Professional Documents
Culture Documents
Table of content
Executive summary
18
27
37
Conclusion
39
Appendix
Author:
Tim Sprissler
Concept and realisation:
Stephan Schreckenberg, Oliver Schelske
Contributors:
Jens Mehlhorn
Marcel Andriesse
David Bresch
Andreas Spiegel
Reto Schnarwiler
Nikhil da Victoria Lobo
Editing:
Josephine Chennell, Richard Heard
Managing editor:
Esther Baur
The editorial deadline for this study was 18 July 2011.
2011
Swiss Reinsurance Company Ltd
All rights reserved.
The material and conclusions contained in this publication are for
information purposes only and the authors offer no guarantee for the
completeness of its contents. The statements in this report may
provide current expectations of future events based on certain
assumptions. These statements involve known and unknown risks,
uncertainties and other factors which are not exhaustive. The authors
of this report undertake no obligation to publicly revise or update any
statements, whether as a result of new information, future events or
otherwise and in no event shall Swiss Re Group or any of its entities
be liable for any damage and financial or consequential loss arising in
connection with the use of information relating to this publication.
Executive summary
On average over the past decade, 120 people have lost their lives as a result of
floods in Brazil per year. And economic losses have amounted to USD 250 million.
In 2010, these peaked at USD 950 million. As a result of growing urban areas and
the effects of climate change, the population at-risk is likely to rise from an
estimated 33 million to 43 million people between 2010 and 2030. Along with the
planned substantial investments and the corresponding growth of exposed values,
we estimate the annual expected losses to increase from USD 1.4 billion to USD 4
billion over this period. The southeastern, southern and northeastern states of Brazil
will be most affected by flooding.
In terms of addressing the flood risk in Brazil today, emergency relief takes
precedence over prevention measures. However, adaptation measures could
potentially reduce the annual expected losses by over one third in 2030. Given the
countrys low insurance penetration, the bulk of the losses is financed from public
and private budgets. Compared to emergency relief financing, prevent financing
through a variety of insurance schemes is often a more effective way of handling
flood risks, because they also set incentives for risk reducing behaviours. Such
measures include strengthening the insurance market for homeowners and
businesses, flood microinsurance schemes and innovative insurance solutions for
the public sector.
River and flash floods combined with landslides are the most frequent and costly natural
disasters in Brazil. Between 2000 and 2010, they killed on average 120 people and
caused economic losses of around USD 250 million per year. There has been a clear
trend towards above average economic damages over the past three years. In 2010,
losses peaked to 450 killed and missing and roughly USD 950 million (BRL 1.7 billion).
Besides the direct damage to life, health and assets, floods can put an increased strain
on personal, corporate and public budgets.
According to our estimates, 33.3 million people (17% of the population) are exposed to
flood risk in Brazil today. A major driver of flood exposure is, and will continue to be, the
rapid urban population growth as paved areas inhibit rainwater infiltration into the soil.
Uncontrolled urban expansion into risk areas is another factor. This means by 2030, a
further 6.5 million people will be exposed to flood risk. Due to the intensification of flood
hazards resulting from climate change, another 2.7 million people will also be affected.
Therefore, we estimate that in 2030 42.5 million people in Brazil will be exposed to flood
risk.
Today, floods can be expected to cause annual economic losses of around USD 1.4 billion
(BRL 2.5 billion). A 100 year flood event could trigger damages of up to USD 14 billion
(BRL 25 billion). Strong capital investments in the context of the Programa de Acelerao
do Crescimento 2 (PAC 2), the FIFA World Cup 2014 and the 2016 Olympic Games
mean that substantially higher economic values will be exposed. In 2030, annual
expected losses could be almost USD 2.5 billion (BRL 4.4 billion) higher than today, due to
asset growth alone. We estimate that an additional USD 200 million (BRL 352 million)
could be caused by flood hazard intensification related to climate change. All effects
combined, we anticipate annual expected losses to reach a little over USD 4 billion.
In terms of both exposed people and assets, the Southeast, Northeast and South of
Brazil are likely to be the most severely affected regions. Over 40% of the population at
risk and half of todays annual expected loss are estimated to be concentrated in the
Southeast, with its large metropolises and economic strongholds. About one fourth of
the exposed population and 13% of todays annual expected loss are likely to be
concentrated in the predominantly poor and agricultural based Northeastern states,
where floods can interfere with the ambitious development agendas in place. An
estimated 15% of todays annual expected loss could potentially occur in the South.
When addressing flood risks in Brazil, emergency relief dominates prevention measures.
From 2004 to 2010, the emergency expenses for all natural and man-made disasters of
around USD 2.6 billion (BRL 4.8 billion) were significantly higher than the money invested
Executive summary
in preventive measures by the various federal budget lines. 1 Given the increase in the
exposed population and assets due to urbanisation and more extreme weather conditions,
the strain on budgets is likely to rise in the near future.
Preventive adaptation measures could reduce the losses and the burden on budgets.
Building codes enforcing the sealing of openings on the ground floor are estimated to
avert losses in 2030 of up to USD 772 million (BRL 1.4 billion). Drainage projects could
reduce damages by almost USD 587 million (BRL 1 billion) and hillside stabilisation by
over USD 94 million (BRL 165 million). Urban planning directed at building settlements
outside floodplains and hill slopes could significantly reduce the flood exposure of
people and assets. The planned early warning system and awareness campaigns could
also save many lives.
Not all losses are avoidable, however. Therefore, planning for the financial consequences
of floods should be part of any loss mitigation strategy. The re/insurance sector offers
various solutions to transfer risks and losses away from homeowners, businesses and the
public sector. Especially with regard to extreme risks such as a 100 year flood event it
might make better economic sense to transfer risks than to take prevention or adaptation
measures designed to withstand such rare events.2
Currently insurance penetration in the non-life sector (3% in 2008), and particularly for
natural catastrophes, is low in Brazil. Thus, the bulk of the losses and reconstruction costs
fall on private and public budgets. A functioning insurance market embedded in a
regulatory framework which allows for product variety and access to international
reinsurance markets could absorb flood risks from homeowners and businesses more
effectively. Natural catastrophe insurance sets incentives for the insured to reduce their
exposure, if they retain part of the disasters cost themselves, or pay risk-adjusted
premiums.
The re/insurance sector can play an important part in increasing Brazils flood insurance
penetration by providing capacity and knowledge. If Brazils recent market liberalization is
sustained, primary insurers could become more willing to provide flood cover, especially in
the corporate/large risk segment, passing on extreme risks to the reinsurance sector.
A particularly exposed part of society is the low-income population that often lives in risk
zones such as hillsides or floodplains. To better protect their homes and possessions, flood
cover could be included in a homeowner microinsurance scheme for fire, explosion and
lightning. Homeowner microinsurance could increase flood insurance penetration
among the low-income population particularly affected. A scheme covering fire,
explosion and lightning offers an estimated market potential of up to 42 million homes
and USD 780 million.3
To include flood cover at still affordable premiums, re/insurers must be able to reduce
administrative costs via group pricing, simple claim turnaround processes and mass
marketing. Public-sector solutions like catastrophe reserve reinsurance, catastrophe
bonds or agricultural/weather insurance can help to reduce the financial strain on public
budgets caused by rising flood losses. Parametric triggers would significantly lower the
costs of such solutions and secure swift payouts.
Brazil would benefit from increasing its focus on flood prevention, adaptation and risk
transfer to ensure that the impacts of flooding do not present obstacles to the countrys
rapid economic development and its poverty reduction efforts. As the Brazilian economy
further expands, investment surges and urbanisation continues to mount, the need for
effective risk solutions will increase. The re/insurance sector stands as a strong partner for
the public sector, corporations and individuals to offer risk assessment and financing
solutions.
According to Contas Abertas 2011 in an article entitled Governo deixou de investir BRL 1,8 bi na preveno
de enchentes em sete anos.
Swiss Re 2010b.
3
Afonso, L. & Sepulveda, M. 2009. p. 199211.
2
River and flash floods combined with landslides have been a risk in Brazil throughout its
history. Although natural catastrophes were not regarded as a major threat in the past,
the countrys rapid socio-economic development and concentration of people and
economic activity in urban centres has substantially increased both the exposure to and
the socio-economic impact of flooding in the past decade. Between 2000 and 2010
floods killed close to 1 300 people and caused economic losses of nearly
USD 2.8 billion.
In the present study we aim to analyse the social and economic impact of floods in Brazil
today and 20 years from now, before going on to examine Brazils current flood
protection system and possible future measures. To contextualise the discussion, the
introduction gives an overview of the damages that flooding can cause in any country. In
chapter 2, we analyse the effects of past floods in Brazil and examine the social and
economic impact of floods today. Taking into account both the socio-economic
development and the flood hazard intensity development in the context of climate
change, in chapter 3 we estimate the flood impact in 2030. To see how Brazil deals
with flood risk, in chapter 4 todays flood protection system, including both the state and
the insurance sector, is analysed before discussing future adaptation and risk transfer
solutions. Finally, we sum up our findings with a discussion.
On the basis of the available information and data, this study aims to give a realistic, yet
typified, assessment of the situation. The loss estimates and development through to
2030 and the quantitative analysis of the adaptation measures are based on
assumptions and expert judgments, and are intended to provide a granular assessment
of flood risk in Brazil. Further in-depth analysis can be conducted at a later stage if more
data and information becomes available.
Tangible
Intangible
Direct
Indirect
Assets (buildings,
infrastructure,
machinery, farmland)
Contents (stocks, crops)
Loss of life
Health effects
Loss of ecological goods
The direct and indirect tangible damages shown mainly refer to the economic sectors
(industry, agriculture, tourism) and to infrastructure, facilities and systems (transport,
electricity and energy, communication, water and sanitation). However, the impact is felt
much deeper: Floods can also cause damage to or adversely affect social sector
housing, education and health. Since many development and poverty reduction
strategies (like those in Brazil) focus on these sectors, the impact of floods goes way
beyond the actual direct damage to infrastructure and facilities.
Housing property often forms a large share of a persons individual wealth, especially in
developing and emerging countries. Moreover, many small and medium-sized
businesses are run from the proprietor's own home. Therefore, floods which damage or
destroy houses and equipment can trigger significant secondary losses in income and
wealth. In the education sector floods can lead to the suspension of educational services
(e.g. teaching), either because direct damage has made facilities unusable or because
they are used as temporary shelters for homeless flood victims. Thus flooding may also
affect human capital formation, which often is a major development goal. In the health
sector, besides direct damage to health facilities (hospitals, laboratories etc.), medical
equipment and stocks of medications /vaccines, floods can lead to a reduction in
available health services and increased health expenses for treatment of victims during
the emergency, as well as in the longer run (trauma etc.). 4
While providing a good overview of the damage floods can cause to different sectors,
this comprehensive approach is difficult to apply, especially regarding the exact
identification of the indirect effects. Most flood damage assessments, therefore,
concentrate on the direct tangible loss assessment. As this study aims to examine the
socio-economic impact of floods in Brazil, we will analyse both the direct tangible
damage (economic impact) and the direct intangible damage (social impact). Where
possible, we will also refer to indirect losses such as business interruptions.
Box 1: Indirect flood damage
Floods can cause indirect damage in both the flooded and non-flooded areas which are
interlinked via supply chains or infrastructure. The damage is regarded as substantial if:
ECLAC, 2003.
1.1)
Over the past two decades, Brazil has experienced a significant increase of flood events
from 20 in the 1990s to 34 in the 2000s. Flash and river floods constitute by far the
most frequent natural hazard in the country (Figure 1.1).
34
35
Figure 1.1
Natural catastrophes in Brazil over the
past two decades
30
25
20
20
15
10
0
Drought
Earthquake
Flood
Extreme
temperature
Events 1990-1999
Storm
Wildfire
Events 2000-2009
Flash floods are caused by high-intensity local precipitation or torrential rain for several
hours. If the water cannot be absorbed by the ground any longer it runs off along the
surface causing a quickly rising flood wave. 5 In mountainous areas landslides often
occur in connection with flash floods if areas of inclination are no longer capable of
sustaining the soil due to heavy infiltration of water. 6 Moreover, Brazil is exposed to river
flood risk when prolonged precipitation over a wide area or short, intense rainfalls cause
the overflow of rivers along their courses.
In the first decade of the new millennium floods killed close to 1 300 people in total, left
almost 800 000 people homeless and caused economic losses of almost USD 2.8
billion (BRL 4.9 billion).7 This corresponds to annual average economic losses of around
USD 250 million (BRL 440 million) and 120 people killed and missing per year. During
the decade, economic losses more than quintupled from roughly USD 420 million in
2001-2005 to USD 2.3 billion in 2006-2010. There has been a clear trend towards
above average economic damages over the past three years. 2010 was recorded as the
costliest year of flooding in the last decade, with almost 450 people killed and economic
losses of roughly USD 950 million (BRL 1.7 billion; Figure 1.2).
947
1 000
Figure 1.2
Increasing economic losses and
fatalities from major floods in Brazil
over the last decade
800
673
600
517
400
200
448
350
189
95
90
97
55
2000
2001
74
13
2002
2003
24
0 0
2005
2006
0
2004
154
131
74
75
48
2007
2008
2009
2010
Due to their mountainous topology and the large number of densely populated urban
areas along river courses, historically, the Southeastern states have been particularly
hard hit by flooding. Regions in the mountainous Serra do Mar which runs from the state
of Esprito Santo via Rio de Janeiro and So Paulo to the north of Santa Catarina have
5
repeatedly been affected. Most recently, in January 2011, major floods in the cities of
Terespolis, Nova Friburgo, Petrpolis and Sumidouro (Rio de Janeiro state) killed over
800 people and caused economic losses of USD 1.2 billion (BRL 2.1 billion).8
Moreover, cities along the Paraba do Sul river, such as Volta Redonda, Barra Mansa or
Resende, which are strong economic centres have also been flooded in the past. The
major metropolitan areas of Rio de Janeiro and So Paulo have been affected, too, as the
former lies in a hilly region and the latter is crossed by the river Tiete (Table 1.1).
Table 1.1
Illustrative cases of flooding in
different regions of Brazil
Southeast
Northeast
Killed
Homeless
>800
>100 000
Jan
2011
Apr
2010
200
256
74 500
Jan
2010
Jan
2007
145
85
6 000
131
48
11 000
55
64
2 500
N/A
34
80 000
602
107
155 000
508
57
267 000
350
161
175 000
406
118
23 000
Dec
2001
Jan
2000
South
Economic
loss in USD
million 2010
1 200
Jun
2010
Apr
2009
Mar
2004
Nov
2008
Also states in the predominantly poor, agriculturally based Northeastern region have
been repeatedly struck by flash and river floods. In June 2010 heavy rains and floods
killed at least 107 people and caused severe damage of around USD 600 million
(BRL 1.1 billion). According to the Civil Defence office, 75 000 people in Alagoas and
80 000 people in Pernambuco were made homeless and around 19 000 houses were
damaged or completely destroyed. 9 In April 2009 river flooding and landslides blocked
highways and cut off the rail link from the major Amazon iron ore mines in Carajs (Par
state) to the Itaqui port in So Louis (Maranho state). Over vast areas communities
were isolated and public life, including schooling, was disrupted for several weeks.10
Major floods and mudslides in the Southern state of Santa Catarina in 2008 killed at
least 118 people. The Itaja river and others overflowed, affecting several cities (e.g.
Blumenau, Itaja) and damaging the major beef-exporting port of Itaja, which led to
economic losses of USD 400 million (BRL 704 million). In addition, business
interruption costs for the port were estimated at USD 35 million (BRL 62 million) per
day. Almost all highways in the region were blocked and the electricity and water supply
were cut off. Flooding also damaged the gas pipeline which transports Bolivian gas to
Southern Brazil, interrupting not only the gas supply in the flooded area but also in the
state of Rio Grande do Sul further south. Besides this, the tragedy also highlighted the
potential for social unrest as desperate flood victims looted supermarkets and
pharmacies.11
1.2)
We determine flood risk zones which will underlie further analysis and quantification of
the flood impact today and in 2030. Given Brazils exposure to flash and river flood risk
in the past, zones for both perils have been defined using digital elevation models (DEM)
of a 1 km resolution (Hydro 1 k DEM) provided by the U.S. Geological Survey (USGS).
Flash flood zones are determined by a 3% slope index. Even though this threshold
seems rather low, USGS studies have shown that it represents flash flood/landslide
zones best when using a Hydro 1 k dataset. Higher slope inclinations would be difficult
to detect at a 1 km resolution.
To determine river flood zones, Swiss Res patented approach of the Geomorphologic
Regression has been implemented on the processed DEM to determine the 100-year
flood boundary across Brazils entire river system. A city at the boundary has an annual
probability of being flooded of 1%, and cities within have a higher probability. The
horizontal and vertical distance to the next river, as well as the characteristics of the
basin area, are used as predictor variables. This approach allows us to derive river flood
risk zones based on the intrinsic characteristics of the river system, substituting for the
rarely available historic data on flooded areas and flood depth. The predictive quality has
been confirmed in studies, showing that in medium-sized catchments the
Geomorphologic Regression method obtains similar results on flood extents as other
methods.12
1.3)
1.3.1)
A major driver of flood exposure is urbanisation, which means both the movement of
people from rural to urban areas and the integration of areas into urban infrastructure. It
leads to more impervious areas, such as roads, pavements or rooftops. They inhibit the
infiltration of rainwater into the soil, which leads to the built-up, channelling and
acceleration of flood waves and decreases the time between the rain event and the
actual flooding. Tucci (2006, p. 8) shows, using data from So Paulo, Curitiba and Porto
Alegre, that the share of impervious areas in these cities is positively related to urban
density. At urban density levels above 100 inhabitants per hectare, the share rises
above 50% until it reaches a plateau at around 65%. Moreover, uncontrolled urban
expansion into risk areas such as flood plains and hillside locations increases the threat
of flooding. As such expansion obstructs natural floodways, the people and assets in
these areas will be affected more frequently by flood catastrophes. According to
estimates, one million of Rio de Janeiro's total 6 million inhabitants live in unplanned,
unlicensed hillside zones.13
Brazil has experienced rapid urbanisation in recent decades, with its urban population
share nearly doubling from 45% in 1960 to 86% in 2009. In 2009, 78 million people
or 40 % of the total population lived in cities with more than 1 million inhabitants (Figure
1.3). The countrys urban population share in 2009 was clearly higher than that of the
other BRIC countries (Russia, India and China). It was also slightly higher than in Mexico
and was only narrowly surpassed by Argentina (Figure 1.4).14
11
10
Figure 1.3
Concentration of people in urban
areas in Brazil
100
90
1.6
80
3.7
70
60
5.9
North
Northeast
Total:
33.3
50
40
8.5
13.6
30
Southeast
South
Central-West
20
10
0
1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008
Urban population (% of total)
Population in urban agglomerations of more than 1 million (% of total
population)
Source: World Bank World Development Indicators.
Figure 1.4
Urbanisation in the BRICs and others
100
80
60
40
20
0
Brazil
Russia
India
China
Mexico
Argentina
15
The total population used in our calculations is around 202 million, based on the LandScan Global Population
Databases (for more information, see http://computing.ornl.gov/cse_home/about/LandScan%20long.pdf).
This compares to 191 million in the 2010 IBGE census.
11
Figure 1.5
Population at flood risk in 2010
(million people)
3.7
1.6
5.9
Total:
33.3
North
Northeast
Southeast
8.5
South
13.6
Central-West
Most people, in absolute terms, are exposed to river flood risk in the states of Amazonas,
So Paulo, Para, Pernambuco and Maranho. With regard to flash floods, most people
live in risk areas in the states of Minas Gerais, So Paulo, Rio de Janeiro, Bahia and Santa
Catarina.16
The large metropolitan areas and economic strongholds of the Southeast are especially
exposed to flood risk (Table 1.2; Figure 1.6). According to our estimates, in the
municipality of So Paulo alone around 1.5 million are exposed to river flood risk, and
around 1 million to flash flood risk. Especially the River Tiet poses a severe risk of
flooding in many parts of the metropolitan region of So Paulo since its course has been
strongly manipulated and the city has expanded into the flood plain. 17 A particularly
critical point is the Marginal do Tiete where major highways meet on the river banks and
flooding can cause serious disruptions in supply chains, transportation networks, access
to major airports, etc. Given the citys importance this, in turn, can impair the states
and even the entire countrys economy. Moreover, low-income residential areas which
span the citys steep hillsides are also affected by extreme precipitation events in this
region.18
Table 1.2
Population at flood risk in the
southeastern states and metropolitan
areas
Population
So Paulo state
44 222 320
5 456 570
17 128 208
3 345 481
10.0%
21 313 029
3 944 170
11.8%
3 656 100
886 046
2.7%
12 507 966
2 641 855
7.9%
6 906 512
895 267
2.7%
33 318 370
100%
16
12
While the city of Rio de Janeiro is less exposed to river flood risk, almost 900 000
people live in hilly areas prone to flash floods. In Rio de Janeiro state, cities in the
mountainous Serra do Mar region (eg Petrpolis, Angra dos Reis, Terespolis, Nova
Friburgo) and municipalities in the economically important and densely populated Paraba
do Sul river valley (eg Volta Redonda, Resende, Barra Mansa, Barra do Pirai) are
particularly exposed. Accordingly, flood risk protection should be made a top priority,
especially in densely populatedurban areas.
Figure 1.6
Flood risk hotspots: population
exposed to river and flash flood risk in
the densely populated southeastern
region of Brazil
1.4)
Besides direct damage to the life and health of the population, floods also trigger direct
losses to physical capital. This section determines the annual expected loss to assets,
which comprises the annual expected loss of the 100 -year flood event (large loss
burden) plus losses from a more frequent event (base loss burden). The large loss
burden will be calculated in the following using the hazard-asset-vulnerability
approach.19
1.4.1) Capital stock at flood risk
Estimation of Brazils capital stock (asset module)
Our estimate of Brazils capital stock in 2010 is based on Morandi (2005) who
calculates the countrys fixed capital stock in the private and public sector as a
proportion of the GDP for the period from 1950 to 2004. Accordingly, we obtain the
total by adding together the capital stock of the sub-aggregate residences, structures,
machinery & equipment and others in both the private and the public sector. The
residential capital stock estimates are based on Reiff et al. (2001)s hedonic price
model. The capital stock of structures is obtained by subtracting the residential capital
stock from the total investment of the construction sector. Machinery & equipment
comprises the purchase of domestic and imported capital goods and Others includes
permanent cultivations, planted woods and reproductive animals. 20
Brazils total fixed capital stock in 2010 amounted to around USD 5.9 trillion (BRL 10.4
trillion) in 2010 currencies. Structures accounted for 44%, the residential sector for
38%, machinery and equipment for 15% and others for 3%. It is thus evident that the
residential sector forms a major share of Brazilian capital accumulation. Moreover, 74%
19
20
For an in-depth description of the approach and the three modules, see Swiss Re, 1998a.
See Morandi, 2005. For the applied method to derive Brazils capital stock of 2010 in 2010 currencies, see
Appendix B.1.
13
Residences
Structures
Machinery &
equipment
Others
Total
Share of
total
North
0.12
0.13
0.04
0.01
0.30
5.1%
Northeast
0.30
0.34
0.12
0.02
0.78
13.1%
Southeast
1.28
1.46
0.49
0.09
3.32
56.0%
South
0.37
0.43
0.15
0.03
0.98
16.6%
CentralWest
Brazil total
0.21
0.24
0.08
0.02
0.55
9.2%
2.28
2.60
0.88
0.17
5.93
100.0%
14
e.g. residential, agriculture, manufacturing industry, trade, corporate services, public and
private services. Intersecting the land cover types with the economic sectors provides
the building density fraction.
The area ratio is a correction factor which considers whether the percentage of a
specific land cover class in a municipality is under- or overrepresented, compared to the
distribution of the land cover class over the total investigation area:
aik = ppik / pi
aik: area ratio of land cover class i in municipality k
ppik: percentage of a polygon of land cover class i in municipality k
pi: percentage of the land cover class i in the total investigation area
Combing the building density fraction and the area ratio provides the weight by which
each polygon contributes to the total municipal capital stock:
fikj = (dij * aik) / ( i (dij * aik))
fikj: weight of a polygon of land cover class i in municipality k for assets of the
economic sector j
dij: building density fraction of land cover class i and economic sector j
aik: area ratio of land cover class i in the municipality k
Afterwards, the municipal capital stock can be disaggregated to the polygon level by
multiplying the municipal capital stock with the weight of the respective polygon (We
can also establish unit asset values by dividing the polygons capital stock by its area).
Finally, we can calculate the capital stock at risk at the raster level by intersecting the
flood risk zones with the raster units capital stock.
Source: Seifert et. al., 2008, p. 465.
We can derive the annual expected large loss burden of a 100-year flood in Brazil
based on the above considerations.26 It amounts to around USD 700 million (BRL 1.2
billion) or USD 0.12 of each USD 1 000 of capital stock. This lies within a reasonable
range, compared to the actually experienced losses during the period 2000-10 (see
section 1.2). It is about 2.5 times the decades experienced annual average loss of
USD 253 million from large events and rather close to the USD 950 million loss
experienced in 2010.
Given our capital stock estimate, the base loss burden, which was estimated at 0.12
per mill of the total capital stock27 amounts to slightly over USD 700 million (BRL 1.2
billion) for the whole of Brazil.28
Therefore, the total annual expected loss today is around USD 1.4 billion (BRL
2.5 billion) or around 0.24 per mill of Brazils total capital stock (Figure 1.7). This is
close to the cost estimates of the severe January 2011 floods which stand at
USD 1.2 billion (BRL 2.1 billion).
26
The annual expected large loss per region is obtained by multiplying the capital stock per region of the
respective asset categories by the population at risk per region and their vulnerabilities, adding them up and
dividing by 100. Adding up the regions annual expected large losses for both river and flash floods gives
Brazils annual expected large loss.
27
Swiss Re Single Snap internal pricing tool.
28
By coincidence, the base loss burden has the same share of 0.12 per mill as the large loss burden for Brazil as
a whole. While the share of the base loss burden remains the same also at state and municipal level, the share
of the large loss burden determined by our model varies.
15
Figure 1.7
Total annual expected loss 2010
(USD million)
111 122
North
180
207
Northeast
Total:
1 406
Southeast
South
Central-West
786
Again, a major driver of the exposure is rapid urbanisation in risk areas, as it leads not
only to the concentration of the population but also of economic activity. The Southeast
where more than half of the countrys economic activity29 is concentrated carries
also the majority (USD 786 million or 56%) of Brazils total annual expected loss. Up to
42% of Brazils annual expected loss can occur in the states of So Paulo (USD 414
million or 29%) and Rio de Janeiro (USD 177 million or 13%) combined. The
municipality of So Paulo alone accounts for nearly 14% (USD 193 million), and that of
Rio de Janeiro for almost 5% (USD 65 million) of Brazils total annual expected loss
(Figure 1.8).
Figure 1.8
Annual expected loss of So Paulo
state and municipality and Rio de
Janeiro state and municipality in
2010 (USD million)
By state
414
By municipality
193
177
65
200
Sao Paulo
195
620
528
400
620
600
Rio de Janeiro
800
1 000
Rest of Southeast
1 200
1 400
1 600
Rest of Brazil
The South and Northeast are the second and third most exposed regions, with annual
expected loss amounting to 15% and 13% of the total, respectively. The North and
Central West are least impacted, but must still expect economic losses of around USD
110 million to 120 million per year from floods.
29
Measured by the regional share in gross value added, which stood at 55% in 2008.
16
Total
Agriculture
15.2
Extractive
industry
12.1
Manufacturing
10.7
Construction
9.8
Minas Gerais
9.5
Esprito Santo
2.2
2.7
10.7
1.6
2.9
Rio de Janeiro
11.2
0.6
53.5
6.7
10.6
So Paulo
32.0
9.1
1.2
43.7
27.6
Southeast total
54.9
27.6
77.5
62.7
50.9
Paran
6.0
10.7
0.4
6.3
5.4
Santa Catarina
4.1
5.1
0.5
5.8
4.3
6.7
11.8
0.4
8.0
6.0
16.8
27.6
1.3
20.1
15.7
South total
Source: Instituto Brasileiro de Geografia e Estatstica (IBGE). Contas Regionais do Brasil 2004-2008.
17
The previous chapter analysed the social and economic impact of floods in Brazil today.
This chapter will do so for the year 2030. Therefore relevant future developments such
as urbanization, economic growth and the flood hazard development in the context of
climate change which lead to variations in the respective impacts will be examined. We
will choose one scenario for each of the three developments assuming that the regional
population and economic structure remains unchanged and that the hazard intensity
evolves as predicted.30
2.1)
250
100
200
80
150
60
100
40
50
20
0
1980
1985
1990
1995
2000
2005
2010
2015
2020
2025
2030
This means that due to annual urban population growth of 0.85% in 2030, almost 40
million people or 6.5 million more than today are likely to live exposed to flood risk in
Brazil. The increase will be strongest in absolute terms in the Southeast and Northeast,
where around 2.7 and 1.7 million people more are estimated to live in risk areas,
respectively (Figure 2.2).
Figure 2.2
Variation of population at flood risk in
2030 due to urban population growth
compared to 2010 (million people)
North
Northeast
Southeast
South
Central-West
0.0
0.5
1.0
1.5
2.0
2.5
3.0
30
We are aware that there might be differences in regional developments and other climate change scenarios,
but we consciously excluded them for illustrative purposes in order to discuss one scenario in detail.
31
IBGE, 2008.
18
1961-90
2030
Change
2010-2030*
111.7
106
-3.4%
Northeast
73.6
95.5
19.8%
Southeast
143.7
143.8
4.4%**
South
138.1
147.3
4.4%
Central-West
114.1
110.5
-2.1%
North
32
Marengo, 2009. He applies the A1B intermediate emission scenario, which assumes a 2.8C temperature
increase in 2090-99 compared to 1980-1999.
Marengo, 2009, p. 18. See also Mota et al., 2010. This pattern is in line with the longer term tendency from
eight global IPCC AR4 models which indicates an increase of extreme rain events in parts of the Southeastern
and in the Northern region in the years 2080 to 2099, compared to 1980 to 1999 using the same scenario
(Marengo et al., 2007, p. 46).
34
IPCC, 2007; Marengo, 2009.
33
19
The results show that the additional effect of climate change on the population at risk in
different regions of Brazil in 2030 is ambiguous (Figure 2.3). A reduction in extreme
rain events leads to an estimated decrease in the exposed population in the North by
about 250 000 and in Central West by about 40 000 people. Yet, this is likely to be
more than outweighed by an increase of about 2 million people in the Northeast and
over 700 000 in the Southeast due to more extreme events there. Overall, the hazard
intensity development could add an estimated 2.7 million people to the population at
flood risk in Brazil up to 2030.
Figure 2.3
Variation in population at flood risk in
2030 due to climate change
compared to 2010 (million people)
North
Northeast
Southeast
South
Central-West
-0.5
0.5
1.5
2.5
2.1.3)
Taking into account both population growth and climate change, the total population
living in risk zones in Brazil is estimated to rise from around 33.3 million people (16.5%
of the total population) in 2010 to 42.5 million people (19.6% of the expected total
population) in 2030 (Figure 2.4). The growing urban population contributes more than
twice as much to the increase as the hazard intensification. Like today, most people are
expected to live exposed to flood risk in the Southeast (17 million). The Northeast is the
second most exposed region, with 12.2 million people. Its share in the total population
is likely to rise from 26% to 29%. The shares of the South and Central West will remain
stable; whereas that of the North will fall from 18% to 16%.35
42.5
45
Figure 2.4
Population at flood risk today and in
2030 (million people)
2.7
40
35
30
33.3
0.85% annual
urban
population
growth rate
25
20
4.6
6.5
1.6
3.7
13.6
17.0
15
10
5
1.8
12.2
8.5
6.9
5.9
0
2010
North
Increase due to
population growth
Northeast
Southeast
South
2030
Central-West
35
Shifts in the relative shares are only due to the varying impacts of climate change per region, since population
growth is assumed to be the same in all regions.
20
2.2)
Figure 2.5
Real GDP and gross fixed investment
growth in Brazil
20
15
10
5
0
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
-10
1994
-5
36
37
21
Table 2.2
Scheduled programmes and
investments
Energia energy
infrastructure
Minha Casa, Minha Vida
housing programme
Post-2014
USD billion
BRL billion
356
627
158
278
Transportes logistic
infrastructure
60
104
32
57
Comunidade Cidada
public services for women
13
23
18
31
543
955
359
632
The lion share of the PAC 2 of USD 262 billion (BRL 462 billion) goes into the energia
infrastructure programme. Slightly over 60% of that is allocated to upstream and
downstream development of the petroleum and natural gas sector (e.g. development of
the Pr-Sal fields, construction and modernisation of refineries). About 30% of the total
is earmarked for investments in the generation and transmission of electric energy. In
generation, priority is given to renewable and competitive energy sources such as
hydropower and clean/low emission sources (nuclear and natural gas). Already today
Brazil obtains 83.5% of its electricity from hydro 41. Investment in transmission is aimed
to guarantee security of electricity and link formerly isolated sections to the national
electricity grid.
38
In 2009 it was revised upwards to USD 363 billion (BRL 638 billion). In this section we use the exchange rate
of 2010: BRL 1.7592 per USD.
Wheatley, 2010c.
40
Unless otherwise indicated, the information on the PAC 2 programmes is taken from
http://www.brasil.gov.br/pac/pac-2/
41
The global average share is 16% (Wheatley, 2010b).
39
22
Taking again the example of the states of So Paulo and Rio de Janeiro, it becomes
evident that already the existing energy infrastructure is exposed to flood risk in several
areas (Figure 2.6). Furthermore, there is a possibility that the transmission of energy and
gas to major urban areas and to the refineries could be interrupted, as parts of the major
lines and pipes run through risk areas.42 Future energy infrastructure investments should
therefore take flood risk into account. Especially since blackouts and energy rationing
(like e.g. in 2001 and 2002 which was, however, due to a lack of rain in hydropower
generation) could have severe detrimental effects on economic growth and Brazils
attractiveness to foreign investors43.
Figure 2.6
Map of critical energy networks in
Southeastern Brazil exposed to flood
risk
The PAC 2 assigns the second highest investment amount of USD 158 billion (BRL 278
billion) to the Minha Casa Minha Vida programme. It aims to reduce the housing deficit
of about 6.2 million residences, stimulate the civil construction sector and create
employment. Nearly two thirds of the funds is for providing financing opportunities for
the construction and acquisition of housing units via the Brazilian System of Savings and
Loans (SBPE). About one fourth of the total will be used for the construction of two
million houses, particularly for low income families. The remaining 11% will be
channelled into the urbanisation of precarious dwellings by providing better, regularised
housing and access to public utilities (water, sewerage, electricity) and services (health,
education, sports and culture).
As this programme aims to build large new housing stock, some significant exposure of
residential values to flood risk can result, depending on the way they are built and where
they are located. Yet the programme also offers an opportunity to guide the new
construction efforts and residential infrastructure development to more secure areas. If
e.g. flood (micro-)insurance were included in the obligatory residential insurance that
comes with the housing loan44, it could set incentives through risk adjusted premiums
for future homeowners to build outside risk areas. It would also transfer the loss burden
away from them. The programmes investments and incentive structure, therefore, have
to be carefully planned.
42
The Santa Cruz thermal power plant, for example, which is central for connecting Rio de Janeiro states
electricity grid to the rest of the country lies at Canal de So Francisco (Guandu river) (see
http://www.furnas.com.br/hotsites/sistemafurnas/usina_term_stacruz.asp).
43
Wheatley, 2010b.
44
See section 3.2.3 for more information.
23
The PAC 2 allocates USD 60 billion (BRL 104 billion) to the transportes programme.
Almost half of that (46%) goes into upgrading the road network, including road
duplication, paving, urban circumnavigations, as well as improved access to ports
(particularly for commodity exports). The main goals are to eliminate congestion points
on strategic axes, include formerly isolated regions and to reduce transport costs overall.
Another 42% is provided for investments to expand the railway system by around 4 700
km. It aims to create a modern, high-capacity network, connect the main urban centres
and link the agricultural and mining regions to the industrial areas, consumer markets
and ports. The investments include three viability studies for almost 2 000 km of high
speed train links (TAV) between the urban centres So Paulo-Curitiba, Campinas-Belo
Horizonte and Campinas-Triangulo Mineiro. Yet the first project of this type the high
speed rail link between Rio de Janeiro and So Paulo as part of the first PAC has not
left the planning phase yet and will not be operational before 2017. Furthermore, flood
risk has to be taken into account during construction planning as its route may run
through the risk areas along the Paraba do Sul river valley (Figure 2.7).
Figure 2.7
Map of critical transportation routes
in Southeastern Brazil exposed to
flood risk
Finally, the Cidade Melhor programme totalling USD 32 billion (BRL 57 billion) aims to
take up the major challenges of the big urban agglomerations. 39% of the investment
funds are channelled into expanding sanitation and sewer treatment systems and 32%
into improving public transport networks, such as metros, trams and buses. It also
earmarks USD 6.7 billion (BRL 12) or one fifth of the total for risk prevention. This
includes flood control, upgrade of the drainage system and hillside stabilisation in order
to reduce the loss of lives and to protect infrastructure, economic activity and tourism
from flooding and landslides.
Annual expected loss in 2030 due to asset growth
Given the forecasted rapid expansion of the Brazilian economy and its capital stock, the
exposure to flood risk is also likely to rise. In this section, we will derive an estimate for
the annual real asset growth up to 2030 to calculate the annual expected loss due to
asset growth in this year.
We will split the period up to 2030 into two periods. The first will last until the 2016
Olympic Games in Rio de Janeiro and will see a high real asset growth rate due to the
immense investment requirements. PAC 2 will last until the end of 2014, yet as
experience from PAC shows, implementation often extends beyond the initial period.
Therefore, it is safe to assume that the above described PAC 2 investments will last until
24
2016. Given the EIUs forecast of 8.8% average annual growth in fixed asset investment
up to 2015, we estimate real asset growth of 8 % for the period 2011-2016.
Afterwards the investment growth spurt is likely to cool down because the immediate
investment requirements decrease. It also remains to be seen if PAC 2 will be followed
by another major investment plan. We assume that during the period 2017 to 2030
real asset growth will return to a similar level as in the 15 years preceding the global
economic crisis in 2008/9, when it stood at 4.6%. We will use a slightly more moderate
average growth rate of 4% for 2017-2030. 45 This still quite high growth rate is
reasonable as the Brazilian economy will still offer considerable growth potential and
continue to attract investments. Furthermore, the socio-political environment is
forecasted to remain stable, partly because of the success of social inclusion
programmes, which is crucial for further investments.
The annual expected loss of the Southeast in 2030 is estimated to rise the strongest
by around USD 1.3 billion (BRL 2.3 billion). It is followed by the South, with around
USD 350 million (BRL 616 million) and the Northeast with around USD 300 million
(BRL 528 million).46 Summing up the results of all the regions, the annual expected loss
in 2030 for the whole of Brazil due to asset growth is estimated to be almost
USD 2.5 billion (BRL 4.4 billion) higher than at present (Figure 2.8).
Figure 2.8
Variation of annual expected loss in
2030 due to asset growth compared
to 2010 (USD million)
North
Northeast
Southeast
South
Central-West
0
200
400
600
800
1 000
1 200
1 400
Figure 2.9
Variation of annual expected climate
loss in 2030 compared to 2010 (USD
million)
Northeast
Southeast
South
Central-West
-25
25
50
75
100
From Figure 2.9 it is evident that hazard development through to 2030 has an
ambivalent effect. The North and Central-West will experience decreases of annual
45
In total, it results in an average decadal real capital stock growth rate of 87% between 2010 and 2030,
which is moderately higher than past rates averaging 71% between 1970 and 2000 (see Figure A.1.2 in the
Appendix).
46
Lacking regional forecasts we apply the countrys growth rates to all the regions. The BRL values are based on
the 2010 exchange rate.
25
expected losses, yet this fall will be outweighed by increased losses in the Northeast,
Southeast and South. The Northeast and Southeast will experience the strongest rise
up by almost USD 100 million (BRL 176 million) each, while in the South the rise will
stand at an estimated USD 25 million (BRL 44 million). In total, Brazils annual expected
loss could be close to USD 200 million (BRL 352 million) higher in 2030 than 2010,
merely due to the intensification of severe precipitation events.
2.2.3) Total annual expected loss in 2030
Including both the variation due to asset growth and climate change, the total annual
expected loss from flooding in Brazil is estimated to nearly triple from USD 1.4 billion
(BRL 2.5 billion) to reach a little over USD 4 billion (BRL 7 billion) between 2010 and
2030 (Figure 2.10). In general, asset growth and concentration will contribute above
ten times more to this increase than climate change. The bulk of the annual expected
loss in 2030 of up to an estimated USD 2.3 billion (BRL 4 billion) is likely to be borne by
the states in the Southeast. This is followed by the Northeast and South, with almost
USD 600 million (BRL 1.1 billion) each. The share of the Northeast, where the flood
hazard is predicted to intensify strongest, will move up from 13% to 15%. 47 The share of
the North and Central West will decline by around 1 percentage point each, while the
shares of the Southeast and South will remain stable.
Figure 2.10
Annual expected loss 2010 and 2030
(USD million)
4 500
+201
+2 455
4 000
194
3 500
25
98
96
-7
-11
361
4 062
298
593
3 000
1 373
2 500
2 255
2 000
1 406
1 500
314
213
111
207
1 000
786
592
180
122
324
500
0
2010
North
Increase due to
asset growth
Northeast
Southeast
South
2030
Central West
47
The variation in the regions relative shares over the 20 years is modest, as equal asset growth rates are
assumed for each region and, thus, shifts are only due to varying hazard developments.
26
It has become clear that floods are having severe social and economic impacts in Brazil
already today and that this will be even more the case in 20 years time, given the fact
that the country is likely to experience continued urbanisation and rapid economic
development. In this section we analyse how the countrys flood protection system
works and where the loss burden is allocated today, before going on to discuss future
adaptation measures and risk transfer solutions.
3.1)
Effective protection against flood risk requires the interplay between the state, the
re/insurance sector and property owners. This is especially true if the risk complexity
and loss burden increase rapidly, as in the case of Brazil, because a single sector can no
longer address the challenge on its own. 48 In the following, we examine the Brazilian
states flood management system and its flood re/insurance market.
3.1.1) State level: prevention and disaster relief
To evaluate the risk and plan prevention measures and emergency and reconstruction
actions for major natural and man-made disasters, the National System of Civil Defence
was established in Brazil in 1988. 49 The central body is the National Secretariat of Civil
Defence, part of the Ministry of National Integration, which coordinates civil defence
measures in the entire national territory at federal level. At the other administrative levels
there are regional civil defence and state and city coordination bodies. Planning is
conducted using master plans at all administrative levels, which comprise contingency
plans laying down the responses to different disasters. The following are specified as
major natural perils.50: For the North, floods and forest fires; for the Northeast, droughts
and floods; for the Central West, forest fires; for the Southeast, landslides and floods;
and for the South, floods and heavy winds.
The financial resources for the actions are provided by the civil defence budgets at
federal, state and municipal level. Looking at the respective federal budget lines for
prevention and emergency programmes51 major mechanisms of the Ministry of
Integration to provide funds to the municipalities illustrates that payouts for
emergency relief clearly trumped those for preventive measures in the period 20042010 (Table 3.1, 3.2). The former stood at around USD 2.6 billion (BRL 4.8 billion)
significantly higher than for prevention which amounted to USD 280 million
(BRL 540 million).52 Moreover, not even one fourth of the authorised allocation for
prevention measures has been paid out, whereas in the case of emergency relief and
reconstruction nearly two thirds were disbursed.
Table 3.1
Federal disaster prevention and
preparation payouts
Year
Total allocation,
in USD million
Paid out,
in USD million
2004
44.2
3.4
2005
58.5
8.7
15
2006
50.7
16.9
33
2007
135.0
27.5
20
2008
336.2
61.4
18
2009
323.4
69.1
21
2010
241.6
95.2
39
Total
1 189.6
282.2
24
48
27
Table 3.2
Federal emergency relief payouts and
the recorded economic losses from
large flood events
Year
Total allocation,
in USD million
Paid out,
in USD million
Economic loss
from large floods,
in USD million*
2004
44.4
26.4
59
350
2005
104.4
44.0
42
2006
117.0
59.0
50
2007
284.7
178.7
63
131
2008
637.3
262.5
41
673
2009
961.6
693.3
72
517
2010
1 731.1
1 309.4
76
947
Total
3 880.5
2 573.3
66
2 618
This situation may have its roots in the organisational structure of the National System of
Civil Defence. The prevention funds are only freed up if a municipality presents a
detailed outline of its prevention projects (e.g. including environmental licenses).
However, often this does not occur because of a lack of qualified personnel or financial
resources to produce such documentation or due to the lack of risk awareness and
political will locally.53 But changes are also needed at federal level, where authorities
should ideally assess if they could increase the payout ratio by making access to the
funds easier.
Regarding the loss burden allocation, it becomes evident that federal emergency relief
payouts and the recorded economic losses from large flood events were within a close
range of around USD 2.6 billion from 2004 to 2010 (Table 4.2). Comparing the two is
troublesome as the first not only includes floods, but all natural and man-made
disasters.54 Nevertheless, it is safe to assume that flood emergency payouts have a high
share in the total expenses, as floods are the most frequent natural catastrophe.
Furthermore, the flood expenses are likely to be not as high as the recorded loss, since
not all economic losses are compensated.
Box 4: Activities of the Brazilian Development Bank and World Bank
The state-owned Brazilian Development Bank (BNDES) is also active in flood emergency
relief, especially in reconstruction funding. It has established programmes for the
provision of low-interest rate loans to (micro)enterprises in affected areas. After the
January 2011 floods in Rio de Janeiro state, it created a programme of USD 227 million
(BRL 400 million) for emergency loans and a second one for debt refinancing. Similar
programmes had been created after floods in the states of Santa Catarina, Pernambuco
and Alagoas before.
Natural disaster risk will be one of the central topics of the World Banks new
partnership strategy with Brazil (which will be approved in July 2011 for 2012-2015).
The World Bank has been active in the provision of funds for disaster prevention and
capacity building. After the January 2011 floods, it announced a USD 485 million
(BRL 853 million) loan for disaster prevention and housing available to the government
of Rio de Janeiro state (in connection with the Metropolitan Urban and Housing Project
which provides better and safer housing to 2 million people in the Rio de Janeiro
metropolitan area). Moreover, it promotes programmes to improve the authorities
capacity to plan and respond to natural disasters.
Source: BNDES, 2011; World Bank, 2011.
53
Buani, 2011; Downie, 2011. According to a report filed by the secretary of the (federal) National Civil
Defence authority to the United Nations, nearly 25% of Brazilian cities do not have a civil defence authority, as
required by the National Policy of Civil Defence.
54
We lack a more detailed disaggregation of the expenses.
28
While typical household insurance does not include flood cover, it can be included under
a comprehensive combined building and contents household policy. Residential all risk
lines, in general, generated premiums of USD 505 million (BRL 888 million) or USD 3
per capita in 2008, corresponding to a very low penetration of 0.032%. In order to
increase competition and reduce premiums in the personal lines market (household and
personal accident), the government recently lifted the ban on life insurers to provide
such non-life policies.56
Motor insurance is one of the few sectors where there have been insured flood losses in
the past. With premium incomes of USD 10.9 billion (BRL 19 billion), motor insurance
was the most important non-life personal line in 2008. Flood cover, according to local
estimates, is included in 90% of voluntary comprehensive motor insurance which,
however, supposedly accounts for only a very low premium fraction. The lion share of
73% of the premiums comes from hull insurance and obligatory motor insurance
(DPVAT).57 Consequently, also flood motor insurance has a low penetration.
Basic industrial and commercial property policies cover fire, lightning and explosion. To
cover water damage, additional optional policies have to be bought. Primary insurers,
however, often exclude or sub-limit flood cover for large industrial or commercial risks,
particularly rail- and highways.58 One reason for the primary insurers limited flood
coverage in the large risk segment might have been that the Instituto de Resseguros do
Brasil formerly the monopoly reinsurer in Brazil offered only proportional contracts. 59
Construction and erection all risk policies (CAR/EAR) generated premium incomes of
around USD 235 million (BRL 413 million) or around 1.4% of total non-life premiums in
2009. There will be significant additional premium potential due to the PAC and PAC 2
programmes (around USD 4.5 billion (BRL 8 billion) over the next 6 years) as well as the
2014 FIFA World Cup (around USD 550 million (BRL 970 million)) and the 2016
Olympic Games in Rio de Janeiro.60 Infrastructure projects like these, and transport and
55
29
energy infrastructure in general, can be exposed to flash and river flood risk, especially in
the flood prone states in the Southeast (Figure 3.6, 3.7) and South, but also in the
Northeast.
The overview shows that flood insurance penetration is low, since flood cover is optional
in most policies. In some important segments, such as household and industrial and
commercial lines, primary insurers are hesitant to assume the risk. Increased reinsurance
options/capacities following the gradual opening of the reinsurance market since 2007
might lead to more primary insurer activity in both fields.61
3.2)
Today, post-event measures dominate Brazils flood protection system and the bulk of
the loss burden is borne by public budgets and, ultimately, by the taxpayers. Prevention
and adaptation measures could reduce the losses and burden on budgets. Therefore, we
discuss possible adaptation measures in section 3.2.1 below. We also outline how,
through risk transfer solutions, the loss burden could be shifted from public and personal
budgets to the re/insurance sector (section 3.2.3).
3.2.1) Risk prevention: adaptation measures
Several adaptation measures have been identified as relevant and feasible since they are
currently debated in public discussions. We assess their loss aversion potential and
make a coarse-grain cost-benefit analysis .62 Averted losses are usually calculated by
running a probabilistic expected loss model incorporating the adaptation measures:
Some measures may reduce the impact severity of the hazard (e.g. drainage systems),
others may alter the vulnerability of asset classes (e.g. building codes) or the asset
distribution (e.g. relocation).63 As we do not dispose of a probabilistic model, we rely on
expert judgment on the loss aversion potential of the different measures.64
Drainage projects
The Cidade Melhor programme which forms part of PAC 2 earmarks USD 5.9 billion
(BRL 10.4 billion) for drainage projects from 2011 to 2014. Half of it is offered to
municipalities which present plans for such projects. 65 Especially in urbanised areas, the
burden of water run-off on drainage systems has increased as more impervious areas
have been created. Estimates for the River Belm basin in the metropolitan area of
Curitiba (Paran state), which has a drainage system of 42 km2 and around 60% of
which is impervious, indicate that the medium flood height increased six times from
when the area was still rural to todays urbanisation level.66
Drainage systems tend to reduce the effect of small and medium, more frequent floods
as the water is collected and safely transported off. According to expert estimates, they
decrease the base loss burden by 30% on average. They do not have an effect on the
large loss burden as the water run-off from large, less frequent floods usually overstrains
the capacity of drainage systems. In view of this and assuming that the projects will be
operational at the end of 2014, drainage systems have the potential to reduce the
annual expected loss in 2030 by USD 587 million (BRL 1 billion).
The sum of the present values 67 of the averted losses amounts to almost USD 3.4 billion
(BRL 6 billion) for the period 2015-2030. 68 If we assume that the cost is equally split
61
Recently, Brazil reintroduced some protection for the domestic industry, obliging insurers to write 40% of the
reinsurance with a local reinsurer (Bradford, 2011).
62
The basic approach is adopted from the Economics of Climate Adaptation Working Group (2009).
63
Economics of Climate Adaptation Working Group, 2009, p. 133.
64
For a more exact cost-benefit analysis, besides a probabilistic loss model, a detailed description of the form
and location, as well as the quality and quantity of the applied measures, is required.
65
Sibaja, 2011.
66
Tucci, 2003, p. 89-92.
67
The discount rate is 6%, which is applied in a study that analyses the financing of infrastructure investment in
Brazil by pension funds (Morandi da Silva et al., 2004). The Agncia Nacional de Transportes Aquavirios
(2010) uses an even higher 8.3% for port infrastructure investment. Given that PAC 2, in addition to private
funds, also includes public investment, which usually has a lower discount rate due to lower risk, the use of th e
lower estimate of 6% is adequate.
68
Calculated by deriving 30% of each years annual expected loss, discounting it to the present value at the end
of 2010 and adding up the results for all years in the period 2014-2030.
30
over the four-year investment period of PAC 2, the present value of the cost is roughly
USD 5.1 billion (BRL 9 billion), exceeding the benefit by about 50%. At first glance,
therefore, investing in drainage systems is not economically profitable, as the costaverted loss ratio is higher than 1. This is a risk-neutral perspective, however. In reality,
most people are risk averse and, therefore, might be satisfied with drainage investments
that pay back about two thirds of their cost.69 Moreover, we have to consider additional
benefits which are hard to quantify: Better drainage systems can lower the indirect
(tangible) flood effects, such as losses from business or infrastructure interruption.
Moreover, they can reduce adverse health effects, such as the risk of water-borne
diseases, or even save lives (direct intangible effects), as the standing period of the
water and the strength of the flood wave is reduced.
Hillside stabilisation
USD 800 million (BRL 1.4 billion) of PAC 2s Cidade Melhor programme flow into
hillside stabilisation projects over the period 2011-2014. Landslides in mountainous
areas, such as the Serra do Mar, have caused large losses in recent years. They are
triggered by heavy rains which, if absorbed by the soil, increase subsurface water flows.
These erode soil particles, reducing the friction and triggering the landslide. There are a
variety of measures available to stabilise steep slopes, e.g. terracing, slope flattening,
soil nailing, soil water pump-out during dry periods, retaining walls etc.. 70
According to expert estimates, slope stabilisation measures can cut the annual expected
loss from flash flood by 10%. As flash flood events usually are local, locally constrained
protection measures only have small scale effects. Assuming that the projects will be
operational at the end of 2014, they could reduce the annual expected loss by USD 94
million in 2030 (BRL 165 million). The sum of the present values of the averted losses
amounts to around USD 540 million (BRL 950 million) for the period 2015-2030. The
present value of the cost, equally split over the four-year investment period, is roughly
USD 695 million (BRL 1.2 billion). It surpasses the benefit by about one third, which
might lie within a reasonable range for a risk adverse investor. Furthermore, direct
intangible effects, such as saved lives, are not included. These measures might also have
cross benefits, such as higher property values in the now safer areas.
Building codes
Building codes direct the construction of houses with regard to structural aspects, used
materials and hydraulic pressure resistance. Rather economic and simple options to
protect houses from flooding include elevation, which raises the lowest inhabitable floor
above the base flood elevation point, or the use of waterproof materials to seal the
shelters openings.71 In the following, we have taken the example of sealing doors and
windows on the ground floor for our further discussion.
According to expert estimates, building codes which implement the use of waterproof
building materials can reduce the total annual expected loss from flooding by 20%. In
2030 the averted loss could, therefore, amount to up to USD 770 million (BRL
1.4 billion). Given the legislative process to pass such codes, we assume that they could
enter into force by the beginning of 2013. For the period 2013-2030, Brazil could
avert an estimated USD 5 billion (BRL 8.8 billion) in present values through such codes.
On the cost side, we have estimated the expenses for upgrading the ground floor of
houses below (Table 3.3). According to expert estimates, sealing could cost on average
up to 5% of the capital value of the ground floor . To calculate the capital value, we take
the capital stock per house in 2010 72 and assume that 50% of it is located on the
ground floor. By multiplying the cost with the number of houses at flood risk 73, we
obtain the cost of around USD 3.1 billion (BRL 5.5 billion) for the building upgrade in all
69
31
houses at risk. If the cost is split equally over the 20 years until 2030, the cost has a
present value of around USD 1.8 billion (BRL 3.2 billion). As that amounts to just one
third of the estimated averted loss, the introduction of building codes which prescribe
the sealing of the buildings openings on the ground floor would be cost-efficient, even
from an economic standpoint.
Table 3.3
Cost of building upgrade in all houses
at risk, 2010 (in USD of 2010)
38 468
19 234
Cost upgrade ground floor per house (5% of capital stock ground floor)
962
3 215 082
3 091 907 786
Source: IBGE, Nacional por Amostra de Domiclios 2008-2009; Swiss Re's own calculations.
32
Adaptation measures
Costs
Drainage projects
5 900
Averted
loss in
2030
587
800
Hillside stabilisation
Building codes
3 092
Urban planning
N/A
288
0.21
PV
cost*
PV averted
loss*
5 111a
3 380 c
Costaverted
loss ratio
1.51
94
693 a
539 c
1.29
772
5 049 d
0.32
1 773
Notes: PV=present value; * end 2010; a costs equally split over 2011-2014;
c
4 500
4 000
3 500
4 062
3 000
- 772
- 587
2 500
- 94
2 610
2 000
1 500
1 000
500
0
2030 without
adaptation
77
Hillside
stabilisation
2030 with
adaptation
We do not breakdown the annual expected loss by region, as information on the regional distribution of the
measures is not available.
33
78
34
process.87 A parametric trigger for payouts would simplify the loss adjustment and claims
turnaround process. Parametric triggers use observable characteristics of an event such
as amount of rainfall in a particular area and period to rapidly and transparently
determine the payment due under the insurance contract. This spares resource-intensive
loss adjustments and secures swift and transparent payout of funds after a disaster. In the
case of river and flash floods, identifying a trigger which is effectively measurable and
correlated with the actual losses is a challenge. Possible options are indices based on
precipitation amounts, flood depths or flood boundaries which can be either determined
on the ground or via satellite data.
Furthermore, to counter the underlying that only those people who are most exposed buy
insurance (issue of adverse selection)88, it is crucial that flood microinsurance would be
mass-marketed beyond the immediate risk areas. Possible options are the inclusion in
World Bank programmes or the expanding mortgage financing of the Minha Casa
Minha Vida programme89. Reinsurers can play an important stabilising role in such a
microinsurance scheme, as they bring in their expertise and experience with innovative
transfer mechanisms, expand the underwriting capacity and take over the extreme
risks.90
Box 6: Case study catastrophic microinsurance in Haiti
The millions of small-scale traders who make up the informal sector in Haiti are
particularly vulnerable to natural disasters. To address this challenge, Swiss Re formed a
partnership with Haitis largest microfinance institution, Fonkoze, to design a microinsurance scheme for catastrophes in Haiti. The project will allow highly vulnerable
Haitians to protect themselves from natural disasters, including extreme rain events, at
reasonable cost.
Clients of microfinance institutions are the primary clients of this micro-insurance. In the
event of a natural disaster, the insurance will provide for a lump-sum payout directly to
affected clients, as well as the repayment of the clients loan thus protecting the client
through both short and medium term economic assistance. Starting in early 2011,
Fonkoze was the first partner to offer the product to its 45 000 credit clients, and the
programme is now being expanded to new microfinance companies around the world.
In June 2011, under the terms of the coverage, Swiss Re made a major payment to the
Fonkozes Haitian clients within two weeks of a major rainfall event, These funds should
allow the microfinance borrowers to pursue their daily lives and livelihoods.
Source: World Economic Forum, 2011; Swiss Re's own sources.
Floods can also pose a severe threat to Brazils public infrastructure, as critical transport
and energy routes run through risk areas (Figure 2.6, 2.7). Besides careful planning of
future investments also with regard to flood risks the public sector can limit the
financial burden of flooding by purchasing re/insurance for its infrastructure. This would
allow the different levels of government to reduce budget volatility and to free up
resources for alternative investments.
Severe flood events such as a 100 year flood can place significant financial strain on
public budgets. People have to be evacuated and temporary housing, food, medication
provided. Public transport, energy, education and health infrastructure have to be
rebuilt. In addition, given Brazils low insurance penetration, the public sector may also
have to support private sector reconstruction efforts (e.g. via subsidised low interest rate
loans).
One option is to purchase reinsurance for public catastrophe reserves which protects
from losses exceeding a certain retention level. Public authorities can also transfer peak
87
See eg MiCRO catastrophe risk insurance scheme in Haiti (Guy Carpenter, 2011).
Swiss Re, 1998a, p. 29.
89
Low-interest rate housing loans of the Minha Casa Minha Vida programme are only available in connection
with insurance against death, disability and property damage. The premiums which in some cases amount
to up to 20% of the total cost must be included in the loan (AXCO, 2011, p. 44). Such mortgage financing
formed the largest part of the completed projects under the PAC between 2007 and 2009 (Wheatley,
2010c).
90
Swiss Re, 2010a, p. 29.
88
35
disaster risks to capital market investors via catastrophe bonds. If a catastrophe strikes,
cat bonds offer public authorities rapid access to funds which they can deploy for
emergency and reconstruction efforts.
Box 7: Case study MultiCat Mexico
In terms of lives lost, the Mexico City earthquake of 1986, which measured 8.1 on the
Richter scale, was the worst disaster to hit Mexico in recent decades, resulting in over
9 500 fatalities. But in economic terms, Hurricane Wilma, which hit in October 2005,
was the most devastating. Wilma caused total damages of over USD 22 billion, but only
USD 13.8 billion of that was insured.
As early as the 1990s, the Mexican government identified disaster risk reduction as a
national priority, creating the Fund for Natural Disasters (FONDEN) in 1999 to improve
its financial preparedness for natural disasters. In 2009, with the intention of further
enhancing FONDENs financial response capacity, Mexico sponsored the first
transaction in the MultiCat Programme, a catastrophe bond shelf programme arranged
by the World Bank in collaboration with Swiss Re.
MultiCat used parametric triggers to transfer USD 290 million of Mexican earthquake
and hurricane risk to the capital markets, and built on Mexicos experience from an
earlier transaction that was designed to cover earthquake risk only. The MultiCat
Programme may be used by public sector sponsors and demonstrates how governments
can use reinsurance and capital markets to reduce the pressure on public budgets in the
event of a natural disaster.
With advances in flood mapping and monitoring technology, such risk transfer solutions
may also be able to help Brazil address the challenges of catastrophic floods.
Source: Swiss Re, 2011.
91
36
Conclusion
37
Conclusion
line for emergency relief for all natural and man-made disasters between 2004 and
2010, around nine times more than for disaster prevention during the same period.
Shifting more public and private funds to preventive measures could potentially reduce
the losses and the strain on the budgets in the long run. According to our estimates,
building codes which enforce the sealing of doors and windows on the ground floor
could reduce the annual expected loss by more than USD 750 million (BRL 1.3 billion)
in 2030. The drainage projects which are planned to be implemented at a cost of USD
5.9 billion (BRL 10.4 billion) in the context of PAC 2 could potentially lower the
expected economic loss in 2030 by nearly USD 600 million (BRL 1.1 billion). The
additional hillside stabilisation measures of PAC 2 costing USD 800 million (BRL 1.4
billion) could avert losses of over USD 90 million (BRL 158 million) in that year. Urban
planning directing settlements to safe areas can significantly lower the flood exposure of
people and assets. Further adaptation measures, such as an early warning system and
awareness campaigns, have the potential to reduce the number of victims from flooding.
Risk transfer solutions to prepare for the financial impact of floods
Besides such adaptation measures, efforts should be undertaken to facilitate the transfer
of risks and losses. Insurance can offer a broad product variety which is crucial to
financially protect homeowners and businesses from flood risks which cannot be
prevented. As it sets incentives for risk-reducing behaviour (eg retention levels, riskadjusted premiums) re/insurance is likely to be more effective in dealing with
catastrophe risks than the currently dominating emergency relief measures. Open
access to international reinsurance markets, which provide capacity and knowledge,
allows primary insurers to provide flood cover on a broader scale.
A possibility to increase the insurance penetration among the strongly exposed poor
population is a flood homeowner microinsurance scheme. A classical scheme covering
fire, explosion, lightning has been shown, in theory, to offer a market potential of about
42 million homes or up to USD 780 million. In order to include flood cover at still
affordable premiums, the possibilities of implementing group pricing, a simple claims
turnaround process (eg parametric trigger) and mass marketing (e.g. in connection with
the Minha Casa Minha Vida programme) have to be explored.
Public sector solutions offer ways to protect public budgets from the financial burden
and volatility of flood emergency expenses. One option is to purchase reinsurance for
public catastrophe reserves which shields from losses exceeding a certain retention
level. Moreover, catastrophe bonds can be used to transfer peak disaster risks to capital
market investors. They also provide public authorities with rapid access to funds which
they can then spend on emergency and reconstruction efforts. Public-private
agriculture/weather insurance schemes in Brazils highly important agricultural sector
could compensate farmers for crop failures and increase their willingness to invest in
fertilizers and seeds.
Avoid major obstacles to Brazils socio-economic development
To conclude, Brazil would benefit from increasing its focus on flood prevention, adaptation
and risk transfer to ensure that the impacts of flooding do not present obstacles to the
countrys rapid economic development and its poverty reduction efforts. As the Brazilian
economy further expands, investment surges and urbanisation continues to mount, the
need for effective risk solutions will increase. The re/insurance sector stands as a strong
partner for the public sector, corporations and individuals to offer risk assessment and
financing solutions.
38
Appendix
Appendix A: Figures and tables
Table A.1
Population at river and flash flood risk
per state92
State
698 921
193 180
303
Population
share at
river flood
risk, in %
27.64%
3 387 118
343 000
77 813
10.13%
2.30%
555 548
46 344
1 124
8.34%
0.20%
3 360 086
2 925 861
1 410
87.08%
0.04%
Bahia
15 393 971
990 726
1 173 494
6.44%
7.62%
Cear
8 818 658
473 640
382 971
5.37%
4.34%
Distrito Federal
2 399 376
1 386
149 912
0.06%
6.25%
Esprito Santo
3 656 100
309 434
576 612
8.46%
15.77%
Goias
5 973 574
280 243
257 905
4.69%
4.32%
Maranho
6 656 816
1 096 861
55 970
16.48%
0.84%
Mato Grosso
3 032 578
371 395
138 467
12.25%
4.57%
2 498 586
320 953
48 204
12.85%
1.93%
21 313 029
998 446
2'945 724
4.68%
13.82%
Par
7 395 373
1 661 379
55 243
22.47%
0.75%
Paraba
4 051 962
255 569
129 597
6.31%
3.20%
Paran
11 462 868
713 206
350 668
6.22%
3.06%
Pernambuco
9 498 650
1 133 010
569 480
11.93%
6.00%
Piau
3 445 328
1 035 070
67 510
30.04%
1.96%
17 128 208
830 069
2 515 412
4.85%
14.69%
3 269 831
335 668
93 719
10.27%
2.87%
12 238 979
643 509
671 184
5.26%
5.48%
1 637 616
578 637
12 477
35.33%
0.76%
384 088
223 237
5 078
58.12%
1.32%
Acre
Alagoas
Amapa
Amazonas
Minas Gerasi
Rio de Janeiro
Rio Grande do Norte
Rio Grande do Sul
Ronondia
Roraima
Santa Catarina
Population
Population at
river flood
risk
Population at
flash flood
risk
Population
share at
flash flood
risk, in %
0.04%
6 347 280
394 546
884 271
6.22%
13.93%
44 222 320
2 728 204
2 728 366
6.17%
6.17%
Sergipe
2 125 857
261 804
57 270
12.32%
2.69%
Tocantins
1 394 681
188 768
34 041
13.53%
2.44%
13 984 225
9.55%
6.91%
So Paulo
Brazil total
19 334 145
92
Table A.1 also includes the relative numbers of flood risk exposure. These provide an insight into how many
people are exposed to river and flash flood risk in relation to the states overall population. Regarding the most
densely populated states, however, the relative numbers suggest an underestimated picture of the exposure
as still a high number of people can be affected.
39
Appendix
Table A.2
Fixed capital stock 2010 (USD million
of 2010), private and public sector, by
state
State
GDP share
weight
2008, in %
Residences
Structures
Machinery
&
equipment
Others
Total
Rondnia
0.6
13 692
15 600
5 275
1 010
35 577
Acre
0.2
4 564
5 200
1 758
337
11 859
Amazonas
1.5
34 231
39 000
13 187
2 525
88 943
Roraima
0.2
4 564
5 200
1 758
337
11 859
Par
1.9
43 359
49 401
16 704
3 199
112 662
Amap
0.2
4 564
5 200
1 758
337
11 859
Tocantins
0.4
9 128
10 400
3 517
673
23 718
Maranho
1.3
29 667
33 800
11 429
2 189
77 084
Piau
0.6
13 692
15 600
5 275
1 010
35 577
45 641
52 001
17 583
3 367
118 591
0.8
18 256
20 800
7 033
1 347
47 436
Paraba
Pernambu
co
0.8
18 256
20 800
7 033
1 347
47 436
2.3
52 487
59 801
20 220
3 872
136 380
Alagoas
0.6
13 692
15 600
5 275
1 010
35 577
Sergipe
0.6
13 692
15 600
5 275
1 010
35 577
Bahia
Minas
Gerais
Esprito
Santo
Rio de
Janeiro
91 282
104 001
35 166
6 734
237 182
9.3
212 230
241 803
81 761
15 656
551 449
2.3
52 487
59 801
20 220
3 872
136 380
11.3
257 870
293 803
99 343
19 023
670 040
So Paulo
33.1
755 355
860 609
290 997
55 723
1 962 684
5.9
134 640
153 402
51 870
9 932
349 844
4.1
93 564
106 601
36 045
6 902
243 112
6.6
150 615
171 602
58 024
11 111
391 351
1.1
25 102
28 600
9 671
1 852
65 225
1.7
38 795
44 200
14 945
2 862
100 803
2.5
57 051
65 001
21 979
4 209
148 239
3.9
89 000
2 282 121
101 401
2 600 122
34 287
879 178
6 566
168 353
231 253
5 929 559
Cear
Rio
Grande do
Norte
Paran
Santa
Catarina
Rio
Grande do
Sul
Mato
Grosso do
Sul
Mato
Grosso
Gois
Distrito
Federal
Brazil
total
40
Appendix
Appendix B: Methods
B.1 Derivation of Brazils 2010 capital stock in USD of 2010
Our estimate of Brazils total capital stock in USD for the year 2010 is based on Morandi
(2005)s capital stock estimates of 2004 in BRL of 2000. We will use the official
exchange rate of 2010 when USD 1 was worth BRL 1.759293. Furthermore we have to
account for the capital stock growth between 2004 and 2010 and the inflation
development between 2000 and 2010. We apply two different methods. In the first,
we add the annual gross fixed capital formation (GFCF) in USD of 2000 to the initial
2004 capital stock (Table B.1.1). In the second, based on Morandis decadal estimates
of Brazils capital stock since 1950, we derive the average decadal growth rate of the
capital stocks proportion of the GDP (7.3% per decade) 94 and add it to the capital
stocks GDP proportion of 2000 to obtain the proportion for 2010 (Table B.1.2).
By multiplying it with Brazils GDP of 2010 in 2000 currencies, we obtain the capital
stock of 2010 in 2000 currencies. Furthermore, to include the inflation development
between 2000 and 2010, we add the rise of Brazils consumer price index (CPI) of
90.25% 95 in this period to our results. Both methods, finally, yield very similar results of
Brazils total capital stock of 2010 of about USD 5.9 trillion (BRL 10.8 trillion) in 2010
currencies, which are used as Brazils 2010 capital stock estimates in our calculations.
Table B.1.1
Method 1: Adding the gross fixed
capital formation
Year
2004
Capital
stock as
proportion
of GDP in
2000
currencies
3.17
GDP (in
USD
million of
2000)
716 960
Capital stock
(in USD million
of 2000)
Capital stock
(in USD million
of 2010)*
2 272 762
2005
2 383 981
111 219
2006
2 506 064
122 083
2007
2 645 059
138 995
2008
2 802 921
157 863
2009
2 944 527
141 606
2010
3 116 720
172 193
5 929 559
93
41
Appendix
Table B.1.2
Method 2: Capital stock extrapolation
Year
Capital
stock as
proportion
of GDP in
2000
currencies
Growth
rate of
proportion
Capital stock
(in USD
millions of
2000)
Capital stock
(in USD
millions of
2010)*
Capital
stock
decad
al real
growt
h rate
1950
2.28
1960
2.3
0.9%
105 343
1970
2.39
3.9%
192 726
460 615
1980
2.6
8.8%
430 379
1 118 985
1990
3.27
25.8%
501 772
1 640 794
142.9
%
46.6%
2000
3.17
-3.1%
644 702
2 043 705
24.6%
2010
estimate
3.40
7.3%
916 220
3 115 207
242 290
5 926 682
average
19702000
52.4%
71.4%
42
Appendix
Table B.2.1
Share in number of housings
approach
State/Region
Rondnia
Acre
Amazonas
Roraima
Par
Amap
Tocantins
North
Maranho
Piau
Cear
Rio Grande
do Norte
Paraba
Pernambuco
Alagoas
Sergipe
Bahia
Northeast
Minas
Gerais
Esprito
Santo
Rio de
Janeiro
So Paulo
Southeast
Paran
Santa
Catarina
Rio Grande
do Sul
South
Mato
Grosso do
Sul
Mato
Grosso
Gois
Distrito
Federal
CentralWest
Brazil total
Residential
capital stock
2010 (USD
million of
2010)
17 481
7 184
33 368
4 629
76 826
5 956
14 956
Number of
housings
2009 (in 1
000)
449
Number of
housings
2009 with
wastewater
collection
(weight: 2)
Number of
housings
2009 without
wastewater
collection
(weight: 1)
20
429
184
43
856
159
119
Weighted
number of
housings
2009
469
0.52%
141
228
0.25%
698
1 015
1.14%
15
104
134
0.15%
1 972
48
1 924
2 020
2.26%
153
151
154
0.17%
384
50
334
433
0.49%
1 701
195
1 506
1 896
2.12%
896
37
860
933
1.04%
2 393
732
1 661
3 125
3.50%
922
138
784
1 061
1.19%
1 079
424
656
1 503
1.68%
2 571
1 018
1 553
3 589
4.02%
874
70
804
944
1.06%
595
258
337
853
0.96%
4 300
1 845
2 454
6 145
6.88%
4.98%
160 401
66 260
34 924
93 217
Share
weighted
number of
housings
35 932
42 042
100 171
34 044
23 179
167 507
22.44%
597 276
6 221
4 871
1 351
11 092
12.41%
1 089
656
433
1 745
1.95%
5 382
3 915
1 468
9 297
10.41%
13 032
11 586
1 446
24 618
27.55%
3 432
1 936
1 495
5 368
6.01%
1 968
488
1 481
2 456
2.75%
3 658
663
2 995
4 322
4.84%
747
138
610
885
0.99%
954
106
848
1 059
1.19%
1 858
675
1 183
2 533
2.84%
786
684
103
1 470
1.65%
242 380
42 434
209 697
507 721
52.33%
1 002 231
133 697
76 686
142 526
13.59%
352 910
29 111
37 162
72 390
30 641
6.66%
169 304
2 282 121
58 577
30 771
27 806
89 349
43
Appendix
As the number of houses is only available at state level, this approach cannot be used to
break the residential capital stock down to the municipal level. To do so, the
municipalities GDP share could be used again.
Approach 3: Share in number of local units
Instead of using the GDP share, we can also use the states share in the number of local
business units (IBGE) to calculate the capital stock, especially for structures and
machinery and equipment.
Table B.2.2
Share in number of local units
approach
State/Region
Rondnia
Acre
Number of
local units
2008
Share local
units 2008
Capital stock
structures 2010 (in
2010 USD)
28 522
0.57%
14 898.4
5 037.6
8 868
0.18%
4 632.2
1 566.3
31 347
0.63%
16 374.0
5 536.5
6 718
0.13%
3 509.1
1 186.5
66 405
1.33%
34 686.5
11 728.5
7 224
0.15%
3 773.4
1 275.9
22 828
0.46%
11 924.1
4 031.9
171 912
3.45%
89 797.7
30 363.3
Maranho
60 528
1.22%
31 616.6
10 690.5
Piau
42 158
0.85%
22 021.1
7 446.0
141 416
2.84%
73 868.2
24 977.0
52 287
1.05%
27 312.0
9 235.0
Amazonas
Roraima
Par
Amap
Tocantins
North
Cear
Rio Grande do
Norte
Paraba
55 327
1.11%
28 899.9
9 771.9
123 903
2.49%
64 720.4
21 883.9
Alagoas
34 985
0.70%
18 274.3
6 179.1
Sergipe
27 984
0.56%
14 617.4
4 942.6
Bahia
240 990
4.84%
125 880.4
42 563.9
Northeast
779 578
15.66%
407 210.3
137 689.8
Minas Gerais
544 908
10.95%
284 631.1
96 242.2
Esprito Santo
97 153
1.95%
50 747.6
17 159.3
Rio de Janeiro
377 015
7.57%
196 932.7
66 588.7
Pernambuco
So Paulo
1 539 025
30.92%
803 905.2
271 824.1
Southeast
2 558 101
51.39%
1 336 216.6
451 814.2
Paran
376 287
7.56%
196 552.4
66 460.2
Santa Catarina
267 781
5.38%
139 874.6
47 295.7
Rio Grande do
Sul
460 071
9.24%
240 316.7
81 258.2
1 104 139
22.18%
576 743.8
195 014.1
57 620
1.16%
30 097.6
10 176.9
South
Mato Grosso
do Sul
Mato Grosso
Gois
Distrito
Federal
Central-West
Brazil total
76 639
1.54%
40 032.2
13 536.1
145 710
2.93%
76 111.2
25 735.4
84 067
1.69%
43 912.2
14 848.0
364 036
7.31%
4 977 766
190 153.1
64 296.4
2 600 121.5
879 177.8
Source: IBGE, Cadastro Central de Empresas 2008; Swiss Re's own calculations.
44
Appendix
Comparison of approaches
As the number of housings share and the number of local units share directly relate to
units of the capital stock, it seems to be more correct to use them to disaggregate the
residential capital stock and the capital stock of structures and machinery and
equipment to the state/municipal level.
Yet as a comparison of the three methods shows, the GDP share can be reasonably used
as a proxy for capital stock distribution in Brazil in many cases. The difference between
the GDP share and the local units share is moderate over all regions. The relative ranking
of the regions remains unchanged. Between the GDP share and the number of housings
share approach, the difference is also moderate for the North and Central-West. Only for
the Southeastern, Southern and Northeastern region, is there some discrepancy. Since,
in total, the difference between the shares which each approach assigns to a region is
moderate and the most complete data set is available for GDP shares (also of all
municipalities), we will use it as a proxy for the regional distribution of the capital stock
to calculate the regions annual expected losses. By doing so, the results of the latter
might become slightly overestimated for the Southeast and underestimated for the
Northeast.
Table B.2.3
Variance between shares
Region
GDP share,
in %
Local units
share, in %
Difference to
GDP share
4.9
3.50
Northeast
12.8
15.60
Southeast
55.8
South
17.4
9.1
North
Central-West
Difference
to GDP
share
1.40
Weighted
number of
housing
share, in %
4.98
-2.80
22.44
-9.64
51.30
4.50
52.33
3.47
22.20
-4.80
13.59
3.81
7.30
1.80
6.66
2.44
-0.08
45
Appendix
Literature
Afonso, L. & Sepulveda, M (2009). Comprehensive Homeowners Microinsurance in
Brazil: Estimation of Pricing and Market Potential. Latin American Business Review,
11, p. 199-211.
Agncia Nacional de Transportes Aquavirios (2010). Portos da Bahia - situao atual e
perspectivas. Seminario de comercio exterior. Salvador.
AON Benfield (2011). January 2011 Monthly Cat Recap Impact Forecasting.
Associated Press (2008). Brazil aims to repair ports destroyed by floods.
AXCO (2011). Insurance Market Report Brazil: Non-Life (P&C).
Barrionuevo, A. (2009). Little Relief Expected for Flood-Ravaged Brazil. New York Times.
Retrieved on 20 May 2011 from
http://www.nytimes.com/2009/05/18/world/americas/18brazil.html
BBC (2010). Brazilian Elections: Infrastructure. Retrieved 23 May 2011 from
http://www.bbc.co.uk/news/world-latin-america-11389613
Bester, H., Chamberlain, D., Hougaard, C. & Smit, H. (2010). Microinsurance in Brazil.
Towards a strategy for market development. Cape Town: The Centre for Financial
Regulation and Inclusion.
BNDES (2011). BNDES cria dois programas de socorro aos municpios atingidos pelas
enchentes no Rio. Press release. Retrieved 23 May 2011 from
http://www.bndes.gov.br/SiteBNDES/bndes/bndes_pt/Institucional/Sala_de_Impr
ensa/Noticias/2011/todas/20110121_perrj.html
Bradford, M. (2011). FERMA protests Brazil's new reinsurance rules. Retrieved 23 May
2011 from
http://www.businessinsurance.com/article/20110318/NEWS/110319905
Buani, C. (2011). Brazil floods and mudslides - A Brazilian perspective. Retrieved 23
May 2011 from http://reliefweb.int/node/382646
Contas Abertas (2011). Governo deixou de investir BRL 1,8 bi na preveno de
enchentes em sete anos. Retrieved 23 May 2011 from
http://contasabertas.uol.com.br/WebSite/Noticias/DetalheNoticias.aspx?Id=402
Do Amaral, R. and Rodrigues Ribeiro, R. (2009). Inundao e Enchentes. In L. Tominga
et al. Desastres Naturais: conhecer para prevenir. So Paulo: Instituto Geolgico.
Downie, A. (2011). Brazil's Murderous Floods: A Disaster Foretold. Retrieved 23 May
2011 from
http://www.time.com/time/world/article/0,8599,2042507,00.html#ixzz1E1Hdt
U1G
Economic Commission for Latin America and the Caribbean (ECLAC) (2003). Handbook
for Estimating the Socio-economic and Environmental Effects of Disasters.
Economics of Climate Adaptation Working Group (2009). Shaping climate-resilient
development: A framework for decision-making.
Economist Intelligence Unit (EIU) (2011). Country Report Brazil. March 2011.
Ellsworth, B. (2010). Brazilian infrastructure lags behind the boom. Reuters. Retrieved
23 May 2011 from http://www.reuters.com/article/2010/11/24/us-brazilsummit-infrastructure-idUSTRE6AN50I20101124?pageNumber=1
Floodsite (2007). Evaluating flood damages: guidance and recommendations on
principles and methods.
Guy Carpenter (2011). Microinsurance Catastrophe Risk Organisation (MiCRO) Created
to Help Protect Haitis Micro Entrepreneurs. Retrieved 7 June 2011 from
http://www.guycarp.com/portal/extranet/press/PDF/2011/MiCRO%20Launch%2
0110330.pdf;jsessionid=Nt1XKVPpzhxw1pKJ7n2dJf45LxTBMcDLvbnhhvLf3dGF
pmpcFWS7!-577718736?vid=1
Han, D., Davis, J., Hu Z., Lan G., Maren E.& Twyman C. (2002). Design Studies on FloodProof House. University of Bristol: Department of Civil Engineering.
46
Appendix
Hazell P., Anderson J., Balzer N., Hastrup Clemmensen A., Hess U. & Rispoli F. (2010).
The Potential for Scale and Sustainability in Weather Index Insurance. International
Fund for Agricultural Development and World Food Programme, Rome.
Instituto Brasileiro de Geografia e Estatstica (IBGE) (2008). Population Projection of
Brazil. Retrieved 24 May 2011 from
http://www.ibge.gov.br/english/presidencia/noticias/noticia_impresSo.php?id_no
ticia=1272
Intergovernmental Panel on Climate Change (IPCC) (2007). Climate Change 2007:
Synthesis Report Summary for Policymakers. Fourth Assessment Report.
Cambridge Univ. Press.
Machado, S. (2010).Microinsurance Concept Definition and Target Audience
Identification. Microinsurance in Brazil: SUSEP Research Series Volume 1.
Marengo, J. (2009). Climate change, extreme weather and climate events in Brazil. In
Climate Change and extreme events in Brazil. Lloyds and Brazilian Foundation for
Sustainable Development.
Marengo, J., Alves, L., Valverde, M, Porfirio da Rocha, R. & Ladorbe, R. (2007). Eventos
extremos em cenrios regionalizados de clima no Brasil e Amrica do Sul para o
Sculo XXI. Projees de clima futuro usando trs modelos regionais. Relatrio 5.
Mehlhorn, J. (2002). Technical memorandum on comparison between flood extents for
the Upper Thames basin modelled with IH-Report 130 method and Swiss Re
Geomorphologic Regression. For internal use only. Swiss Re.
Mendiondo, E. M. (2005). Flood risk management of urban waters in humid tropics:
early-warning, protection and rehabilitation.
Merz, B., Kreibich H., Schwarze R. and Thieken A. (2010). Assessment of economic
flood damage. Natural Hazards and System Sciences.
Ministerio das Relaes Exteriores (2010). Brazilian Strategies on Disaster Risk, Climate
Change and Humanitarian Assistance. Presentation given at the 19th Meeting of the
Special Committee on Disaster Risk Reduction, Association of the Caribbean States,
Santo Domingo.
Moran, A. (2010). Heavy rains, floods in Brazil displace 150,000 people. Retrieved 20
May 2011 from http://www.digitaljournal.com/article/293855#ixzz1Mtwn6nRq
Morandi, L. (2005). Estoque e Produtividade de Capital Fixo Brasil, 1940-2004.
Universidade Federal Fluminense.
Morandi da Silva, A., Lustosa, E., Bastos, E. & Gasparini, L. (2004). Financiamento de
Projetos de Infra-Estrutura pelos Fundos de Penso. Retrieved 23 May 2011 from
http://bibliotecadigital.fgv.br/ocs/index.php/ebf/4EBF/paper/viewFile/1513/632
Morris, H. (2010). Aim is to bring slums into the mainstream. Financial Times Special
Report. Brazil Infrastructure.
Mota, J., Ganzoni J. & Ges, G. (2010). Economia das Mudanas Climticas. Instituto de
Pesquisa Econmica Aplicada.
Oritago, J., & Sayao, A. (2004). Handbook of Slope Stabilization Engineering. Berlin:
Springer.
Reiff, L., Reis E. & Tafner P. (2001). Distribuio de riqueza imobiliria e de renda no
Brasil, 1992-1999. Srie Seminrios DIMAC, n. 75, IPEA-RJ.
Sandink, D., Kovacs, P., Oulahen, G., & McGillivray, G. (2010). Making Flood Insurable
for Canadian Homeowners: A Discussion Paper. Toronto: Institute for Catastrophic
Loss Reduction & Swiss Reinsurance Company Ltd.
Seifert, I., Thieken, A., Merz, M., Borst D. & Werner, U. (2009). Estimation of industrial
and commercial asset values for hazard risk assessment. Springer.
Serpone Bueno, F. & Sedeh, V. (2010). No excuses slum upgrading in So Paulo.
Retrieved on 23 May 2011 from http://globalurbanist.com/2010/03/24/noexcuses-slum-upgrading-in-s%C3%A3o-paulo.aspx
Sibaja, M. (2011). Brazil to create disaster-prevention, alert system. Associated Press.
Retrieved on 23 May 2011 from http://www.thestar.com/article/925736
Swiss Re (1998a). Floods an insurable risk? Zurich.
47
Appendix
48
49