Tax Revenues Are a Windfall for Louisiana”, The
New York Times, 26 June.
Ewing, B. and J. Kruse (2001), Hurricane Bertha and Unemployment: A Case Study of Wilmington, NC, Proceedings of the
Americas Conference on Wind Engineering.
Ewing, B. and J. Kruse (2002), “The Impact of Project Impact on the
Wilmington, North Carolina Labor Market”, Public Finance Review 30, 296–309.
Ewing, B., J. Kruse and D. Sutter (2007), “Hurricanes and Economic
Research: An Introduction to the Hurricane Katrina Symposium”,
Southern Economic Journal 74, 315–325.
Ewing, B., J. Kruse and M. Thompson (2003), “A Comparison of
Employment Growth and Stability before and after the Fort Worth
Tornado”, Environmental Hazards 5, 83–91.
Ewing, B., J. Kruse and M. Thompson (2004), “Employment
Dynamics and the Nashville Tornado”, Journal of Regional Analysis
and Policy 34, 47–60.
Ewing, B., J. Kruse and M. Thompson (2005a), “Empirical
Examination of the Corpus Christi Unemployment Rate and
Hurricane Bret”, Natural Hazards Review 6, 191–196.
Ewing, B., J. Kruse and M. Thompson (2005b), “Transmission of
Employment Shocks before and after the Oklahoma City Tornado”,
Environmental Hazards 6, 181–188.
Ewing, B., J. Kruse and M. Thompson (2009), “Twister! Employment
Responses to the 3 May 1999 Oklahoma City Tornado”, Applied
Economics 41, 691–702.
Guimaraes, P., F. Hefner and D. Woodward (1993), “Wealth and
Income Effects of Natural Disasters: An Econometric Analysis of
Hurricane Hugo”, Review of Regional Studies 23, 97–114.
Landry, C., O. Bin, P. Hindsley, J. Whitehead and K. Wilson (2007),
“Going Home: Evacuation-Migration Decisions of Hurricane
Katrina Survivors”, Southern Economic Journal 74, 326–343.
Skidmore, M. and H. Toya (2002), “Do Natural Disasters Promote
Long-Run Growth?”, Economic Inquiry 40, 664–687.
Stock, J. and M. Watson (1989), “New Indexes of Coincident and
Leading Economic Indicators”, NBER Macroeconomics Annual 4,
351–393.
The White House (2006), The Federal Response to Hurricane
Katrina: Lessons Learned, Washington DC.
Thompson, M. (2009), “Hurricane Katrina and Economic Loss: An
Alternative Measure of Economic Activity”, Journal of Business
Valuation and Economic Loss Analysis 4, Article 5.
West, C. and D. Lenze (1994), “Modeling the Regional Impact of
Natural Disasters and Recovery: A General Framework and an
Application to Hurricane Andrew”, International Regional Science
Review 17, 121–150.
85
CESifo Forum 2/2010
Focus
struction in Haiti. Indeed, with the multitude of market failures complicating the process, it is difficult to
even contemplate what an optimal reconstruction
would look like in Haitian society. At the heart of the
uncertainty lies a difficult question: in the face of
endemic disaster risk, is a society best off concentrating or dispersing its population? Had the Haitian
population been spread more widely throughout its
territory, the proportion affected by the earthquake
would have undoubtedly been lower. Concentration
of the affected population, however, introduces the
possibility of exploiting economies of scale in relief
and recovery efforts.
WHAT SHOULD THE WORLD
DO ABOUT PORT-AU-PRINCE?
AN ECONOMIC ASSESSMENT
JACOB VIGDOR*
On 12 January 2010, a major earthquake caused catastrophic damage to the capital of one of the world’s
poorest nations. Although many communities in the
southern portion of Haiti suffered grievous damage,
the devastation wrought upon Port-au-Prince was
both massive and almost instantly communicated
throughout the world. The nation with the lowest
GDP per capita in the Western hemisphere now
faces a daunting but sadly familiar dilemma – how to
rebuild in the face of vastly insufficient resources,
without reintroducing the problems that exacerbated natural risk in the first place.
This essay proceeds by describing the economic historical context of Port-au-Prince, and considering the
various market- and non-market mechanisms that
led the city to become by far the largest in Haiti. It
then offers some thoughts on the basic tasks that
should be high priorities in reconstruction, and then
some additional thoughts on the proper spatial distribution of those tasks. What emerges more clearly
than anything else is that the best prospects for an
economically rational reconstruction process hinge
on foreign cooperation. The Haitian government
enjoys – and deserves – an exceptionally poor reputation; government corruption and inefficiency is in
fact the primary reason that Port-au-Prince became
a large city in the first place. The earthquake, tragic
though it was, could provide an opportunity to remedy the consequences of decades, if not centuries, of
poor planning decisions.
Among the best tools for understanding this dilemma are those found in the economist’s toolbox. For
each individual, rebuilding decisions rest upon a
calculation of costs and benefits. As such, one could
imagine an uncoordinated, laissez-faire vision of
rebuilding, with some individuals deciding that
reconstructing their home or business in Port-auPrince offered greater value than the next best
alternative. On the other hand, though, a host of
potential market failures suggest that the decentralized solution to the rebuilding problem might
be anything but optimal. Even in the most developed economy, one would be concerned about
proper investment in infrastructure and other public goods, the potential role of externalities, and the
proper allocation of risk. In the Haitian context, we
must add to this list concerns about property rights,
underdeveloped capital markets, and the competence of government.
A brief economic history of Haiti and
Port-au-Prince
Port-au-Prince was chosen as the site of the capital
of the French colony of St. Domingue in 1738, a few
decades before Haiti earned its independence in
1804. For more than a century, the city served as one
of several transshipment points and administrative
centers in an economically undeveloped country.
Geographer Georges Anglade, writing in 1982, identified the period between independence and
American occupation in 1915 as the era of ‘regions’,
Given all these considerations, it appears highly
unlikely that a laissez-faire process will result in anything that could be considered an ‘optimal’ recon-
* Duke University.
CESifo Forum 2/2010
86
Focus
oil or natural gas reserves. Bauxite mining was a fairly prominent industry in the nation’s southeast for a
time; the nation produced 650,000 tons of ore per
year in the early 1980s and exported most of it (Tata
1982). By 2009, however, trade statistics for the
United States – Haiti’s most significant trade partner
by a wide margin – show that the nation exported
neither ores nor primary metal manufactures. The
nation’s sole metal export to the United States
amounted to USD 5,000 worth of heavy-gauge metal
tanks (US Census Bureau 2010).
where the output of coffee plantations, as well as
mahogany wood and other forest products, were sent
to one of several regional port towns for export
(Anglade 1982 as quoted in Lundahl 1992).
Although sugar cane has traditionally been an
important crop in Haiti, no sugar whatsoever was
exported during this period (Girard 2005).
Anglade (1982) describes the period after 1915 as
one of ‘centralization’, as both political power and
economic activity were increasingly concentrated in
Port-au-Prince. The typical narrative of urbanization
in initially agrarian societies begins with agricultural
surpluses sufficient to feed a population that does
not work the land. Cities begin as centers of trade,
and in some cases evolve into centers of production.
As technological change proceeds in the agricultural
sector, shifting emphasis from subsistence to export,
farm labor eventually becomes urbanized industrial
labor.
The disastrous performance of resource-based
industries in Haiti can be attributed to a combination of fate, poor decision-making and political
instability. The nation has few resources to exploit
and only 28 percent of its land area is considered
arable. Mismanagement of forestry resources and
of coffee plantations can be blamed in those sectors. The nation’s political woes have often carried
negative economic repercussions. The 1991 military
coup which deposed president Jean-Bertrand
Aristide, for example, led to a trade embargo associated with simultaneous economic contraction and
inflation.
In Haiti, the traditional narrative breaks down along
several dimensions. Agriculture and other resourcebased industries have declined, in the absence of any
clear technological improvements. Haiti has been
almost entirely deforested for decades. Subsistence
agriculture continues to play a strong role in the
economy, while the production of cash crops has
dwindled. Coffee has long been the nation’s most
important cash crop; however poor agricultural practices have earned the sector’s output a reputation as
a low-quality product. In 2009, Haiti exported just
USD 35,000 worth of coffee to the United States,
while simultaneously importing USD 120,000 worth
(US Census Bureau 2010).
Consistent with the traditional narrative, the rise of
Port-au-Prince can be attributed in part to the emergence of a manufacturing sector in that city. The
nation’s apparel industry, which accounts for some
93 percent of the value of exports to the United
States, is concentrated in Port-au-Prince. Upon closer inspection, though, the role of industry in the capital’s expansion is quite minor. Haiti’s manufacturing
sector overall is astonishingly small. The nation’s
total exports to the United States in calendar year
2009 amounted to about 550 million US dollars. By
contrast, the Dominican Republic, which occupies
the same island and has a population of comparable
size to Haiti, exported 3.3 billion US dollars to the
same partner in the same year. For every dollar in
income received through exports, Haiti receives
about USD 2.50 in remittances. In a very real sense,
then, Haiti’s most significant export is labor.
In the 1960s and 1970s, Haiti produced over 4 million
tons of sugar cane per year. This output increased in
the 1980s. Sugar refineries were present in several
cities, and collectively produced tens of thousands of
metric tons around this time. In more recent years,
though, the sugar industry has essentially vanished.
Haiti exported no sugar to the United States in 2009,
while importing USD 378,000 worth. In 2009, Haiti’s
primary agricultural exports to the United States
were fruit – predominantly mangoes – and tree nuts.
Total exports of agricultural products to the United
States amounted to 16.5 million US dollars, against
USD 53 million worth of agricultural imports (US
Census Bureau 2010).
Port-au-Prince’s emergence as Haiti’s leading manufacturing center cannot be attributed to any natural
advantage.1 Fabric used to create apparel is almost
1 Puzzlement at the location of the capital dates back at least two
centuries. British historian Marcus Rainsford, writing in 1805,
opined that “It must have been one of the unaccountable caprices
that sometimes direct the settlement of towns, that could have
obtained for [Port-au-Prince], indefensible at all points, the distinction it received” (Heinl and Heinl 1978, 31).
Even setting agriculture aside, Haiti’s resource
endowment is quite poor. The nation has no proven
87
CESifo Forum 2/2010
Focus
Thirty-five years into Anglade’s period of centralization, Port-au-Prince was by any measure a modestly sized city. In 1950, at the time of Haiti’s first
official census, the Port-au-Prince metropolitan
area, which at that time essentially consisted of the
city itself and the nearby suburb of Petionville,
counted 144,000 inhabitants in a nation of just over
3 million (Institut Haïtien de Statistique et
d’Informatique, Division d’Analyse et de Recherches Démographiques 1983). Nearly half of
these residents had been born elsewhere in the
country (Lundahl 1992). Seven of every eight
Haitians lived in rural areas at that time, the vast
majority engaged in subsistence agriculture. Haiti
in 1950 was still a quintessential pre-industrial
society.
exclusively imported. While the city does have container port facilities, it is not necessarily the most
obvious choice for shippers. Cap Haitien, on
Hispaniola’s northern coast, is closer by sea to
Miami, and during the era of bauxite production the
port at Miragoâne handled a larger volume of
exports.
If not for natural advantages, what explains Port-auPrince’s rise to primacy in manufacturing? In a word,
the proximate cause is infrastructure. Tata (1982, 64)
remarks that Port-au-Prince in the early 1980s was
“the only real choice for a modern manufacturing
plant”. Port-au-Prince possessed the nation’s only
runway longer than 1,600 meters, the only modern
dock facilities and the most reliable electricity generation.
By the time of the nation’s third Census, in 1982,
the metropolitan region counted 720,000 inhabitants in a nation of just over 5 million – over
32 years, the share of national population in the
capital city rose from under 5 to over 14 percent. In
the early 1980s, the Port-au-Prince region was home
to two-thirds of Haiti’s manufacturing establishments and over 90 percent of all manufacturing
employees (Tata 1982).
These infrastructural advantages, in turn, reflect the
investment decisions of Haitian government. The
tendency for population to cluster in the capital
cities of nations with unstable governments is both
general and widely noted (Ades and Glaeser 1995).
Several hypotheses explain the link. When governments are corrupt, proximity to the seat of power
may facilitate business and rent-seeking – indeed,
Tata (1982) notes that the ease of acquiring necessary government licenses as one factor leading
Haitian manufacturing towards Port-au-Prince.
Governments may favor infrastructure investments
in the capital because the hinterland is weak, or
because access to superior infrastructure helps
guard against uprisings in outlying areas. Ultimately,
then, the emergence of Port-au-Prince as a manufacturing hub is far more clearly attributable to
these political considerations than to any innate
locational advantage.
In the early 1980s, the Haitian government evaluated the prospects of establishing industrial facilities
in the outlying population centers of Cap Haitien
and Les Cayes (Tata 1982). These efforts did not
lead to any break in the trend toward primacy in
Port-au-Prince. Population estimates from 2009 suggest that the capital region was home to about
28 percent of the nation’s population. Over the six
decades between the nation’s first census and the
devastating earthquake of 2010, the Port-au-Prince
region grew at an average annual rate of 4.8 percent,
more than twice the growth rate of the population
as a whole. This era witnessed the emergence of suburban slums that would become household names in
the aftermath of the quake: Cite Soleil, established
in the 1950s, Carrefour, a former coastal tourist destination and artist colony that tripled in population
between 1982 and 2009, and Delmas, which was not
even enumerated in the 1982 Census but was home
to 359,000 people by 2009.
The greatest share of employment in the Port-auPrince region associates not with manufacturing,
but with services. The rapidly growing metropolis
has supported a significant construction sector,
transportation and communication services, education and health care, and banking and insurance.
Once again, these sectors have gravitated towards
Port-au-Prince not because of any innate advantage
in providing them there, but because they must
closely track the distribution of the population
more generally, and must appeal to the same government-allocated resources that drive industrial
location decisions. Haitian service industry has followed people to Port-au-Prince, not the other way
around.
CESifo Forum 2/2010
The earthquake of 12 January 2010 wrought destruction on a city that owed its existence, above everything else, to the actions of government. The region’s
infrastructure, which made Port-au-Prince the only
feasible location for most export-oriented activity,
88
Focus
minican program implemented in the 1960s: the distribution of subsidized natural-gas cook-stoves
(Oppenheimer 2010a). Of course, in a nation with no
natural gas reserves, such a program would only
exacerbate an already burdensome reliance on
imported fuel. The deforestation of Haiti represents
a classic negative externality problem, however, and
actions to correct it would almost certainly pay dividends in the long run.
was decimated. Indeed, the destruction of critical
infrastructure undeniably worsened the human toll
of the earthquake, as facilities to offload cargo containers were rendered unusable, and the singlerunway Port-au-Prince airport was quickly overwhelmed.
Port-au-Prince’s population has dispersed into a
countryside where subsistence agriculture is by and
large the only economic activity – the sustainability
of which is severely threatened by a combination of
poor prior practice and vulnerability to further natural calamity. The country is now in a precarious
position, and does not possess the resources to pull
itself up by its own bootstraps.
What Haiti needs more than any commodity, however, is a stable government. A stable, democratic, representative government would face stronger incentives to properly guide investments of the nation’s
own public resources. The government’s past reputation for corruption and incompetence serves as a
substantial impediment to policy and planning in the
wake of disaster. A stable government would not
only credibly guide the actions of private citizens, but
also encourage investment by foreign firms.
Reinvesting in Haiti: what needs to be done?
Haiti’s precarious position prior to the earthquake
creates great uncertainty regarding the pace and
form of recovery. From an economic perspective,
there are several important questions to ask about
the recovery process. What are the most sensible
investments to make in Haiti? Should these investments be geographically concentrated, to mirror the
economic situation in 2009, or would it make more
sense to disperse them? There are a few commonsense answers to the first question, but the second
one is actually quite difficult.
Unfortunately, stable government is perhaps the
most difficult thing to bequeath a nation. The presence of a United Nations Stabilization Mission in
Haiti since 2004 has undoubtedly helped to stabilize
the country, but preventing armed conflict is not
exactly the same as ensuring stability in government.
Foreign intervention may need to be a significant
aspect of Haitian governance for many decades to
come.
Haiti has, in fact, experienced two periods of intense
foreign intervention in the past century. The first
American occupation, between 1915 and 1934, yielded incredible investments in infrastructure. A country with 5 kilometers’ worth of highway in 1915 had
nearly 2,000 by the end of occupation. Irrigation
canals dating to the colonial era were renovated, and
investments in urban water supply and hospitals
improved health. Most strikingly, these investments
were accomplished largely without the use of foreign-source revenue. The occupiers diverted the flow
of domestic revenue from the pockets of bureaucrats
to infrastructure (Girard 2005). Much of this infrastructure crumbled after the end of occupation, however, for want of maintenance. By 1971, three-quarters of the American-built road infrastructure was no
longer passable in all weather conditions (Heinl and
Heinl 1978).
Some of the most sensible investments in Haiti
would undo the damage inflicted by the mistakes of
earlier eras. Deforestation has created incredible
problems for the country; the soil has been damaged
by erosion and has greater difficulty retaining rainwater. As a consequence, hurricanes are now more
likely to lead to catastrophic flooding and mudslides.
A proposal to allocate United States government
resources to the reforestation of Haiti was introduced in the Senate in June 2009, and to the House
of Representatives in December, but these legislative bodies have taken no definitive action on the
proposals.
Programs of reforestation have been tried in the
past, without substantial success. Legal uncertainties
regarding land ownership have impeded efforts
(Oppenheimer 2010b). The widespread use of wood
for cooking fuel leads to situations where trees are
consumed as rapidly as they are planted. The finance
minister of the Dominican Republic, Morales
Tonosco, has suggested the replication of a Do-
The first American occupation had its flaws. The
occupying force was entirely white and insensitive to
the country’s racial history. Many roads were built
89
CESifo Forum 2/2010
Focus
with unpaid labor, as the occupiers enforced a law on
the Haitian books since the 1860s, which required all
citizens to contribute to road-building with either
taxes or physical work. Investments in educational
institutions were also not well-aligned with the needs
of a largely illiterate, Creole-speaking population
(Girard 2005).
city enjoyed a rail connection to Port-au-Prince,
though Haiti is now entirely devoid of rail service. St.
Marc now serves as a minor port, with a small wharf
in the city center. The harbors of Port-au-Prince, Cap
Haitien and Miragoane are naturally deeper and
thus more amenable to modern shipping, but each
lies in a more seismically active zone.2
The second American occupation of 1994–95, which
had the primary goal of re-establishing Jean
Bertrand Aristide as president, earned much higher
marks in terms of cultural sensitivity but accomplished comparatively little in the way of investment. The ideal form of foreign intervention would
couple the first American focus on efficient investment with the second American attention to cultural sensitivity.
St. Marc is hemmed in by modest 200m hills. Just
10 km by road across those hills, however, lies the flat
and sparsely inhabited plain of the Artibonite river,
where it is not difficult to imagine the creation of an
entire city, replete with airfields and other commercial amenities.3
Even if building an entirely new city were cheaper
than rehabilitating a devastated one, however, the
question remains whether the concentration of population in one city is in any sense optimal. Urban
economists have devoted much attention to the
question of optimal city size. Larger cities are desirable to the extent that they permit the realization of
economies of scale in labor-intensive industries, and
create other benefits associated with greater specialization in production and consumption, or agglomeration economies rooted in knowledge spillovers,
labor market pooling, or other mechanisms. On the
other hand, larger cities incorporate diseconomies of
scale associated with congestion, pollution, and other
maladies. The ‘optimal’ city size equates the marginal costs and benefits of population expansion.
Where should reinvestment take place?
The notion that redevelopment efforts should aim to
disperse population away from Port-au-Prince is as
old as the nation itself. Jean-Jacques Dessalines, the
self-styled first emperor of Haiti, reportedly desired
to abandon the city (Heinl and Heinl 1978). These
plans were lost with his assassination in 1806. More
recently, Georges Anglade concluded his 1982 Atlas
of Haiti with a call to decentralize the population – a
call that echoed the government’s own study of
development outside the capital. Since that time, of
course, the proportion of Haitians living in Port-auPrince has doubled.
In the case of an impoverished and seismically
unstable nation such as Haiti, the creation of a large
city incorporates risks that became all too apparent
in the aftermath of the earthquake. In a society
where vehicle ownership is rare, and public transportation networks primitive, urban life requires
high population density. The costs of seismic protection increase much more than linearly with density.
The cost of seismically retrofitting a large concreteframe structure in the United States, on a persquare-meter basis, are more than twice that of
retrofitting a smaller structure of the same vintage.4
Rather than face these higher costs, Haitian builders
As noted above, there is no natural reason for Portau-Prince to be the center of export activity. Given
the devastation to the capital region’s infrastructure, it might well be less expensive in the long-run
to develop the region around an alternative port.
Cap Haitien, the most frequently cited alternative
port facility, is itself in a seismically active zone,
afflicted by an earthquake most recently in 1842,
and lies along a fault that some consider overdue
for a major quake. From a seismic perspective, the
safest portion of the country lies near the geographic center, north of Port-au-Prince along the
Golfe de la Gonâve.
2
As of 1982, the harbor depths in Port-au-Prince, Cap Haitien,
Miragoane and St. Marc were 9.8, 6.4, 5.4 and 3.6 meters, respectively (Tata 1982). The wharf at St. Marc extends only 70 meters
from the adjacent shoreline; construction of a larger-scale wharf
could potentially ameliorate the depth problem.
3 While the Peligre dam has provided the Artibonite valley with
some measure of flood control since 1956, further infrastructure
investments would be required to assure flood protection and
water supply to any large-scale habitation.
4 Estimates can be derived from the US Federal Emergency
Management Administration’s Seismic Retrofitting Cost
Calculator.
The small city of St. Marc is perhaps the most
promising location in this zone. St. Marc traces its
origins to the colonial era. Its immediate environs
have not been directly affected by any recorded
earthquake since the 17th century. At one time, the
CESifo Forum 2/2010
90
Focus
dren of Haitians born elsewhere, and the fact that
many residents are now relocated anyway, because
of the general disarray in Port-au-Prince. Absent a
concerted effort to change citizens’ expectations
regarding the future locus of economic activity in
Haiti, however, the force of inertia will likely be
strong.
and their clients accepted the risk of catastrophic
collapse. From a seismic perspective, it would be
much more cost-effective to redistribute the residents of Port-au-Prince into a series of smaller-scale,
lower-density cities.
Such a redistribution would introduce other types of
costs, however. To make Haiti’s new hinterland cities
attractive to potential employers, these cities would
need to be connected to reliable transportation and
electric grids. The absence of these networks was a
major factor in the emergence of Port-au-Prince as a
prime destination in the first place. This type of
tradeoff between economies and diseconomies of
scale is inevitable.
In the end, the hypothetical tradeoffs between
economies and diseconomies of scale are difficult
to conclusively reconcile. What is more clear, however, is that the location of Port-au-Prince is seismically unfortunate and rebuilding the city in a way
that mitigates risk of future destruction would be
quite costly. Even clearer than this conclusion, however, is the reality that organized, planned recovery
in a nation as chaotic as Haiti will be impossible
without concerted efforts on the part of the developed world.
Catastrophic but localized risk, such as that associated with hurricanes and earthquakes, introduces
another set of arguments into the dispersal issue.
Dispersing the population could be seen as a simple
risk diversification policy: the best way to avoid a
human disaster on the scale of what was seen in Portau-Prince would be to not have cities as large as
Port-au-Prince. The aggregate risk faced by the population might not change much, or could even
increase. The correlation in risk across persons
would decline, however.
References
Ades, A. F. and E. L. Glaeser (1995), “Trade and Circuses:
Explaining Urban Giants”, Quarterly Journal of Economics 110,
195–227.
Anglade, G. (1982), Atlas Critique d’Haïti, Montreal: Centre de
Recherches Caraïbes de l’Université de Montréal.
Girard, P. R. (2005), Paradise Lost: Haiti’s Tumultuous Journey from
Pearl of the Caribbean to Third World Hot Spot, New York: Palgrave
Macmillan.
From another perspective, though, the correlated
risk introduced by population concentration could
bring countervailing benefits, to the extent that there
are economies of scale in disaster recovery. For all its
flaws, Port-au-Prince benefited from its close proximity to deep water and the nation’s longest airstrip.
To decentralize the population would be to increase
the average citizen’s distance from these facilities. In
a wealthier country, this disadvantage would be offset by the creation of redundant infrastructure:
building more airfields and port facilities outside the
capital. Given the scarcity of available funds and
more pressing priorities, these sorts of investments
are unlikely to take place anytime soon.
Heinl, R. D. Jr. and N. G. Heinl (1978), Written in Blood: The Story
of the Haitian People, 1492–1971, Boston: Houghton Mifflin.
Institut Haïtien de Statistique et d’Informatique, Division
d’Analyse et de Recherches Démographiques (1983), “Le
Recensement Haïtien de 1982”, Population 38, 1055–1059.
Lundahl, M. (1992), Politics or Markets? Essays on Haitian
Underdevelopment, London: Routledge.
Oppenheimer, A. (2010a), “In Haiti, Reforestation Should Be Part
of Rebuilding Process”, Miami Herald, 31 January.
Oppenheimer, A. (2010b), “Commentary: Reforestation of Haiti
Gets Off to a Slow Start”, Miami Herald, 22 March.
Tata, R. J. (1982), Haiti: Land of Poverty, Lanham, MD: University
Press of America.
United States Census Bureau (2010), U.S. International Trade in
Goods and Services Report,
http://www.census.gov/foreign-trade/data.
Even if a concrete plan for population dispersal were
adopted, it may prove difficult to coordinate the
actions of independent households and firms to settle in alternative locations. Government could take
certain actions to encourage resettlement: improving
the infrastructure in regional towns, perhaps even
moving some government functions – or the entire
operation – away from Port-au-Prince. The prospects
for relocation are improved by the fact that many
residents of the capital are the children or grandchil-
91
CESifo Forum 2/2010
Focus
when aircraft engines were compromised without
any reported airplane crash, however. The interest of
the mass media initiated by this type of event,
together with the raised risk perception of travellers,
may also alter economic behaviour. Thus, while flight
operations rapidly go back to normality, it is likely
that potential air passengers (for example, holiday
makers) decide to cancel their trip or alter their
plans (e.g. change the mode of transport or the destination country). When psychology comes into play,
an accurate estimation of economic outcomes is
even harder. Travelling with the worry that your
flight schedule might be disrupted or – worse – that
your aircraft engines might be damaged could by
itself generate a welfare loss, even if the trip and aircraft are not affected. The literature on the response
to risk information (see e.g. Becker and Rubenstein
2004) also shows that economic agents invest their
money to take countermeasures, thereby generating
profits for other economic agents. For example, air
passengers may opt to take a train, rent a car or stay
in a hotel, with different costs and economic outcomes, including (possibly) additional income for
railways, car rental agencies and hotels.
THE 2010 VOLCANIC ASH
CLOUD AND ITS FINANCIAL
IMPACT ON THE EUROPEAN
AIRLINE INDUSTRY
MARIO MAZZOCCHI,
FRANCESCA HANSSTEIN AND
MADDALENA RAGONA*
Introduction
Last April, European air traffic was heavily disrupted by the volcanic ash cloud generated by the eruption of the Icelandic volcano Eyjafjallajokull. Even
though the explosion was of low intensity, it produced an enormous cloud of ash moving through the
European skies. The fact that the ash was much finer
than usual, moving quickly and possibly affecting
aircraft engines, led the aviation authorities of the
concerned countries to declare most of European
skies as no-fly zones (NFZs). Based on the initially
available information, there were claims of a huge
economic impact on the air travel industry, even bigger than the direct impact of the US air traffic halt
after the terrorist attacks on 11 September 2001
(European Commission 2010).
A further reason why it is difficult to estimate the
overall economic impact is the number of offsetting
factors that need to be considered. Accountability is
quite complicated for airplane and airport industries, due to the weight of offsetting effects as, for
example, the tonnes of fuel saved or parking lots
filled. The International Air Transport Association
(IATA) has estimated that the airline industry uses
around 4.3 million barrels of fuel per day. At the
peak of the airspace closure, the demand for fuel is
estimated to have fallen by 1.2 million kerosene barrels per day (about 110 million US dollars saved)
(IATA 2010).
It is obviously difficult to obtain accurate estimates
of the overall economic impact imputable to a natural disaster like this. Besides the unpredictable
behaviour of nature – in this case not only the eruption but also weather conditions – one should consider the adaptive behaviour of people, whose complexity increases with the number of actors involved.
For example, the threshold at which flying was
admitted was raised by the relevant authorities after
five days of air disruptions, a decision which is likely
to have softened the potential impact.
There are a few industries that could have benefitted
from the volcano’s eruption, at least in the short
term, especially those linked to alternative modes of
transport, like car rentals and railways. Eurostar
reported that it carried 50,000 extra passengers on
15 April, and registered an increase of 33 percent on
17 April. P&O Ferries of France declared that their
Restrictive measures that caused the closure of the
greatest part of European skies had been established
on the basis of two previous accidents in the 1980s,
* University of Bologna.
CESifo Forum 2/2010
92
Focus
tions, the key results will be discussed and some conclusions drawn.
services between Britain, Spain, France and the
Netherlands were fully booked and that they
employed extra personnel to handle the huge volume of phone calls to their info centres.1 On the
other hand, other industries besides airlines have
been negatively affected by the disruption: for example, courier services, air cargo businesses or industries that rely on air transport for trading perishable
commodities. The European Commission has pointed out that air cargo traffic suffered a fall of 61 percent within the EU27 between the scheduled flights
the week before and the ‘ash days’ (compared to a
decrease of 64 percent in passenger traffic) (European Commission 2010).
Economic impact on the airline industry
As it usually happens with crises of this scale, a number of attempts at quick – albeit rough – quantification of the magnitude of the economic impact have
been produced. Right after the eruption, the majority of estimates were provided by the main airline
association, the IATA and the European Commission. Table 1 reports some of these quantifications,
all referring to the week following the first closure of
airspace.
A (partial) shortcut to the all but impossible task of
estimating all those impacts is an analysis of financial markets. Assuming that investors make the best
use of available information and base their purchasing or selling behaviour on their rational economic
assessment (which rules out speculative behaviours), returns on equities may be seen as the ‘thermometer’ of the current and predicted performance
of an individual industry. In this way one might be
able to account for both the losses suffered by the
air industry and the potential gains of other stakeholders. A comprehensive analysis is still impossible, at least in the short term. We cannot evaluate
the welfare loss of stranded travellers or the productivity losses of all firms indirectly affected by the
ash crisis. Another strong assumption is that the
behaviour of listed securities is also representative
of industries not listed on stock exchanges, such as
small and medium-size enterprises. However, the
analysis of stock returns seems the easiest and most
reasonable way to explore the magnitude of such a
complex event.
In terms of operations, on 17-18 April 2010, 17 EU
Member States had a full airspace closure, 2 Member
States had a partial closure and 6 non-EU States also
decided on a full closure. On April 22, airspace was
fully operational again, apart from a partial closure in
southern Finland (European Commission 2010). In
terms of passenger flows, the biggest domestic markets affected by the closure were Britain, France and
Germany, while the biggest decline of airline revenues was due to the cancellation of US-UK flights.
In terms of economic impact, the revenue loss for
airlines from scheduled services was estimated at
1.7 billion US dollars (this figure is considered ‘conservative’) during the period of 15-21 April 2010. The
revenue loss per day varies according to the daily
airspace closure, and reached 400 US dollars per day
during the peak period (17-19 April). During the five
days of disruption, British Airways reported a loss of
20 million British pounds per day as did Air-France
KLM.2
In this paper, we will provide a first basic estimate of
some of the economic effects of the Eyjafjallajokull
volcano’s eruption, focusing on the airline industry,
together with some evaluations of potential gainers,
namely alternative transport industries. Toward this
end we employ a basic event study analysis
(MacKinlay 1997) with the most recent available
data, in order to provide some estimates of financial
losses that may be ascribed to the volcanic ash cloud.
After giving an overview of initial attempts to estimate the economic consequences for the airline
industry, we will briefly summarize the basic procedure for conducting an event study. In the final sec-
1
The classical event study method
A rapid assessment of the financial impact of the
volcanic ash cloud may be made with an event study
analysis (MacKinlay 1997). The method basically
ascribes to specific events significant deviations of
individual returns on equities from the overall market trend, based on the ‘ordinary’ behaviour of share
prices. For this purpose, one may benchmark ordinary behaviour using the market model, which
assumes a linear relationship between returns on
individual securities and the market return:
2
From BBC News, www.bbc.co.uk.
93
From The Economist, www.economist.com.
CESifo Forum 2/2010
Focus
Table 1
Estimated effects of the volcanic ash cloud on the airline industry (15–23 April 2010)
Source
ACI Europe
Outcome
313 airports
IATA
100,000 flights
EUROCONTROL
Source
IATA
IATA
European airports totally disabled (75% of the European
Airport Network)
Flights cancelled within the EU, to/from the EU and
overflying the EU
Peak of flights cancelled on 18 and 19 April
Estimated passengers unable to travel
Average of scheduled passengers affected each day
Reduction of the within-Europe and Europe-rest of the
world passenger flows
19,000 flights
10 million passengers
1.2 million passengers
24% (and 9%
worldwide) passengers
flow reduction
Economic impact
US$ 1.7 billion
US$ 400 million
Revenue loss for airlines during the period 15-21 April
Per day revenue lost for airlines over the peak period
(17–19 April)
AEA
850 million
Loss for airlines including profitability, assistance to
passengers, costs for stranded crew, parking and
positioning of aircraft and other cost issues (for the
period 15–23 April)
ERAA
110 million
Estimated loss for members of ERAA
ELFAA
202 million
Estimated loss for members of ELFAA
IACA
310 million
Estimated loss for members of IACA
ACI Europe
250 million
Overall European airports losses
IAHA
200 million
Direct financial loss for independent handlers pertaining
to the IAHA
ANSPs
25 million
Loss per day for Air Traffic Management (ATM)
EC
61%
Fall in air traffic cargo between the scheduled flight per
week in the EU-27
Notes: ACI: Airport Council International; IATA: International Air Transport Association; AEA: Association
of European Airlines; EUROCONTROL: European Organisation for the Safety of Air Navigation; ERA:
European Regions Airline Association; ELFAA: European Low-Fare Airlines Association; IACA:
International Air Carrier Association; IAHA: International Aviation Handlers Association; ANSPs: Air
Navigation Server Providers and EC: European Commission.
Source: Websites of the sources indicated in the first column and listed above.
Rit = i + i RMt + uit .
conditioning on the market return, and allows to predict the expected return had the event not occurred.
Even if tests of individual securities and time periods
are possible, these are less powerful compared to a
test based on aggregations of time periods and/or
securities. Once the parameters α and β are estimated, the normal returns may be predicted over the
event window and are computed as:
(1)
where Rit is the return of a security at time t, RMt is
the market return and uit are independent normally-distributed residuals. The benchmark model is
estimated on a sample prior to the event, which
should not be affected by other major security-specific events. Then the ordinary behaviour is projected through a time window after the event occurs
(the ‘event window’) and significant deviations
from the ordinary behaviour are detected based on
tests of the excess returns with the following null
hypothesis:
H 0 : Rt = E ( Rt RMt )
with
with
t ,
E ( Rit RMt ) = ˆ i + ˆi RMt
t .
(3)
So that excess returns correspond to the forecast
error uit and are computed over the event window
as
(2)
uit = Rit ˆ i ˆi RMt
where τ is the event window set which covers the
period following the occurrence of the event and
E(Rt|RMt) is the ‘normal’ return, which is obtained by
CESifo Forum 2/2010
with
with
with t with
(4)
Under the null hypothesis these results are normally
distributed with the following standard error (Patell
1976):
94
Focus
2
(
R
R
)
1
M
= i Cit
uit = i2 1 + + T Mt
T
2 (
)
R
R
Mj
M
j =1
Application and results
(5)
For this event study, we make use of Datastream
daily data on share prices for selected securities on
the London, Frankfurt, Paris and Stockholm stock
exchanges. The estimation window is chosen to be
relatively small (100 observations from 24 December
2009 to 14 April 2010) in order to minimize the risks
of major structural changes associated with the economic crisis and to emphasize the short-run dynamics.3 Nine airlines (7 flag carriers and 2 low-cost companies) were included, considering their listing on
the most relevant stock exchange. We also selected
six potential gainers: five car rental companies and
Eurotunnel, the company which runs Eurostar trains
and the Eurotunnel. Our assumption – partially confirmed by the results – is that these companies may
have benefitted from a decrease in airplane transport by an increase in the demand for car rental and
train services. The selected companies and the reference stock exchange are listed in Table 2.
with t with
where σi is the regression standard error of the i-th
–
security, RM is the average market return over the
estimation period, T is the number of observations in
the estimation sample and Cit (i.e. the expression in
brackets) is the variance inflation factor due to prediction outside the estimation period. A test for the
null hypothesis for individual securities and individual time periods of the event window is provided by
the Patell Standardised Residuals (PSRs):
zil =
uil
uil Cil
:G N (0,1)
where l .
where
(6)
The period affected by the ash cloud ran from
15 April to 20 May (day when European skies were
declared ‘ash-free’). A detailed timeline of events
and effect on air operations is provided in Table 3.
Based on the events listed in Table 3, several event
windows (EW) were explored, as summarized in
Table 4.
For large T it is possible to obtain aggregated tests
over the L time periods of the event window τ,
across S securities of a given set ψ or across both τ
and ψ, as follows:
L
zi 0 =
zSl =
u
l =1
il
i LCil
~ N (0,1),
uSl
~ N (0,1)
SE (uS ) CSl
(7)
Despite using the basic event study approach, results
were econometrically robust to choices on the size of
estimation and event windows. Figure 1 shows the
day-by-day Patell Standardised Residuals (PSRs)
aggregated over the two groups of firms through the
overall event window running from 15 April to
20 May, computed as indicated in equation (8).
Significant abnormal returns at the 95 percent confidence level are those below or above the two horizontal lines. The aggregate value of the airlines (blue
line) is below the zero line over most of the considered period, although it is only significant during the
first peak period (17–19 April), on 28 April, over the
second ‘wave’ of ash (5–7 May) and with the last
occurrence of the ash cloud (around 17 May). On
21 April there is a positive significant return, a
‘rebound’ after the negative peaks of previous days.
In contrast, in the aggregate, evidence on positive
returns for potential gainers is quite weak, with positive values during the first few days (only significant
on 15 April).
(8)
and
L
zS 0 =
u
l =1
Sl
~ N (0,1),
L
SE (uS ) CSl
(9)
l =1
where
l , uSl =
1
1
uil , uS = uSl , SE (uS ) =
S i
L l
1 T
(uSt uS )2
T 1 t =1
.
3
However, estimation windows with 200, 300 or 400 observations
produced similar results.
95
CESifo Forum 2/2010
Focus
Table 2
Selected securities and reference market returns
Security
Aerlingus
Air France – KLM
British Airways
Easyjet
Finnair
Iberia
Lufthansa
Ryanair
SAS
Avis Europe
Avis Budget Group
Eurazeo
Eurotunnel
Hertz
Sixt
Source: Datastream.
Reference market return (Stock Exchange)
Airlines
FTSE-All (London Stock Exchange)
CAC (Paris Stock Exchange)
FTSE-All (London Stock Exchange)
FTSE-All (London Stock Exchange)
DAX (Frankfurt Stock Exchange)
DAX (Frankfurt Stock Exchange)
DAX (Frankfurt Stock Exchange)
FTSE-All (London Stock Exchange)
OMX (Stockholm Stock Exchange)
Potential gainers
DAX (Frankfurt Stock Exchange)
DAX (Frankfurt Stock Exchange)
CAC (Paris Stock Exchange)
CAC (Paris Stock Exchange)
DAX (Frankfurt Stock Exchange)
DAX (Frankfurt Stock Exchange)
Table 3
Timeline of events associated with the ash cloud and effects on flights
Date
Actual Estimated
EUROCONTROL’s expectations
flights flights in a
declared each day (summary)
normal day
Wed 14 April 28,087
28,000 None
Thu 15 April 20,842
28,000 None
Fri 16 April 11,659
28,000 Airspace is not available for operation of civilian aircraft in the following
countries/areas: Ireland, UK, Belgium, Netherlands, Denmark, Sweden,
Norway, Finland, Estonia, northen France, parts of Germany, parts of Poland.
Forecasts suggest that the cloud of volcanic ash is continuing to move east
and south-east and that the impact will continue for at least the next 24 hours.
Sat 17 April 5,335
22,000 No landings and take-offs are possible for civilian aircraft across most of
northern and central Europe: Austria, Belgium, Croatia, Czech Rep.,
Denmark, Estonia, Finland, northern France, most of Germany, Hungary,
Ireland, northern Italy, Netherlands, southern Norway, Poland, Romania,
Slovakia, Slovenia, Sweden, Switzerland and UK. Forecasts suggest that the
cloud of volcanic ash will persist and that the impact will continue for at least
the next 24 hours.
Sun 18 April 5,204
24,000 Air traffic control services are not being provided to civil aircraft in the major
part of European airspace: Austria, Belgium, Croatia, Czech Rep., Denmark,
Estonia, Finland, most of France, most of Germany, Hungary, Ireland,
northern Italy, Netherlands, Norway, Poland, Romania, Serbia, Slovenia,
Slovakia, northern Spain, Sweden, Switzerland, Ukraine and UK.
Mon 19 April 9,330
28,000 Air traffic control services are not being provided to civil aircraft in the major
part of European airspace: Belgium, Czech Rep., Denmark, Estonia, Finland,
parts of France, Germany, Hungary, Ireland, Netherlands, northern Italy,
Poland, Romania, Slovenia, Switzerland, parts of Ukraine and UK.
Tue 20 April 13,101
28,000 The new procedures agreed yesterday have been in place since 6.00 UTC. Air
traffic control services are not being provided to civil aircraft, or are being
provided with significant restrictions, in the lower airspace in north-western
Europe: Denmark, Estonia, Finland, northern France, northern Italy, Latvia,
Slovenia, Slovakia and UK. In the upper airspace above 20,000 feet, all
European airspace is available. In the evening almost 75% of the total
continent area is free of any restrictions.
Wed 21 April 21,916
28,000 All European airspace is available above 20,000 feet. Below 20,000 feet,
restrictions are still in force in a few areas (southern Sweden, part of Finland,
parts of Scotland). It is anticipated that these restrictions will gradually be
lifted throughout the day. It is anticipated that almost 100% of the air traffic
will take place in Europe tomorrow.
CESifo Forum 2/2010
96
Focus
Thu 22 April 27,284 28–29,000 A small number of cancellations can be expected due to some limited
restrictions and the logistical problems of airlines. Almost all European
airspace is available, with a few exceptions in parts of southern Finland,
southern Norway, northern Scotland and western Sweden.
Fri 23 April
29,000 Almost all European airspace is available, with the exception of part of
northern Scotland.
Wed 28 April Normal Normal The Ash Concentration Charts produced by VAAC London show that there
has been no area of high potential volcanic ash coverage within the CFMU
area for several days now.
Tue 4 May
28,000 Airspace in Ireland, northern Ireland and small parts of western Scotland was
closed between 8.00 and 14.00 CET (cancellation of some 150 flights). The
latest Ash Concentration Charts show that the area where ash concentrations
could exceed engine manufacturer tolerance levels has shrunk and is no
longer affecting any substantial part of European airspace. This situation is
expected to remain stable for the coming hours.
Wed 5 May 27,904
29,000 Several Irish airports will be closed for limited hours. Edinburgh is
currently operating at reduced capacity and the western part of Scottish
airspace is closed. The situation is not expected to improve in this area during
the day. The whole of Ireland, western Scotland and north-western England
could be affected. Greek airspace is also closed for all traffic as a result of
industrial action.
Thu 6 May
30,202
28,500 No closures of airspace or airports within the Europe. The predicted area
where ash concentration could exceed engine manufacturer tolerance levels
lies to the west/north-west of Ireland. In the night of 5 to 6 May, renewed and
more intensive ash eruptions took place.
Fri 7 May
Sat 8 May
Sun 9 May
Mon 10 May
Tue 11 May
Wed 12 May
Thu 13 May
Fri 14 May
Sun 16 May
30,342 28–29,000 Some airports were closed in western Ireland overnight. The main predicted
area where ash concentration could exceed engine manufacturer tolerance
levels lies to the western part of North-West Europe. Renewed and more
intensive ash eruptions took place overnight, and the area of potential higher
ash contamination is forecast to extend from Iceland as far south as the
western edge of the Iberian Peninsula during the day. Transatlantic flights are
being re-routed south of the affected area which could cause delays to these
flights.
22,424
22,600 Ash eruptions are ongoing. Airports are closed or expected to close in
northern Portugal, northern Spain and parts of southern France. Transatlantic
flights are being re-routed around the affected area which is causing
substantial delays to these flights.
23,491
25,000 Ash eruptions are still substantially affecting European airspace. Airports in
northern and central Portugal, north-western Spain, northern and central Italy
are unavailable, and are expected to open later. Transatlantic flights continue
to be affected by the ash cloud (re-routings, delays).
29,155
29,000 Areas of high ash concentration have dispersed overnight over continental
Europe. There is an area of ash cloud in the middle of the North Atlantic
impacting transatlantic flights (re-routings, delays). No airports are closed in
Europe. During the afternoon, areas of higher ash concentration could move
in a north-easterly direction from the Atlantic into the Iberian Peninsula.
27,807
29,000 Airports on the Canary Islands, some in south-west Spain and some in
Morocco are closed. At the same time, ongoing work by the UK Met Office
and the UK CAA has confirmed the effectiveness of the model used to
determine the areas where ash concentration could be above engine tolerance
levels.
29,935 Normal Areas of high ash concentration at lower altitudes, which are still causing
some difficulties for trans-Atlantic flights, are currently found in the
Mediterranean between the Spanish mainland and the Balearic Islands, and
are moving north-east. All airports are available, however with the Balearic
Islands airports operating at reduced capacity. The areas of higher ash
concentration are expected to dissipate further during the day.
26,852 Normal The areas of high ash concentration at high altitude have now dispersed. The
areas of higher ash concentration are not expected to cause any disruption to
air traffic during the next 24 hours.
Normal Normal The areas of ash concentration are mainly at low levels in the vicinity of
Iceland, and are not expected to cause any disruption to air traffic during the
next 24 hours.
25,088
25,000 None.
97
CESifo Forum 2/2010
Focus
Mon 17 May
29,000
The areas of ash concentration are mainly at low levels. During the course of the
day, the current cloud is expected to disperse somewhat. The cloud is expected
to mainly affect northern Ireland, parts of Scotland and parts of south-western
UK. On Sunday 16 May, the disruptions in Ireland and north-western UK
resulted in a reduction in expected number of flights by about 400.
Note: Blanks in the actual flight number indicate that EUROCONTROL did not provide official information,
which happened in days when air traffic was normal.
Source: EUROCONTROL News – Update on European Air Traffic Situation (www.eurocontrol.it).
Table 5 clearly shows that a persistent negative impact on the
selection of listed airlines may be
Event window (EW)
Flights affected
observed throughout all event
EW1
15–19 April
Peak negative effect on most EU flights
EW2
15–23 April
Total period until return to normal flight
windows, with the exception of
operation (April)
the second event window that
EW3
3–7 May
Negative effects on UK, Ireland and
ended on 23 April. This might be
southern European flights
explained by the rebound
EW4
17–19 May
Some disruption of flights to/from
observed on 21 April and presumUK and northern Europe
EWTOT 15 April–
Total period until return to normal flight
ably by the positive expectations
20 May
operations (May)
driven by a return to normality
Source: ?
and the debate about rising ash
concentration thresholds to allow
Figure 1
air traffic, which happened
towards the end of the event winAGGREGATED PATELL STANDARDISED RESIDUALS
dow. The reoccurrence of the ash
BY GROUP OF FIRMS
4
cloud in May resulted in new
Potential gainers
3
major negative effects, probably
2
worsened by the expectation that
1
the issue could be long-lasting.
0
The overall effect over the longest
-1
event window is strongly nega-2
tive, despite the fact that increas-3
Airlines
ing the length of an event window
-4
usually reduces the power of tests
because of forecast errors. Again,
the evidence on potential gainers
Note: Continuous horizontal lines show 95% confidence level thresholds.
is mixed and confined to the
Source: Datastream; authors' calculation.
short-term at the first occurrence
of the crisis, while on the longest
More powerful tests can be obtained by aggregating
event window there is no detectable effect.
over time as well as over securities, according to
equation (9). We consider the various event windows
Finally, in Table 6, we provide an estimate of the
described in Table 5.
overall effect of the crisis on the individual compa-
Table 4
19/05/2010
17/05/2010
13/05/2010
11/05/2010
07/05/2010
05/05/2010
03/05/2010
29/04/2010
27/04/2010
23/04/2010
21/04/2010
19/04/2010
15/04/2010
Event windows
Table 5
Aggregate financial impact over various event windows (EW): Patell Standardised Residuals
EW1
EW2
EW3
EW4
15–19 April
15–23 April
3–7 May
17–19 May
Airlines
– 3.48*
– 1.29
– 4.04*
Potential gainers
2.66*
0.98
– 3.13*
Note: * Significant abnormal returns at the 95% confidence level.
Source: Datastream; authors’ calculation.
CESifo Forum 2/2010
98
–3.80*
– 1.65
EWTOT
15 April–
20 May
– 3.79*
– 0.18
Focus
Table 6
Impact by individual security (Patell Standardised Residuals and impact on firm market values)
PSR
Airlines
Aerlingus
Air France - KLM
British Airways
Easyjet
Finnair
Iberia
Lufthansa
Ryanair
SAS
Potential gainers
Avis
Budget
Eurazeo
Eurotunnel
Hertz
Sixt
Aggregate results
Airlines
Potential gainers
Note: * Significant abnormal returns at the 95% confidence level.
Firm value impact
(million )
– 0.12
– 1.34
– 1.43
– 2.20*
– 1.91
– 1.52
– 2.20*
– 1.58
– 1.25
–7
– 368
– 365
– 338
– 83
– 525
– 670
– 791
– 229
1.10
– 1.14
0.56
– 1.02
– 0.46
– 0.56
123
– 225
110
– 201
– 306
– 76
– 3.79*
– 0.18
– 3,374
– 575
Source: Datastream; authors’ calculation.
an efficient instrument for monitoring the patterns
nies, based on PSRs as computed in equation (7). We
also give a rough estimate of the associated financial
loss, based on the excess residuals on single days and
the market value of firms.4 Of course, these figures
only refer to a small selection of companies and
stock exchanges, but they may give an idea of the
magnitude of the effect. Considering the period
between 15 April and 20 May, the nine airlines considered for these studies experienced a loss of about
3.3 billion euros in terms of market value.
of these complex effects.
With respect to the ash cloud application, a few key
results may be summarized. First, as experienced in
other risk-related events, while the first occurrence
of the crisis generated major negative impacts on airlines, the return to normal financial operations was
quite rapid; one week after the first closure of
European airspace and airports there was no major
sign of significant losses. However, as the ash cloud
returned to affect flight operations in May, despite
Concluding remarks
the relatively low impact in terms of disrupted flights
and grounded passengers, the financial reaction was
This study provides a timely exploration of the
impact of the volcanic ash cloud, using financial
data and a traditional event study approach. The
volcanic ash cloud and its effect on air traffic represent a major example of the complexities that
economists face in producing a rapid estimate of
the monetary effects of a natural disaster.
Although the event study approach is a dated
instrument and was applied in its basic form, which
consists in running a set of linear regressions and
out-of-sample forecast tests, the procedure is still
quite strong. Reoccurring events raise the risk level
for affected companies and may engender a structural impact on the economic performance of firms,
at least in the short to medium term. Although our
limited sample of securities does not allow for general conclusions, potential gains for economic agents
who might benefit from the disruption of air travel
seem to be short-lived, consistently with the adaptive
behaviour of agents and the time needed for structural adjustments (e.g. increasing capacity for car
rentals). Our overall estimate for 9 selected
European flag carriers is a loss of about 3.3 billion
4 As a reference company market value we use Datastream estimates for May 2010.
euros over one month, a figure which is well above
99
CESifo Forum 2/2010
Focus
the overall economic impact estimated in the aftermath of the event for all European airlines.
References
Becker, G. S. and Y. Rubinstein (2004), Fear and the Response to
Terrorism: An Economic Analysis, Revised Version of February
2010, University of Chicago, http://www.econ.brown.edu/fac/
yona_rubinstein/Research/Working%20Papers/BR_FEB2010.pdf.
European Commission (2010), The Impact of the Volcanic Ash
Cloud Crisis on the Air Transport Industry, SEC(2010) 533, 27 April,
Brussels.
International Air Transport Association (IATA, 2010), IATA
Economics Briefing: The Impact of Eyjafjallajokull’s Volcanic Ash
Plume,
http://www.iata.org/whatwedo/Documents/economics/VolcanicAsh-Plume-May2010.pdf.
MacKinlay, A. C. (1997), “Event Studies in Economics and
Finance”, Journal of Economic Literature 35, 13–39.
Patell, J. M. (1976), “Corporate Forecasts of Earning Per Share and
Stock Price Behaviour: Empirical Tests”, Journal of Accounting
Research 14, 246–276.
CESifo Forum 2/2010
100
Special
almost half of German exports are to the euro area
and thus not immediately affected by the devaluation of the euro – just as they were also not affected
by the foregoing revaluation. As originally intended
by its founders, the European Monetary Union has
contributed to immunising the real economy against
exchange rate fluctuations.
A EURO RESCUE PLAN
WOLFGANG FRANZ*, CLEMENS FUEST**
MARTIN HELLWIG*** AND
HANS-WERNER SINN****
In light of the gravity of the European debt crisis and
the radical measures that were decided on in May
2010, we feel obliged to make this public appeal to
the German federal government, in which we set out
basic considerations for the pending negotiations for
the reform of the euro area. In our opinion, these
considerations are essential for the survival of the
European Monetary Union.
The euro is a key element of European integration, but
it will be endangered if we do not succeed in establishing more fiscal discipline in the future. What Europe
needs is not an economic government but political and
economic mechanisms that effectively limit public and
private indebtedness in the member states. It needs
these mechanisms not only to stabilise national
finances and the common currency but also to achieve
a better balance of growth forces in Europe.
The present crisis is not a currency crisis.
Interpretations that regard the devaluation of the
euro vis-à-vis the dollar as a currency crisis are pure
hysteria. The devaluation is only a partial correction
of the overvaluation of the euro relative to OECD
purchasing power parity that had been building up
since 2003. The improvement in competitiveness for
Europe’s exports on international markets that
comes with the devaluation should be welcomed by
the countries in Europe, especially since they have
not yet recovered from the recession. However,
Professor of Economics, University of Mannheim, President
of the Centre for European Economic Research (ZEW);
Chairman of the German Council of Economic Experts.
** Professor of Economics, Oxford University; Research
Director of the Oxford University Centre for Business
Taxation; Chairman of the Academic Advisory Board of the
German Federal Ministry of Finance.
*** Professor of Economics, University of Bonn, Director of the
Max-Planck Institute for Research on Collective Goods,
Bonn; Chairman of the Advisory Committee of Wirtschaftsfonds Deutschland.
**** Professor of Economics and Public Finance, University of
Munich (LMU); President of the Ifo Institute for Economic
Research and the CESifo Group.
The current crisis is due to the debt and financing
problems of some euro member states. The introduction of the euro induced a convergence of interest
rates in the eurozone countries. As of 1995 the initial
members of the European Monetary Union (except
for Germany) had to pay interest rates on their government bonds that were on average 2.6 percentage
points higher than the rate on German government
bonds, in some cases even 6 percentage points higher. With the introduction of the euro this interest
premium disappeared almost completely. A common
capital market was created that provided favourable
financing conditions of a kind that especially the
countries on the southwest periphery of Europe had
not known previously. These favourable conditions
applied to both public and private debtors. As a
result there was a huge capital flow to these countries that induced a boom in construction and investment. However, effective mechanisms for limiting
public and private debt were lacking. In many countries the booms in construction activity developed
into speculative bubbles. The bursting of these bubbles is now threatening the solvency of banks. This
has resulted in considerable risks for the public
finances of some countries. Current government
deficits in Ireland, Greece, Portugal and Spain
amount to more than three times the level allowed
by the Stability and Growth Pact (3 percent of
GDP). In Greece, the problem is exacerbated by an
already very high level of government debt.
*
In the international discussion on sharing the burdens from the crisis, the claim is often made, particularly by the French, that Germany was the main economic beneficiary of the single currency, the high
German trade surplus being a reflection of the trade
deficits of the other euro countries. The only truth in
this statement is that the booms in construction and
101
CESifo Forum 2/2010
Special
investment in the countries of the periphery caused
inflation there that reduced the price competitiveness of these countries and led to external trade
deficits. In Germany the reverse situation was true. A
large part of savings was channelled abroad through
the banks and relatively little money was invested at
home. Domestic demand weakened, and wages and
prices increased only slowly. This improved the price
competitiveness of the country, and led to a large
external trade surplus. This trade surplus was the
counterpart in the real economy to the capital flows
out of Germany.
In the current political discourse, those who regard
the trade surplus and the combined loss of investment capital as a sign of Germany’s strength display
an almost tragic misunderstanding of the underlying
economics. Germany’s net investment rate from
1995 to 2008 was the lowest of all OECD countries.
In 2008 only 40 percent of savings in Germany was
invested in the country itself. Economic growth was
correspondingly low. For 1995 to 2008 as a whole,
real growth in Germany was 22 percent, almost the
lowest in the entire OECD. In contrast, during this
period economic growth amounted to 33 percent in
Portugal, 56 percent in Spain, 61 percent in Greece
and an impressive 124 percent in Ireland.
In principle, the flows of capital out of Germany can
be seen as part of a useful convergence process in
Europe, which has also benefited German savers and
property owners. However, these flows were excessive. Too large a part of savings flowed out of Germany and into the countries of the southwest periphery of Europe. The resulting trade imbalances, which
gave rise to large imbalances in external trade, were
too high. The dynamics of the German economy, in
contrast, were impaired.
This critical assessment is based not only on the observation that the construction and investment boom
degenerated in part into a speculation bubble and that
individual member states are now caught up in a debt
crisis. It is mainly based on the assessment that even
before the crisis broke out there were insufficient constraints on public and private borrowing.
Markets and prices did not sufficiently take into
account the risks of investments. It was evident for
some time that the government bonds of different
member states, for example, did not all have the same
degree of creditworthiness. Prior to 2008, however,
investors did not demand interest premiums from
CESifo Forum 2/2010
102
member states whose budget policies created default
risks for their government bonds. No sanctions were
imposed on the violators of the European Stability
and Growth Pact, neither from the capital markets
nor from officials in the European Union.
It was not until the 2008 financial crisis that interest
rates on government bonds began to include premiums for default risks. Contrary to what some of the
debtors have suggested, this is not reprehensible but
is actually necessary for the long-term survival of the
euro. The risk premiums are indeed much smaller
than in the pre-euro era, but they help re-impose the
fiscal discipline that had been lacking and they are
fundamentally necessary for the functioning of the
eurozone capital market. Capital flows are slowed
down, government debtors are effectively disciplined and an overheating is avoided. It can now be
expected that more capital will remain in Germany
and will be available for investments in construction
and in the rest of the economy. Domestic demand in
Germany should become stronger and growth will
accelerate again. And, of course, the external trade
surplus should decrease again.
The rescue decisions made by the Council and the
member states on 9/10 May, however, suspended the
no-bailout stipulation of the Maastricht Treaty in order
to prevent a default by Greece and possibly other
member states on their public debts. Without passing
judgment on these decisions, we are nevertheless convinced that these measures should not be extended in
the present form beyond the agreed period of time.
Maintaining them would further destabilise the eurozone. The recent packages foresee the complete
bailout of creditors without requiring them to bear any
part of the risks they had taken on. Sparing the creditors would again lead to carelessness in their lending
policies and to an excessive degree of interest-rate
convergence. The disciplining function of the capital
markets would again be undermined, and the incentive to keep government budget deficits under control
would be weakened. German voters are not likely to
accept such a carte blanche guarantee for other countries. A regime in which Germany is obligated to pay
for other countries’ debts could endanger the euro if
not European integration as a whole. Even if this prevents further crises, the imbalances in foreign trade
and in capital flows would be perpetuated, with negative consequences for German economic growth.
Proponents of a continuation of the rescue policies
argue that the associated lowering of interest rates
Special
will allow those countries with high private and public indebtedness to reduce their deficits faster in the
current crisis. But the highly indebted countries in
the eurozone have already gained considerable relief
from the rescue packages. Going beyond this and
having additional interest-rate convergence induced
by continued joint liability would be damaging.
Many member states already received considerable
relief from lower interest rates when the euro was
introduced. But they did not redeem their debts as a
result. Evidently, the effects of lower interest burdens were weaker than the incentives for additional
debt financing that went along with lower interest
rates – and with reduced capital market discipline. It
was not until the current crisis made interest spreads
grow again and imposed constraints on new borrowing that policymakers are now seriously attempting
to limit state indebtedness.
For this reason it would be a grave mistake to introduce eurobonds, i.e. common government bonds of all
states of the eurozone, as some countries are currently
proposing. This would perpetuate the mutual bailout
principle and would ultimately destroy the institutional basis for a sound fiscal policy in Europe.
Before the rescue package expires, policymakers
must develop a viable concept for future fiscal-policy
rules in Europe. This concept must contain two elements: more effective political constraints on government deficits and debt and most importantly a quasiinsolvency procedure for member states. The quasiinsolvency procedure is intended not only to help
overcome future crises. It should also ensure that the
member states keep their debt under control and that
the financial-market participants proceed with
greater caution. It is essential that the rules of the
quasi-insolvency procedure be credible. The institutions responsible for enforcing these rules must be
willing to do so effectively. Unlike the Stability and
Growth Pact or now with the no-bailout stipulation,
the rules must not be circumvented the moment they
are intended to take hold.
In order to understand the conditions under which
such a quasi-insolvency procedure would function, it
is helpful to ask why even the German federal government agreed to the euro rescue plan in the current
crisis, despite its violation of the Maastricht Treaty.
Political pressure from the highly indebted countries
certainly played a role. But the decisive factor behind
the German government’s acceptance of the plan
seems to have been the fear that a sovereign default
103
in the eurozone would trigger a general capital market crisis with concomitant bank failures and yet
another financial and economic crisis. Regardless of
whether this fear was ultimately justified or not, the
risk itself was regarded as too large. As it turned out,
the eurozone was not prepared for this situation. The
lack of preparation contributed to the emergence of
the crisis because investors knew that the eurozone,
in the case of the over-indebtedness of an individual
member state, would hardly risk a formal default.
Because the no-bailout stipulation was not credible,
these states could take on much more debt, without
the interest rates reflecting the risks.
In order to enforce government budget discipline in
Europe, the capital markets must receive credible
signals that in the case of one country’s over-indebtedness, the creditors bear liability before help from
EU or other member states can come into play. This
sequencing is essential for inducing creditors to be
cautious when granting loans. But this sequencing is
only credible if there is no risk that the losses that
the creditors must bear will not set off a general
financial and economic crisis.
For this purpose, the capital requirements of the
banks should be increased. If the government bonds
of highly indebted countries must be backed up with
more capital, this in itself will ensure that lending to
these countries will be reduced or that higher interest premiums will be required. It is also necessary to
have a simple, transparent, and rule-determined procedure for the restructuring of public debts. Only
with such a procedure can one expect that a sovereign default will not induce a financial panic.
To enhance the credibility of rules for crises it is necessary to significantly strengthen the EU’s competence for enforcing budget discipline both before and
during a crisis itself. This competence should be
assumed directly by the European Commission. In the
present crisis the Commission did not perceive its
intrinsic interest as lying in the enforcement of the
Maastricht Treaty but in the expansion of the EU’s
competence that came with the bailout measures. If
the Commission is given the competence for the determination and enforcement of the conditions under
which aid is granted, it will perceive its own interest to
lie in exercising this competence when a crisis occurs.
How should the fiscal-policy rules for the eurozone
be reformed? In our opinion, the following ten
points are necessary:
CESifo Forum 2/2010
Special
1. Distressed countries can expect help only if an
imminent inability to fulfil its financial obligations is unanimously confirmed by the countries
providing the help together with the IMF.
2. Assistance can be provided by covered bonds
bearing interest, or by loans, the yield of which
must be set at a reasonable percentage (possibly
3.5 percentage points) above the European average. The loans must not exceed a given percentage maximum of the distressed country’s GDP.
3. At the same time assistance is granted, the original creditors must waive a portion of their
claims through a so-called ‘haircut’. The maximum percentage to be waived must be clearly
defined beforehand, in order to prevent a panicfuelled intensification of the crisis. A reasonable
haircut would be 5 percent per year since the
issuance of the respective government bond.
This would limit the interest premium demanded upfront by the creditors to a maximum of
around 5 percentage points. In addition, an
extension of maturity could be considered for
bonds with less than three years to maturity. The
crucial factor is to provide the capital markets
with a clear calculation framework.
4. The budget of the countries facing quasi-insolvency must be placed under the control of the
European Commission. Together with the
country in question, the Commission would
work out a programme to overhaul the state’s
finances, including reforms aimed at strengthening economic growth. Disbursement of rescue funds must be contingent on compliance
with the conditions set forth by the rescue programme.
5. This quasi-insolvency process must under no
circumstances be undermined by other assistance systems that could provide incentives for
opportunistic behaviour, in particular by such
mechanisms as the eurobonds, i.e. commonly
issued government bonds, favoured by some of
the overly indebted countries. Eurobonds entail
an across-the-board equalisation of interest
rates regardless of the creditworthiness of each
debtor country and, for that reason, would be
tantamount to a subsidy to capital flows to those
countries. Eurobonds would give carte blanche
to new debt excesses, in addition to exerting
negative effects upon economic growth in
Germany. A particular risk in the coming negotiations is that Germany will be pressured to
accept eurobonds in return for a quasi-insolvency procedure.
CESifo Forum 2/2010
104
It is also necessary, in our opinion, that the political limits to public debt be strengthened by
rules along the lines of the Stability and Growth
Pact. It must be stressed, however, that these limits cannot substitute for the discipline imposed
by the markets through the interest spreads
resulting from the individual countries’ creditworthiness. The following provisions should be
established:
6. The deficit limit set by the Stability and Growth
Pact should be modified in accordance with each
country’s debt-to-GDP ratio, in order to demand
more debt discipline early enough from the highly indebted countries. As an example, the limit
could be tightened by one percentage point for
every ten percentage points that the debt-toGDP ratio exceeds the 60-percent limit. A country with an 80-percent debt-to-GDP ratio, for
instance, would be allowed a maximum deficit of
1 percent of GDP, while a country with a 110-percent debt-to-GDP ratio would be required to
have a budget surplus of at least 2 percent.
7. Penalties for exceeding the debt limits must
apply automatically, without any further political
decisions, once Eurostat has formally ascertained the deficits. The penalties can be of a
pecuniary nature and take the form of covered
bonds collateralised with privatisable state
assets, and they can also contain non-pecuniary
elements such as the withdrawal of voting rights.
8. In order to ascertain deficit and debt-to-GDP
ratios, Eurostat must be given the right to directly request information from every level of the
national statistics offices and to conduct independent controls on site of the data gathering
procedures.
9. Finally, in case all the above assistance and control systems fail and insolvency approaches, the
country in question may be asked to leave the
eurozone by a majority of the eurozone members.
10. A voluntary exit from the eurozone must be possible at any time.
We are Europeans by conviction and favour the further integration of the EU. However, we firmly
believe that the European model will fail if it does
not manage to strengthen once again the individual
responsibility of the countries of Europe. For this
reason we see no alternative to the proposals we
have outlined here.
Special
COMMENTS ON RECENT
FISCAL DEVELOPMENTS AND
EXIT STRATEGIES
VITO TANZI*
General background
In a note published in The Financial Times of
12 August 2003, the author of this comment
advanced the hypothesis that the world might be
moving towards a future fiscal crisis. The reasons
for such a prediction were three-fold. Firstly, significant structural fiscal deficits and high public
debts characterized many countries (including six
of the G7 countries) at that time. Among the G7
countries, the only exception was Canada.
Secondly, widely anticipated demographic developments would become significantly unfriendly to
the countries’ public finances by around 2010.
These demographic changes would require important and painful reforms in public pensions, health
care, care for the aged and other welfare programs.
These reforms would be politically unpopular, as
President George W. Bush quickly discovered
when he attempted to reform the US social security system in 2005. Third was a development that
had attracted little attention but that, a few years
earlier, had inspired a cover story in The
Economist of 6 June 1997. The story had the catching title of ‘The Disappearing Taxpayer’. This
hypothesis was associated with the existence of ‘fiscal termites’. It argued that globalization of economic activities and financial markets, combined
with ongoing technological developments (internet-use, trading in non-tangible goods, etc.), was
making it progressively more difficult for countries
(and, ceteris paribus, especially for high-tax countries) to raise tax levels, or even maintain the high
* Former Director of Fiscal Affairs Department, IMF and Undersecretary for Economy and Finance, Italian Government. This
paper was delivered at the Vancouver Seminar on ‘The Crisis
Response and Road to Recovery’, hosted by the Department of
Finance and the Bank of Canada, Vancouver, 30 November to
1 December 2009. The views expressed are strictly personal.
105
levels reached in past years. Some observers had
dismissed the hypothesis. However, the latest data
available for OECD countries indicate that, in the
first decade of this century, the ratio of taxes to
GDP fell in most of them, and even in those countries that, because of high fiscal deficits, would
have been expected to increase taxes. A talk with
any country’s tax administrator would provide
strong support for this hypothesis. Ceteris paribus
taxes are always more difficult, politically, to
increase than to cut. On the other hand, public
spending is always easier to increase than to
reduce. These important asymmetries should not
be forgotten in the pursuit of fiscal policy.
The 2003 Financial Times note had been written at a
time when the economies of many countries were
doing relatively well, and governments were not facing urgent ‘exit strategies’.
The crisis and the response to it
In 2008 there was the unwelcome visit of a major
financial crisis, and, in late 2008 and especially in
2009, that of an economic crisis. The financial crisis
had not been as unanticipated as generally reported – see, for example, Tanzi (2007a). The huge economic imbalances that had accumulated among
countries and the ongoing bubbles in some sectors
within countries had been sending strong warning
signs that should have been listened to. Most countries felt the full impact of the crisis in 2009. At that
point, concerns that might have existed about
future fiscal developments – because of the anticipated demographic changes as well as theoretical
and practical doubts that had been raised over the
years about the effectiveness of discretionary fiscal
policy (such as the existence of various lags, possible ‘Ricardian equivalence’ reactions, impacts of
high debts and high fiscal deficits on the confidence
and the psychology of investors, questions about
future fiscal sustainability, etc.) – were pushed aside
or ignored. Closet Keynesians came out of the closets and some became very vocal in encouraging or,
better, in pushing governments to increase public
CESifo Forum 2/2010
Special
spending to fight the crisis, or to take advantage of
it to promote public sector activities that they had
wanted to see promoted. Some did this with the
same degree of spending enthusiasm shown by
sailors, when they go on shore after having spent
many months at sea.
The calls by some articulate and well-placed economists became loud and even shrill. These calls were
supported, and amplified, by similar calls coming
from international institutions including the IMF.
Those advocating counter-cyclical Keynesian fiscal
policies were not satisfied with letting the automatic
stabilizers do their work. Because of the severity of
the crisis, and especially because of its impact on the
incomes of sectors that (partly because of bubbles)
had contributed significantly to tax revenue (especially the financial industry), without any discretionary action by governments, the fiscal deficits
would still have increased rapidly and significantly,
and would have provided some large, automatic stabilization to the countries’ economies. The calls were
for additional, discretionary fiscal stimulus directed
especially towards higher public spending. There was
a repeat of the calls, in the 1990s, on the Japanese
government to increase Japan’s fiscal deficit to fight
that country’s undergoing financial crisis. Those calls
made a mess of the Japanese fiscal accounts but contributed little or nothing to economic growth (see
Tanzi 2008). It was forgotten that a fiscal expansion,
which starts when the fiscal accounts are already out
of balance and already face large future problems, is
less likely to be effective than when it starts from
balances and sustainable fiscal accounts.
It was argued that the larger the fiscal stimulus, the
better it would be in fighting the crisis. Large fiscal
stimuli were presumably needed to ‘prevent another Great Depression’, a possibility assumed to be
very likely without additional public spending.
There were frequent references, by the supporters
of large fiscal stimulus packages, to abysses being
faced and to the need to step back from them with
the help of higher public spending.
These calls ignored many significant differences
between the current situation and that of the 1930s.
During the 1930s, for example:
• Government expenditure and taxes were very low
(as shares of GNP). In the United States, for
instance, general government expenditure was
only 9.9 percent of GNP in 1929, while general
CESifo Forum 2/2010
•
•
•
•
government receipts amounted only to 10.9 percent of GNP. Federal expenditure and receipts
were respectively only 2.5 and 3.7 percent of GNP
in the same year (see Stein 1984). In other countries these ratios were not much higher.
Consequently, there were hardly any automatic
stabilizers during the Great Depression.
Fundamental errors (such as letting the money
supply fall sharply, engaging in protectionist ‘beggar thy neighbor’ policies, and others) were committed. These errors contributed to transforming
the 1930s crisis into a Great Depression.
Bank deposits were not insured, making runs on
banks a common experience.
Only few individuals received pensions or other
fixed incomes.
The situation during the current crisis was dramatically different. Public spending and tax levels were
much higher, in many countries well over 40 percent
of GDP. These levels provided major automatic stabilizers that would increase the fiscal deficits during
the crisis without discretionary fiscal stimulus packages. Central banks were ready to intervene, and
they did intervene, to inject huge amounts of liquidity in the system, and to rescue ‘too-big-to fail’ financial institutions in difficulties with their loans and
their purchases of toxic assets. To a large extent central banks became large ‘off-budget budgets’.
Protectionist tendencies were largely contained by
better collaboration among G7 and G20 countries.
Most bank deposits were insured, preventing runs on
banks. Many individuals depended on pensions and
on other fixed incomes (for example, salaries from
public employment) that were not, or were not much
affected, by the downturn in the economy.
Furthermore, in several countries, and especially in
the United States, there was no convincing evidence
of ‘under-consumption’ or of ‘excessive saving’ that
is a major justification for Keynesian policies. Even
during 2009 the United States continued to run very
large trade deficits and to have personal saving
ratios close to zero.
The current economic crisis was thus not a classical
Keynesian crisis of deficient aggregate demand (as
might have been the Great Depression). Therefore,
it could not be corrected by demand management
and by an injection of fiscal stimulus. Rather, it was a
crisis created by monumental structural imbalances,
especially in the United States, imbalances both in
current accounts and among sectors within the countries. The imbalances had been created, or had been
106
Special
stimulated, by inadequate policies, both monetary
and fiscal, in recent years. These imbalances required
special attention. They required policies specifically
directed to the causes that had led to the crisis.
The imbalances were particularly pronounced in
foreign trade and among important sectors, such as
housing, the automobile industry and, of course, the
financial sector. In the United States the financial
sector had seen its share of total profits rise from
about 5 percent in the 1940s to more than 40 percent in this decade. Thus, the cost of financial intermediation for the economy had increased phenomenally, inviting the important question as to the
value the financial market was contributing to general welfare to justify such a large share of total
profits. Some sectors, and especially the housing sector, had grown far too much because of the actions
of the financial market. It needed to be scaled down
to reduce the imbalances. The right policy should
have been to reduce the bubbles and to correct the
structural imbalances and not to follow policies that
would allow these sectors to maintain their inflated
incomes or claims on total resources. But this is
largely what current expansionary policies have
tried to do.
An important point to make is that, while automatic
stabilizers expand (automatically), during an economic slowdown, and can thus be expected to
reverse themselves (automatically) when, and if, the
economic situation returns to normal, fiscal stimuli,
created discretionally, require legislative changes,
both to create them and to undo them. Unless these
changes have clear and inflexible (that is politically
resistant) sunset provisions, and unless they have
been specifically directed at correcting the structural
imbalances (rather than at just injecting money in
the economy), they are likely to make the long-run
fiscal situation worse. Often they do not correct the
structural imbalances but allow them to continue in
the future, making ‘exit strategies’ in the fiscal area
more difficult.
This happened in Japan in the 1990s, where the fiscal
expansion created many useless government programs strongly defended by interest groups. It is now
happening in the United States, where the huge fiscal stimulus package of February 2009 will allow
some clearly unproductive activities or some overgrown sectors (car industry, housing and financial
market) to continue to operate at excessive levels.
Some of the policies being followed make no sense
in terms of correcting the existing structural imbalances.
As Pearlstein (2009) put it, politicians in Washington
are proposing to spend a lot of money that they do
not have, in ways that will not work, to help too
many people who are neither desperate nor deserving. He listed among ‘the idiotic ideas’ the bipartisan
push to re-inflate the housing bubble, and called ‘this
$10 billion boondoggle’ a giveaway to the real estate
industrial complex and one of those strategies that
are as nonsensical in theory as they are in practice.
The bill passed in the US Senate on a 98 to 0 vote!
This shows the power of lobbyists in determining the
details of fiscal policies and why an exit strategy for
the fiscal sector will be difficult to devise.
Confusing what was largely a structural crisis with a
traditional Keynesian under-consumption crisis has
led to policies that have created a fiscal mess and
that are likely to prolong the existence of structural
imbalances and to reduce potential economic growth
in future years.
A character in Charles Schultz’ popular ‘peanuts cartoons’ used to say that with enough ketchup, he
could eat anything. It seems that many policymakers
(and some vocal economists) believe that with
enough public spending, any country can be rescued
from any economic crisis, even when the crisis is
structural in origin and has been created by poor and
unsustainable policies. The existence of high unemployment is assumed to guarantee that the huge
injections of liquidity in the systems by central
banks, and in part by the large fiscal deficits (indirectly finance by the central banks) will not lead to
inflation. In this connection two comments may be
worthwhile:
Firstly, Morgan (1947, 84) states that “it is possible
that increases of private or public expenditure will
lead to sharp price rises in given areas or occupations, while there is still heavy unemployment elsewhere. An increase in the hiring of labor in
Massachusetts will not diminish unemployment
much in California and a road building program may
not much alleviate distress in the local textile industry”. We find here another important asymmetry.
Prices tend to increase more easily than they tend to
decrease. This point is particularly important when
the crisis is structural and has thus a geographical or
sector-specific impact. We tend to forget that the
mobility of factors like labor is limited even in the
107
CESifo Forum 2/2010
Special
United States where people are supposed to move
more readily than elsewhere. People will not move
from Massachusetts to Arizona to occupy houses left
empty by the housing bubble; or from the automobile industry in Michigan to Texas where jobs may be
more abundant in the oil industry.
Morgan’s sixty-year old view is clearly relevant
today, when the economic crisis has a major structural dimension. Because of this, it is puzzling to
listen to the statements made by Governor Ben
Bernanke and others that high unemployment
guarantees that inflation will not become a concern – in spite of the huge expansion which has
taken place in bank reserves and the huge fiscal
deficits which, to some extent, are being or will be
indirectly financed by the actions of the Federal
Reserve.
Secondly the view that inflation cannot coexist with
a high unemployment rate is not consistent with
much historical evidence from other countries. For
example, anyone who worked in Latin America in
the 1970s and 1980s would know better: in the 1980s
Argentina experienced one of the certified ‘great
depressions’ of the twentieth century. It happened
with huge inflation and enormous falls in output
(Tanzi 2007b). It would be wise to keep these experiences in mind.
Exit strategies
Let me now turn briefly to the exit strategies. We are
told (for example, by the Managing Director of the
IMF and by others) that they should be prepared
now but should not yet be acted upon. As StraussKahn said in a speech in London on 23 November
2009, “it is too early for a general exit [and] exiting
too early is costlier than existing too late”.1 One is
reminded of Saint Augustine who asked God to give
him chastity, but not yet. Thinking of exist strategies
from the current fiscal mess one is tempted to recall
a story about an Irishman who got lost in a backward, rural area of Ireland. When he asked someone
how he could get to Dublin, the answer he got was:
‘mister, if I were going to Dublin, I would surely not
want to start from here’. If we wanted to move
toward genuine, sustainable fiscal situations in the
future, we would not want to start from where we are
now. Unfortunately, there is no choice.
1
CESifo Forum 2/2010
See The Financial Times, 24 November 2009.
108
Three comments could be made in connection with
exit strategies from the fiscal mess. First, the longer
exit policies are postponed, the larger will the public
debts become that will, in turn, send negative signals
to economic operators in general. Thus waiting is not
a neutral policy. Second, if economic growth remains
week, there may never be an ideal, or even good,
time to exit. The time of exit will become the subject
of political discussion as it became in Japan. Third,
the more time passes, the stronger will become the
vested interests that protect the new spending programs introduced. Also, the idea of developing and
announcing the exit strategies now but waiting until
the right moment to enact them would be an invitation to all the lobbies of this world to organize themselves to prevent the needed changes.
Those economists who contributed to pushing the
countries into the current fiscal mess, believing
that they were rescuing them from a great depression, are not likely to have fully appreciated how
difficult the exist strategies will be in the fiscal
area. This area requires political decisions and
coordination among various institutions and political groups at each junction. In this respect exit
strategies in the monetary area are much less
demanding because they require far fewer political decisions. The exit strategies will be especially
difficult for countries that went into the crisis with
already high public debts, large fiscal deficits and
worrisome demographics. Obviously, the few
countries that started with better fiscal situations,
Canada among them, will have an easier, though
still tough time. It will not be a walk in the park for
any country.
The IMF (2009) has provided useful, but obviously
tentative, estimates for G20 countries of the fiscal
effort that they will need to exit from the fiscal
mess. Reminding readers that many advanced
economies entered the crisis with relatively weak
structural fiscal positions, this report highlights the
fact that government debt in advanced G20
economies would amount to 118 percent of GDP in
2014, even assuming some discretionary tightening
the following year. According to this report, Japan,
Britain, Ireland and Spain would require the largest
fiscal adjustment, and across the G20 the average
overall deficit would reach 7.9 percent of GDP in
2009. This is surely an extraordinary level. In the
advanced economies the structural primary balances would deteriorate by 4 percentage points of
GDP between 2007 and 2010.
Special
It should be added that the above estimated deficits
have not been inflated by the interaction of inflation
and high public debts, as so often happened in past
episodes. It is well known that this interaction can
distort and inflate conventional measures of fiscal
deficits. The current deficit estimates basically
assume zero inflation. Thus, in some sense, they have
greater ‘density’ (are greater per percentage unit of
GDP) than were the deficits in past years in countries that had significant public debts and some
inflation.
The exit strategy, defined as one that brings government debt-to-GDP ratios below 60 percent (the
original Maastricht level) by 2030 for advanced
countries “would require steadily raising the structural, primary balance from a deficit of 3.5 percent
of GDP in 2010 to a surplus of 4.5 percent of GDP
in 2020 – an 8 percentage point swing in one
decade – and keeping it at that level for the following decade” (IMF 2009, 23). The required
adjustment between 2010 and 2020 amounts to 13.4
percent of GDP in Japan, 12.8 percent in Britain
and 11.8 percent in Ireland, followed by 10.7 percent in Spain and 8.8 percent in the United States.
In Canada, the necessary adjustment is a more
modest 3.1 percent of GDP. Only Canada would be
within reach of the famous Maastricht criteria that
had been considered too strict and were consequently relaxed some years ago.
This adjustment would come at a time when demographic developments will become particularly unfriendly and when growth rates (and tax revenue)
are likely to be lower than in recent years. The reason for the expected lower growth rates is that the
crisis must have reduced potential growth rates of
countries by (a) lowering investment during the crisis years, (b) increasing unemployment, and (c)
lowering labor force participation caused by the
difficulty in finding jobs. It could be added that
recent years’ growth rates had been artificially
inflated by existing bubbles, and the demographic
trends will contribute to the reduction of the working-age population.
The estimated fiscal adjustment required reflects relatively optimistic developments on the financing
front. Many observers expect that interest rates will
increase due to the pressures coming from high fiscal
deficits to be financed and high public debts to be
serviced (in practically all countries), and the reversal of ‘quantitative easing’ policies carried out by
central banks. Should interest rates begin to rise, the
required fiscal adjustment could easily become much
larger than that estimated by the IMF.
Past adjustment experiences
The IMF (2009) also shows that in past years some
countries had been able to significantly improve
their cyclically-adjusted primary balances, expressed
in terms of a percentage share of GDP, over periods
that extended from 3 to 15 years. In other words, this
study stresses that large fiscal adjustments are possible. This is obviously an important message.
However, a few comments may be appropriate
regarding this matter.
First of all, the argument refers to experiences of single countries operating in isolation. Unfortunately,
this finding does not adequately reflect collective
experiences within similar global developments.
Furthermore, for the reasons mentioned earlier, the
fiscal deficits had probably been inflated somewhat
by the interaction of inflation (and increased nominal interest rates) together with significant public
debts.
Second, large reductions in public debts and in fiscal
deficits can be possible or easier to achieve by four
developments: (a) significant falls in real interest
rates, (b) rapid economic growth, (c) unanticipated
inflation and (d) major reforms that redefine the
proper role of the state in the economy and that
lead to large reductions in public spending (or to
some increases in the level of taxation).
The first two of these developments are unlikely to
characterize future developments. It is unlikely that
the present, extremely low real interest rates could
fall even more or that the growth rate could be significantly higher in the next several years. The third
(higher inflation) may play a role, just like in the
past, as a consequence of the large injection of liquidity made by central banks. The expectation and
the hope is that central banks will have the tools, the
independence, and the political courage not to allow
inflation to become a serious problem. However,
what economists call the ‘time consistency’ temptations could play a role, as they have occasionally
done in the past. Some governments are likely to
push central banks to inflate, especially if the maturity of the public debts is significant. However, in
several countries that maturity was reduced recently
109
CESifo Forum 2/2010
Special
to benefit from lower rates on shorter maturity
instruments.
The last alternative is clearly the most desirable,
although it is not an easy one. Because of the difficulties and the economic cost in raising the level of
taxation in today’s world, the better strategy would
be one that required a significant rethinking of the
spending role of the state – a rethinking aimed at
significantly reducing the now much inflated level
of public spending (see Tanzi 2009). A few countries
successfully followed this strategy in the past two
decades. These countries have shown that it is a
possible, though not an easy strategy, and that public welfare does not suffer as a consequence.
Tanzi, V. (2003), “To Scrap Fiscal Prudence Would Be a Mistake”,
The Financial Times, 12 August.
Tanzi, V. (2007a), “Complexity and Systemic Failure”, in: Estrin, S.,
G. W. Kolodko and M. Uvalic (eds.), Transition and Beyond,
London: Palgrave Macmillan, 229-246.
Tanzi, V. (2007b), Argentina: An Economic Chronicle, New York:
Jorge Pinto Book Inc.
Tanzi, V. (2008), Peoples, Places and Policies: China, Japan and
Southeast Asia, New York: Jorge Pinto Books, Inc.
Tanzi, V. (2009), The Economic Role of the State Before and After the
Current Crisis, Keynote speech made at the 65th International
Institute of Public Finance Conference, Cape Town, 13–16 August
2009, www.iipf.org/speeches/Tanzi_2009.pdf.
Conclusion
The exit strategy and the kind of adjustment that will
be needed by many of the G20 countries in future
years will require very difficult choices and great
political determination. It cannot be based on the
often-heard, recent slogan of ‘less market and more
state’. Future reforms and a sustainable exit strategy
should rather be guided by the slogan: ‘more efficiently regulated markets with less, but more efficient, public spending’. Reducing the level of public
spending and making it as efficient as possible would
have the double objective of allowing a reduction in
the fiscal deficit and promoting growth. This would
be an effective exit strategy. A better-regulated private market would facilitate the reduction in public
spending by making it possible to shift some current
government functions to the private sector. Also, to
the extent that there are possibilities for introducing
taxes on harmful activities, including environmental
charges, they could be relied upon. An immediate
introduction of these reforms is recommended.
References
International Monetary Fund (IMF, 2009), The State of Public
Finances Cross-Country, Fiscal Monitor, November 2009,
SPN/09/25.
Morgan, T. (1947), Income and Employment, New York: PrenticeHall, Inc.
The Economist (1997), “The Disappearing Taxpayer”, 6 June, 21–23.
Pearlstein, S. (2009), “In the Name of Jobs, Ideas That Won’t Work”,
The Washington Post, 6 November, A15.
Stein, H. (1984), Presidential Economics, New York: Simon and
Schuster.
Tanzi, V. (1995), Taxation in an Integrating World, Washington DC:
The Brookings Institution.
CESifo Forum 2/2010
110
Spotlight
ratio of 77 percent), Italy (116 percent) and Greece
(115 percent) struggle particularly with excessive
debt. Figure 1 also suggests, however, that countries
like Belgium (97 percent), Hungary (78 percent) and
France (78 percent) must also deal with high debt-toGDP ratios.
GOVERNMENT DEBT
IN EUROPE
CHRISTIAN BREUER AND
MATTHIAS MÜLLER*
The financial and economic crisis during the past two
years exerted a strong pressure on government budgets
in the EU member states. According to Eurostat
(2010),1 public debt (as a percentage of GDP) in the
euro area (EU16) rose to a peak of around 79 percent
in the 4th quarter of 2009 compared to a level of 66 percent in the beginning of 2008. In addition to the negative effects on government debt caused by the automatic stabilizers, the rapid increase in public indebtedness was mainly driven by government fiscal stimuli
and especially financial support to the banking system
aimed at avoiding a long-lasting economic downturn.
Since the beginning of the crisis some countries’
financial positions have worsened dramatically leading to rising concerns about these countries’ solvency.
As a consequence of its liabilities and due to the weak
ability to sell government bonds, Greece asked the
European Monetary Union (EMU) and the
International Monetary Fund (IMF) for financial aid
in April this year. The risk of a national insolvency in
an EU member state has never been as imminent
since the establishment of the EMU.This implies challenges of a new degree to policy makers in the EU.
To appreciate the size of government indebtedness
in an international comparison, Figure 1 depicts the
stock of government debt as a percentage share of
GDP in 27 European countries in the 4th quarter of
2009.2 In recent years, countries at the southern
periphery of Europe, such as Portugal (debt-to-GDP
* Ifo Institute for Economic Research.
1 The data on government debt in Europe for the 4th quarter of
2009 was published by Eurostat on 22 April 2010. It includes adjustments of previously published data for some countries: for example
the debt-to-GDP ratio for Greece in the 3rd quarter of 2009 was
revised from 113.2 percent towards 113.8 percent and the same
value for Germany from 71.9 percent towards 72.6 percent.
2 The figures show gross government debt, which indicates that the
data does not account for the country’s assets. This fact may raise
some doubts about the comparability of the data (German Council
of Economic Experts 2007).
Some northern and eastern European countries
show public indebtedness below or close to 40 percent of GDP. In Sweden, for instance, the share
reached 42 percent in 2009. Germany (73 percent),
Austria (67 percent) and the Netherlands (61 percent) capture the medium range on the scale
between 60 to 75 percent, even though these countries also exceed the Maastricht criterion of the
Stability and Growth Pact which allows a ratio of
government debt to GDP of 60 percent.
In order to assess the impact of the financial crisis on
national debt, we compare current government debt
with the corresponding figures before the crisis.
Figure 2 illustrates the change of each country’s indebtedness (expressed in a percentage of GDP) since
the beginning of 2008. The strongest increase of debt
was observed in Ireland, where public debt grew by
39 percent of GDP, followed by Latvia (around 27 percent) and Britain (23 percent). Greece and Spain have
widened their debt by approximately 19 and 17 percent
of GDP, respectively. With a stock of debt-to-GDP of
ca. 53 percent in the 4th quarter of 2009, Spain is one of
the few countries which still fulfill the Maastricht criterion with respect to the debt-limit (see Figure 1).
However, it remains to be seen whether Spain will be
able to stabilize its public finances since the country has
been heavily hit by its real estate crisis and by an
increase of unemployment.
In Germany (8 percent increase) and Austria (7 percent increase) the crisis has not yet resulted in new
government debt to the same extent as elsewhere in
Europe (Figure 2). However, recent economic forecasts for Germany have predicted a strong increase in
government deficits for 2010 (Carstensen et al. 2009).
The financial and economic crisis has resulted in accelerating government debt for the majority of European
111
CESifo Forum 2/2010
Spotlight
Figure 1
GOVERNMENT DEBT AS A PERCENTAGE OF GDP IN THE
4TH QUARTER OF 2009
countries. In the medium term, a
break of this trend and a step
towards fiscal consolidation seems
to be indispensable in order to
safeguard the stability of the
EMU. In this context, fiscal consolidation packages have already
been introduced (or planned) in a
number of countries including
Greece, Spain, Italy, Britain and
Germany. The current lively discussion in these countries not only
highlights the immediate political,
economic and social burden
caused by the implementation of
such restrictive measures but also
suggests a danger that an excessive simultaneous consolidation in
the individual countries may, in
turn, create large scale negative
economic spillovers in Europe.
References
Carstensen, K. et al. (2009),“ifo Konjunkturprognose 2010: Deutsche Wirtschaft
ohne Dynamik”, ifo Schnelldienst 62(24),
17–64.
Source: Eurostat (2010).
Figure 2
CHANGE OF GOVERNMENT DEBT AS A PERCENTAGE OF GDP BETWEEN
THE 4TH QUARTER OF 2007 AND THE 4TH QUARTER OF 2009
Eurostat (2010), Economic and Finance
Database, http://epp.eurostat.ec.europa.
eu/portal/page/portal/statistics/search_
database.
German Council of Economic Experts
(2007), Staatsverschuldung wirksam begrenzen, http://www.sachverstaendigenrat-wirtschaft.de/download/publikationen/fipo07.pdf.
Source: Eurostat (2010).
CESifo Forum 2/2010
112
Trends
FINANCIAL CONDITIONS
IN THE EURO AREA
8
Nominal Interest Rates a)
%
%
Stock Market Indices
8
7
7
6
6
5
5
380
340
300
long-term
Euro STOXX 50
260
4
4
short-term
3
3
220
2
180
1
1
140
0
0
2
yield spread
-1
100
-1
German share index (DAX)
June
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
a) Weighted averages (GDP weights).
Source: European Central Bank.
Source: Deutsche Bundesbank, Dow Jones & Co., Inc.
In the three-month period from March to May 2010 short-term interest
rates increased slightly. The three-month EURIBOR rate grew from an
average 0.65% in March to 0.69% in May. Yet the ten-year bond yields
declined from 3.99% in March to 3.68% in May. In the same period of
time the yield spread decreased from 3.34% (March) to 2.99% (May).
%
Dow Jones industrial
60
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
14
January 1996=100
Change in M3 a)
%
The German stock index DAX declined in June 2010, averaging
5,966 points compared to 6,136 points in April. The Euro STOXX
also decreased from 2,937 in April to 2,642 in June. The Dow Jones
International declined as well, averaging 10,159 points in June compared to 11,052 points in April.
Monetary Conditions Index
14
12
12
10
10
-5.0
1994=0 (inverted scale)
-4.0
monetary
easing
-3.0
8
8
6
6
ECB reference value
4.5 %
4
2
4
2
0
-2.0
-1.0
average
1994–2009
0.0
monetary
tightening
0
1.0
-2
-2
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
a) Annual percentage change (3-month moving average).
Source: European Central Bank.
The annual growth rate of M3 stood at – 0.2% in May 2010, unchanged
from the previous month. The three-month average of the annual growth
rate of M3 over the period from March to May 2010 stood also at – 0.2%,
unchanged from the previous period.
113
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Note: MCI index is calculated as a (smoothed) weighted average of real short-term interest rates
(nominal rate minus core inflation rate HCPI) and the real effective exchange rate of the euro.
Source: European Central Bank, calculations by the Ifo Institute.
Between April and November 2009 the monetary conditions index
remained rather stable after its rapid growth that had started in mid2008. Yet the index started to grow again since December 2009, signalling greater monetary easing. In particular, this is the result of
decreasing real short-term interest rates.
CESifo Forum 2/2010
Trends
EU
SURVEY RESULTS
Gross Domestic Product in Constant 2000 Prices
5.0
EU27 Economic Sentiment Indicator
Percentage change over previous year
%
Index 2000=100, seasonally adjusted
120
EU27
EA16
4.0
3.0
110
2.0
100
1.0
0.0
90
-1.0
-2.0
80
-3.0
-4.0
70
-5.0
June
-6.0
60
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Source: Eurostat.
Source: European Commission.
According to the first Eurostat estimates, GDP increased by 0.2% in both
the euro area (EU16) and the EU27 during the first quarter of 2010, compared to the previous quarter. In the fourth quarter of 2009 the growth
rate had amounted to 0.1% for the euro area and 0.2% for the EU27.
Compared to the first quarter of 2009, i.e. year over year, seasonally
adjusted GDP increased by 0.6% in the euro area and by 0.5% in the
EU27.
In June 2010, the Economic Sentiment Indicator (ESI) remained broadly unchanged at 100.1 (down by 0.1 points) in the EU27 and at 98.7 (up
by 0.3 points) in the euro area (EU16). The ESI reached its long-term
average, albeit it will still require further improvement for the economic
activity to reach its pre-crisis level
EU27 Industrial and Consumer Confidence Indicators
8
Percentage balances, seasonally adjusted
%
20
balances
EU27 Capacity Utilisation and Order Books
in the Manufacturing Industry
Capacity utilisation
(left-hand scale)
(right-hand scale)
0
%
88
Assessment of order books
84
0
-8
80
-20
-16
76
-40
-24
72
Consumer confidence **)
Industrial confidence *)
-32
Jun-10
Apr-10
Feb-10
Dec-09
Oct-09
Aug-09
Jun-09
Apr-09
Feb-09
Dec-08
Oct-08
Aug-08
Jun-08
Apr-08
Feb-08
64
Source: European Commission.
Managers’ assessment of order books improved from – 32.4 in April to
– 25.7 in June 2010. In March the indicator had reached – 38.2. Capacity
utilisation increased to 75.6 in the second quarter of 2010 from 73.1 in the
previous quarter.
In June 2010, the industrial confidence indicator decreased by 1 point in
the EU27 but remained unchanged in the euro area (EU16). On the other
hand, the consumer confidence indicator remained unchanged in the
EU27 but increased by 1 point in the euro area. However, these indicators
stood below the long-term average in both areas in June 2010.
CESifo Forum 2/2010
Dec-07
Oct-07
* The industrial confidence indicator is an average of responses (balances) to the
questions on production expectations, order-books and stocks (the latter with inverted sign).
** New consumer confidence indicators, calculated as an arithmetic average of the
following questions: financial and general economic situation (over the next
12 months), unemployment expectations (over the next 12 months) and savings
(over the next 12 months). Seasonally adjusted data.
Aug-07
Source: European Commission.
-80
68
Jun-07
Jun-10
Apr-10
Feb-10
Dec-09
Oct-09
Aug-09
Jun-09
Apr-09
Feb-09
Dec-08
Oct-08
Aug-08
Jun-08
Apr-08
Feb-08
Dec-07
Oct-07
Aug-07
Jun-07
-40
-60
114
Trends
EURO AREA
INDICATORS
Exchange Rate of the Euro and PPPs
Ifo Economic Climate for the Euro Area
150
USD per EUR
2005=100
1.65
140
1.50
130
Exchange rate
1.35
120
110
German basket
1.20
100
1.05
90
OECD basket
80
0.90
long-term average (1990-2009)
70
US basket
0.75
60
Juni
50
0.60
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10
94
95
96
97
98
The Ifo indicator of the economic climate in the euro area (EU16) rose
somewhat in the second quarter of 2010 and continues to stay below its
long-term average. The assessments of the current economic situation
improved only slightly vis-à-vis the first quarter. The expectations for the
coming six months weakened somewhat but remain positive on the
whole. These results indicate that the economic recovery will continue in
the second half of the year, albeit at a slower pace.
01
02
03
04
05
06
07
08
09
10
The exchange rate of the euro against the US dollar averaged 1.22 $/€
in June 2010, a decrease from 1.26 $/€ in May. (In April the rate had
amounted to 1.34 $/€.)
Unemployment Rate
10.4
00
Source: European Central Bank, Federal Statistical Office, OECD and calculations by the Ifo Institute.
Source: Ifo World Economic Survey (WES) II/2010.
Inflation Rate (HICP)
ILO definition, seasonally adjusted
%
99
4.5
EU27
EA16
9.6
Percentage change over previous year
%
4.0
3.5
8.8
3.0
Core Inflation
2.5
(total excl. energy and
unprocessed food)
2.0
1.5
8.0
Total
1.0
0.5
0.0
7.2
-0.5
Jun-10
Apr-10
Feb-10
Dec-09
Oct-09
Aug-09
Jun-09
Apr-09
Feb-09
Dec-08
Oct-08
Aug-08
Jun-08
Apr-08
Feb-08
Dec-07
Oct-07
Aug-07
Apr-10
Feb-10
Dec-09
Oct-09
Aug-09
Jun-09
Apr-09
Feb-09
Dec-08
Oct-08
Aug-08
Jun-08
Apr-08
Feb-08
Dec-07
Oct-07
Aug-07
Jun-07
Jun-07
-1.0
Apr-07
6.4
Source: Eurostat.
Source: Eurostat.
Euro area (EU16) unemployment (seasonally adjusted) amounted to
10.1% in April 2010, compared to 10.0% in March. It was 9.2% in April
2009. EU27 unemployment stood at 9.7% in April 2010, unchanged compared to March. The rate was 8.7% in April 2009. In April 2010 the lowest rate was registered in the Netherlands (4.1%) and Austria (4.9%),
while the unemployment rate was highest in Latvia (22.5%) and Spain
(19.7%).
115
Euro area annual inflation (HICP) was 1.6% in May 2010, compared to
1.5% in April. A year earlier the rate had amounted to 0.0%. The EU27
annual inflation rate reached 2.0% in May 2010, unchanged compared
to April. A year earlier the rate had been 0.8%. An EU-wide HICP
comparison shows that in May 2010 the lowest annual rates were
observed in Latvia (– 2.4%), Ireland (– 1.9%) and the Netherlands
(0.4%), and the highest rates in Greece (5.3%), Hungary (4.9%) and
Romania (4.4%). Year-on-year EU16 core inflation (excluding energy
and unprocessed foods) fell to 0.85% in May 2010 from 0.92% in
March.
CESifo Forum 2/2010
Online information services of the CESifo Group, Munich
The Ifo Newsletter is a free service of the Ifo Institute and is sent by e-mail every month. It informs you (in German) about new research results, important publications, selected events, personal news, upcoming dates and many more items from the Ifo Institute.
If you wish to subscribe to the Ifo Newsletter, please e-mail us at: [email protected].
CESifo publishes about 20 working papers monthly with research results of its worldwide academic network. The CESifo Newsletter presents selected working papers (in English) in an easily
understandable style with the goal of making its research output accessible to a broader public.
If you wish to subscribe to the CESifo Newsletter, please e-mail us at: [email protected].
If you wish to receive our current press releases, please e-mail us at: [email protected].
You can also request these services by fax:
Ifo Institute for Economic Research, fax: (089) 9224-1267
Please include me in your mailing list for:
Ifo Newsletter
CESifo Newsletter
Ifo Press Releases
Name:
………..........................................................................................................
Institution: .....................................................................................................................
Street:
.....................................................................................................................
City:
.....................................................................................................................
Telephone: ...................................................................................................................
Fax:
.....................................................................................................................
E-mail: …………………..............................................................................................