Habitat International 45 (2015) 106e113

Contents lists available at ScienceDirect

Habitat International
journal homepage: www.elsevier.com/locate/habitatint

Economic impacts of natural disasters in megacities: The case of floods

~o Paulo, Brazil
in Sa
Eduardo Amaral Haddad a, b, c, *, Eliane Teixeira a, b, c
a
Department of Economics at the University of Sao Paulo, Brazil
b
NEREUS, The University of Sao Paulo Regional and Urban Economics Lab, Departamento de Economia, FEA-USP, Av. Prof. Luciano Gualberto, 908, FEA I,
Cidade Universita ria, Sa
~o Paulo, SP, 05508-900, Brazil
c
Institute of Economic Research Foundation, FIPE, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: The city of Sa~o Paulo, home to 11 million people, suffers constantly the effects of flooding caused by
Available online 26 July 2014 extreme precipitation. Localized floods occur every summer in various parts of the city. Besides the losses
and inconvenience felt by the residents, floods produce damages that cross the city boundaries, affecting
Keywords: income and output in the metropolitan area as well as in other parts of the state and the country. The
Climate change ~o Paulo through the
objective of this study is to evaluate the economic impacts of floods in the city of Sa
Natural disasters
use of a Spatial Computable General Equilibrium (SCGE) model integrated to GIS information related to
~o Paulo
Flooding in Sa
the location of points of floods and the firms within their influence. It is estimated that floods
Spatial computable general equilibrium
model
contributed to reduce city growth and residents' welfare, as well as hampering local competitiveness in
GIS both domestic and international markets. An intra-city total impact-damage ratio of 2.2 and an economy-
wide total impact-damage ratio of 5.0 were found.
© 2014 Elsevier Ltd. All rights reserved.

Introduction and intensity of short lasting extreme events, and point to an in-
crease in the number of days with heavy rainfall until the end of the
The relationship between urbanization of large agglomerations century, aggravating further the issue of flooding.2
and flooding has been receiving more attention recently in the The consequences of floods in urban areas are relevant, ranging
context of forecasted changes in the frequency and intensity of from impacts on human health (Huntingford, Hemming, Gash,
extreme events due to climate change (Gu, Hu, Zhang, Wang, & Gedney, & Nuttall, 2007) to effects on housing prices (Harrison,
Guo, 2011; Linnekamp, Koedam, & Baud, 2011; Nobre et al., 2011). Smersh, & Schwartz, 2001), urban transportation infrastructure
There is a significant correlation between human occupation (Suarez, Anderson, Mahal, & Lakshmanan, 2005), and other
and urban drainage systems.1 Urban development promotes damaging effects such as time lost in work and education, damages
changes in land use that increase flood hazards, as changes to to property and psychological stress (Linnekamp et al., 2011);
stream channels can limit drainage during heavy rainfall (Konrad & moreover financial costs are also important, through potential ef-
Booth, 2002). Megacities such as Sa ~o Paulo have numerous social fects in the insurance sector and public compensation schemes
and environmental problems associated with patterns of develop- (Botzen & Van den Bergh, 2008).
ment and transformation of space, which have been aggravated by Van der Veen and Logtmeijer (2005) discusses [economic]
increases in temperature and intensification of extreme weather vulnerability to flooding as a function of dependence e the degree
events (Dias, Dias, Carvalho, Freitas, Dias, 2013; Nobre et al., 2011). to which an activity relates to other economic activities in the rest
Projections of climate changes indicate a step up in the frequency of the country; redundancy e the ability of an activity (or system)

2
* Corresponding author. Department of Economics at the University of Sao Paulo, Even though projections of changes in rainfall and rainfall extremes may still
Brazil. Tel.: þ55 1138131444. show uncertainties due to climate model difficulties in representing some physical
E-mail addresses: [email protected] (E.A. Haddad), [email protected] processes related to rainfall formation at higher resolutions (Marengo, Jones, Alves,
(E. Teixeira). & Valverde, 2009), significant positive trends are found in the evolution of daily
1 ~o Paulo has always been linked to the
The worsening of drainage problems in Sa ~o Paulo from 1933 to 2010 historical data (Dias
rainfall extremes in the city of Sa
occupation of the Tiete River plain (Nobre et al., 2011; Ostrowsky, 1991). et al., 2013).

http://dx.doi.org/10.1016/j.habitatint.2014.06.023
0197-3975/© 2014 Elsevier Ltd. All rights reserved.
 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113 107

to respond to a disruption by overcoming dependence by deferring, Table 1

using substitutes or even relocating; and susceptibility e the Basic indicators.

probability and extent of flooding. Understanding this functional Area Population GDP Per capita HDI 2000
tripod would be essential to identify risk-related economic hot- (000 km2) (000 000) (USD billion) GDP (USD)
spots associated with a broader concept of damage which should ~o Paulo 1.5
City of Sa 11.3 194.6 17,221 0.841
include not only traditional direct damages related to infrastruc- SPMR 7.9 19.7 306.5 15,558 0.813
ture, property and business disruption, but also indirect costs Brazil 8,514.9 190.8 1,619.2 8,486 0.665

measured through inputeoutput linkages.

We operationalize this concept using an alternative methodo-
logical approach. The focus of the paper is on one specific dimen- million inhabitants), and responsible for 19% of Brazilian GDP. The
sion of the economic impacts of flooding in an integrated ~o Paulo is the core of the metropolitan area and accounts
city of Sa
interregional system. We use as our case study all the flooding for 5.9% of the country's population and 12% of its GDP (Table 1).
events that occurred in the city of Sa ~o Paulo in 2008. As we will From a stylized perspective, one can notice the existence of an
analyze every single point of flood in that year, treatment of sus- extended central business district (CBD) associated with the spatial
ceptibility becomes of lesser importance; from the economic configuration of economic activities in the metropolitan area. The
standpoint, once the flood took place in a point of the city, we will CBD concentrates the jobs, while households are located in the
just need to verify the existence of affected business firms within its surroundings of the center, with population density decay in the
territorial extent. The other two elements are also explicitly boundaries of the territory of the metropolis. As suggested by Figs. 1
considered. Dependence is fully captured through spatial linkages and 2, the internal organization of the SPMR can be approached by
associated with the income flows and value chains embedded in a Muth-Mills-Alonso urban model, having as the CBD the extended
the metropolitan inputeoutput system used to calibrate the spatial center of the city of Sao Paulo. According to the 2010 population
computable general equilibrium (SCGE) model, and on hypotheses census, the city of Sa ~o Paulo receives daily an inflow of almost 1
on interregional factor mobility; redundancy is contemplated on million commuters, representing 15.4% of workers in the city.3
the strength of substitution effects induced by price effects and Dense urbanization is a significant source of heat. The densest
supply constraints in the SCGE model. parts of the metropolitan area tend to be the warmest, and tem-
In an economic perspective, three main driving forces come into perature decreases as urban density declines (Nobre et al., 2011).
play: at first, businesses in the influence area of the flood points The “urban heat-islands” create more summer rain over and
may have to shut down temporarily, hampering potential GDP/GRP downwind of major cities (Shepherd, Pierce, & Negri, 2002), rein-
growth through interruptions in their value chains; secondly, forcing the forecasted trends of increasing frequency and intensity
damage in the economic infrastructure may generate either a ~o Paulo (Dias et al., 2013).
of daily rainfall extremes in the city of Sa
reduction in the capital stocks available for production or, more The rapid process of urban expansion around the CBD (Fig. 3)
common in flood events, temporary disruptions of infrastructure was not followed by the implementation of adequate infrastruc-
services (e.g. electricity disruption, closing of roadways), also ture, causing important urban problems. Due to the lack of urban
negatively impacting the potential GRP/GDP; and thirdly, recon- planning and the illegal occupation of the environmental protec-
struction efforts and additional maintenance measures operate in tion areas (around watercourses and on slopes), the percentage of
the opposite direction, activating investment-oriented activities impervious areas in the Upper Tiete ^ River Basin, the main water-
(e.g. construction sector), starting more vigorously after the flood course in the city, has increased significantly. As a result, flood
periods. This paper aims to evaluate the effects of the first of these frequency in the basin has also increased (Barros, Branda ~o, Silva,
three driving forces. We look at the economies of the city of S~ ao Ono, 2005).
Paulo and of Brazil in 2008 and estimate what would be the hy- Among the main issues currently faced by the city are those
pothetical economy-wide impact had all flood events not occurred related to climate change. The intense rainfall in the summer causes
in that year. In doing that, we are able to derive the estimates of the floods in several points of the town. Besides the loss and inconve-
economic costs of the floods related to the value chains disruptions nience suffered by people directly affected, these floods produce
associated with businesses' temporary close downs during the wider effects that surpass the limits of the city, reflecting on eco-
events. By not taking into account the effects of neither disruptions nomic sectors of the state and the country. The impacts of floods
in infrastructure services nor financial flows associated with re- affect households, industrial and commercial activities, public and
covery of the natural disasters, we are able to isolate the economic private services, and the urban transportation system. Given the
effects of the flooding and its spatial propagation providing an increasing concentration of people and economic activities in the
approximation of the regional consequences from a value chain area in the past decades, impacts and economic losses due to nat-
perspective. Thus, these estimates should be considered as lower ural catastrophes have increased substantially.
bounds of the broader economic costs. Since the early 1970s, the SPMR has undergone major
The remaining of the paper is structured as follows. In the next structural changes, transformed from dominance by traditional
section, we provide more information on the study area, followed manufacturing to more sophisticated services.4 The hollowing-out
by a section in which the methodology is described and estimates process of the metropolitan area associated with the geographical
of direct effects are shown. The following section presents the de-concentration of the manufacturing sector has imposed a new
economy-wide impacts of the 2008 floods in the city of Sa ~o Paulo, structure of interregional dependence in which SPMR plays a spe-
emphasizing their systemic nature. Final remarks conclude. cific role. On one hand, a typical establishment is now less
dependent both on sources of inputs and on markets within the
The study area metropolitan area. On the other hand, fragmentation is now a

~o Paulo Metropolitan Region (SPMR), the main economic

The Sa
and financial center of Brazil, consists of 39 municipalities in an 3
Around 170,000 of S~ao Paulo residents commute daily to other cities, especially
intense process of conurbation. It is the fourth largest urban in the SPMR.
agglomeration of the world, and the largest urban agglomeration in 4
The share of SPMR in the national manufacturing output dropped from over
the country, with about 10% of the national population (around 20 40% in the 1970s to less than 20% in the recent years.
108 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113

Fig. 1. Where people live e Population density in the SPMR.

Source: Companhia do Metropolitano de S~ao Paulo

Fig. 2. Where people work e Employment density in the SPMR.

Source: Companhia do Metropolitano de S~ao Paulo

characteristic of production with longer value chains based on the dependence of S~ ao Paulo city on primary and manufactured
organization of production to exploit economies of scale in indi- products from outside the metropolitan boundaries and a relative
vidual plants in specialized component production e and shipping self-sufficiency in services.
to other plants elsewhere to add further components (Hewings, It is in this context that the assessment of the impacts of floods
2012). Table 2 reveals the structure of interregional and interna- in the city of S~ ao Paulo is to be undertaken. Even though there is a
tional trade flows with special attention to the city of S~
ao Paulo and strong perception that, apart from the damage and inconvenience
the remaining metropolitan area. The Sa ~o Paulo city is directly suffered by those directly affected, flooding in the Upper Tiete^ Basin
involved in 14.1% of all trade flows of the country, including both in Sa ~o Paulo could produce wider effects stretching beyond
domestic and international trade partners. Intra-city trade corre- regional boundaries, with devastating effects on the state and
sponds to 36.6% of total trade flows of the city, while the remaining national economies (Nobre et al., 2011), this is, to our knowledge,
63.4% are distributed between interregional trade (17.0% within the the first attempt to address this issue in an integrated modeling
metropolitan area and 38.6% with other Brazilian regions) and in- approach.
ternational trade (7.8%). This important share of domestic trade
outside the metropolitan area reflects the aforementioned process The integrated modeling framework
of hollowing-out and spatial fragmentation that started during
the 1970s in Brazil, when the manufacturing sector relocate to The purpose of this study is to evaluate the wider economic
other regions outside the SPMR (Diniz, 1994). A closer look at the impacts of floods in the city of Sao Paulo through a bottom-up
structure of trade flows by sectoral products reveals the stronger modeling approach. In this context, the economic consequences
 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113 109

Fig. 3. Urban sprawl of the SPMR, 1881e1995.

Source: CESAD-FAU/USP e http://www.cesadweb.fau.usp.br/

of localized floods are assessed from the municipal to the national appropriate level of disaster assistance, to improve recovery de-
level. As highlighted by Rose (2004), the quantification of economic cisions, and to inform insurers of their potential liability.
losses is necessary to gauge individual and community vulnera- Natural disasters can cause physical destruction of installed
bility, to evaluate the worthiness of mitigation, to determine the capacity. However, the city of Sao Paulo has been adapting to the
type of flooding that it experiences. Despite the inconvenience
caused, such as the interruption of the flow of people and goods
due to the lack of urban drainage, the loss of machinery and
Table 2
Structure of trade flows in Brazil, 2008 (in % of total). equipment are negligible in such events. In our approach, the
economic losses of the main economic and financial center of the
Destination
country are directly related to the shutdown of production in the
SPC SPMR RSP RB ROW affected sites. While the direct losses are assessed through the
Primary (8.5% of total flows) disruption of economic activities, indirect losses can be calculated
Origin S~ ao Paulo City (SPC) 0.00 0.01 0.00 0.00 0.00 by considering associated interruptions in the value chains. Such
Rest of SPMR (SPMR) 0.01 0.02 0.01 0.00 0.00 estimation of indirect losses demands an economic model capable
Rest of S~
ao Paulo State (RSP) 0.54 0.35 3.60 1.88 0.26
of capturing the flow of goods between economic agents, which
Rest of Brazil (RB) 1.93 1.58 13.25 47.67 16.88
Rest of the world (ROW) 0.28 0.35 3.46 7.93 0.00 poses another challenge, given the uniqueness of each disaster
Manufacturing (36.6% of total flows) (Okuyama, 2007).
Origin S~ ao Paulo City (SPC) 1.81 1.03 0.82 1.46 0.55
Rest of SPMR (SPMR) 1.77 2.14 1.00 1.65 0.89
Rest of S~
ao Paulo State (RSP) 2.31 1.52 8.10 7.06 3.32 Estimating the direct impact of the 2008 flooding events in the city
Rest of Brazil (RB) 1.00 0.64 3.14 39.87 6.50 ~o Paulo
of Sa
Rest of the world (ROW) 1.28 1.06 2.36 8.74 0.00
Services (54.9% of total flows) Using detailed GIS information on flooding points and on labor
Origin S~ ao Paulo City (SPC) 8.22 1.82 2.01 3.32 0.46
Rest of SPMR (SPMR) 0.68 3.02 0.49 0.81 0.18
profiles of firms, and assuming that production is a continuous on
Rest of S~
ao Paulo State (RSP) 0.28 0.22 9.04 1.89 0.40 weekdays, foregone losses can be indirectly approached by a
Rest of Brazil (RB) 0.20 0.18 1.34 61.13 1.98 measure of foregone wages paid during the interruption of pro-
Rest of the world (ROW) 0.27 0.14 0.33 1.59 0.00 duction in businesses affected by flooding events.
Total (100.0% of total flows) ~o Paulo is provided by the
The information on flooding in Sa
Origin S~ ao Paulo City (SPC) 5.17 1.38 1.41 2.36 0.45
Rest of SPMR (SPMR) 1.02 2.44 0.64 1.05 0.42 Emergency Management Center (EMC) of Sa ~o Paulo local govern-
Rest of S~
ao Paulo State (RSP) 1.04 0.71 8.23 3.78 1.46 ment. The EMC database informs, for each flooding point, its exact
Rest of Brazil (RB) 0.64 0.46 3.01 52.20 4.90 location, date of occurrence, starting time and duration, as well as a
Rest of the world (ROW) 0.64 0.49 1.34 4.75 0.00 qualitative assessment of the intensity of the event. With the geo-
Source: SCGE database. referencing of the flooding database for 2008, and integrating these
110 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113

Fig. 4. Example of GIS-based influence area of flood points, for different scenarios (50 m, 100 m, 150 m, 200 m).

data with firms' location through GIS techniques, it is possible to that the extension of the foregone output will depend on the time of
locate all businesses impacted by floods in a pre-defined impact the day the flood event takes place: for events that take place before
radius.5 noon (after 6am), it will be assumed that the whole daily output is
The information on payments to workers in the affected firms is “lost”, while only half of it will be considered to be “lost” in cases the
essential to capture the direct losses caused by flooding. These data flood occurs in the afternoon, during working hours. Thus, once the
are obtained from a GIS database on labor firms, known in Brazil as companies directly affected by the flood events are identified under
RAIS (Annual Report of Social Information).6 The RAIS database our working hypotheses, one can estimate foregone losses in the
informs, for each single firm, its address, total wages paid to periods of interrupted production. Such information is used to
workers, and the SIC code of its main activity. This statistical survey prepare the shock vector that feeds the SCGE model so that one can
covers 97% of the formal labor market, which represents 57% of the evaluate the total economic impacts of floods in S~ ao Paulo.
total employment in the country,7 providing a measure of the Estimated direct foregone labor income due to floods in 2008 in
representativeness of the sample. A simple reweighting procedure ~o Paulo is presented in Table 3. Direct impact is heavily
the city of Sa
is used based on the information on sectoral labor payments in the concentrated in tertiary activities (around 90%), especially services,
CGE database, so that the aggregated micro data on the workers' commerce and transportation. In the more plausible scenario, i.e.
wage bill in a given sector in the city of Sa ~o Paulo would be 100 m, almost 20,000 firms would have been directed affected by
consistent with the macro data. flood events in 2008, with a labor income loss of over BRL 17
The combination of these two databases allows us to identify million. In terms of foregone output, this is close to BRL 80 million
those businesses potentially affected by a flood event (Fig. 4). To (BRL 43.5 million in terms of value added).
calibrate the area of influence of a flood point in our simulations, 25
interviews were undertaken with firms locate in the vicinity of two Estimating the impacts at the regional and national levels
specific flood points in our database.8 The main conclusions of the
field research were that (i) up to 100 m from the flood point ~o
In order to estimate the total impacts caused by floods in Sa
(approximately one block), firms are definitely affected e depend- Paulo we use a spatial computable general equilibrium model
ing on the time of the event, workers either do not arrive at work or (SCGE). It is a model based on simultaneous optimization of the
leave the job place earlier; (ii) firms further away (e.g. 200 m) are
only affected in the most severe events; and (iii) in the rare occa-
Table 3
sions when the water invades the establishment, precautionary
Estimated foregone (reweighted) labor income (in BRL thousand).
adaptive measures are promptly undertaken for the next event.
Thus, the most likely scenario would be to define an impact radius Impact radius
of 100 m and not to consider damage to capital stocks. Moreover, to 50 m 100 m 150 m 200 m
calibrate our simulations, we have an additional ad hoc assumption Primary 7.08 9.92 22.39 33.13
Manufacturing 454.01 2,014.94 3,131.36 5,557.37
Utilities 673.48 1,444.4 1,795.81 2,146.47
5
We excluded from the analysis flood events that took place in the middle of the Construction 185.72 589.96 1,034.77 1,778.95
night. Since the vast majority of flood-related problems last fewer than several Commerce 1,057.58 2,046.03 3,702.56 6,024.32
hours, as businesses are probably closed anyway and there is no disruption for the Transportation 663.62 948.01 1,483.51 1,933.59
most part (Rose & Guha, 2004). Services 3,132.72 9,034.60 16,445.59 31,519.12
6 Public administration 300.43 1,354.86 15,233.56 16,317.47
Ministry of Labor and Employment.
7 TOTAL 6,474.63 17,442.72 42,849.55 65,310.42
IBGE (Brazilian Institute of Geography and Statistics).
8 # of affected firms 8,577 19,784 35,896 57,938
Lapa and CEAGESP areas.
 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113 111

Fig. 5. Input-output relations embedded in the spatial CGE model.

behavior of individual consumers and firms, subject to resource municipalities that compose the metropolitan area, the rest of the
constraints. When extended into a multi-regional framework, the ~o Paulo and the rest of Brazil. Furthermore, it maps the
state of Sa
model is able to provide the spatial distribution of impacts. inter-industrial relations by place of production; the payments to
The specification of the SCGE model uses as its departure point labor factor by place of residence; and the consumption structure
the B-MARIA model, developed by Haddad (1999). The B-MARIA by place of consumption (Fig. 5).10
model e and its extensions e has been widely used for assessing
regional impacts of economic policies in Brazil. Since the publica- ~o Paulo city
The economy-wide impacts of the 2008 floods in Sa
tion of the reference text, various studies have been undertaken
using, as their basic analytical tool, variations of the original model. What if floods had not occurred in the city of Sa~o Paulo in 2008?
The theoretical structure of the B-MARIA model is well docu- What would be the difference in terms of value added (GRP/GDP)
mented. Results are based on a bottom-up approach e i.e. national for the city and for other regions of the country? Results of the SCGE
results are obtained from the aggregation of regional results. The simulations for the four scenarios were computed via a 1e2e4
model identifies different production/investment sectors in each Euler procedure with extrapolation, under a short-run closure
region producing different commodities, one representative (exogenous capital stocks). In what follows, we focus our analysis
household in each region, regional governments and one Federal on the 100 m-scenario. Results for the remaining simulations are
government, and a single foreign area that trades with each do- qualitative similar.
mestic region, through a network of ports of exit and ports of entry. Table 4 shows the results for the macroeconomic effects of the
Three local primary factors are used in the production process, 2008 floods generated by the SCGE simulation. It can be seen that it
according to regional endowments (land, capital and labor). is expected floods to lower the Sa ~o Paulo city GRP by 0.0263% and
The B-MARIA framework includes explicitly some important the national GDP by 0.0071%. Despite the localized occurrence of
elements from an interregional system, needed to better under- floods within the city limits, they reduce output growth beyond its
stand macro spatial phenomena, namely: interregional flows of territory. They also contribute to a decline in welfare of city resi-
goods and services, transportation costs based on origin- dents (lower real household consumption), a reduction in tax
destination pairs, interregional movement of primary factors, revenue,11 and a decrease in the city's domestic and international
regionalization of the transactions of the public sector, and regional competitiveness, as verified by the worsening of the interregional
labor markets segmentation. We have also introduced the possi- and international balances of trade. The national effects go in the
bility of (external) non-constant returns in the production process, same direction, with lower welfare and government expenditures,
following Haddad and Hewings (2005). This extension is essential and a decrease in the country's competitiveness in international
to adequately represent one of the functioning mechanisms of a markets.
spatial economy. The model used here is structurally calibrated for From a spatial perspective, Fig. 6 presents the impacts on mu-
2008, the same year for which data on floods are available. The nicipalities in the SPMR. Noteworthy is that the economic effects
calibration of the SCGE model is based on a fully specified inter- are not only local e they spread across the space through produc-
regional inputeoutput system considering 41 regions, 56 sectors tion and income linkages. We see that the effects of the floods are
and 110 products.9 It is a database at the municipal level for the year estimated to vary considerably across municipalities in the SPMR.
2008 which focuses on the SPMR, that is, it covers the 39

10
These different spatial dimensions follow the tradition found in metropolitan
inputeoutput models (Hewings et al., 2001; Jun, 2004).
9 11
Due to computational constraints, we have used an aggregated version with 8 Our assumption regarding adjustment in the real government expenditure
sectors and 8 products. considers constant real budget deficits for both regional and federal governments.
112 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113

Table 4 Table 5
Macroeconomic impacts of floods, 2008 (in percentage change). Direct and total GRP/GDP impact (in BRL million).

Impact radius Impact radius

50 m 100 m 150 m 200 m 50 m 100 m 150 m 200 m

~o Paulo
City of Sa Direct loss
Real GRP 0.0112 0.0263 0.0418 0.0692 S~
ao Paulo City (SPC) 16.63 43.54 88.3 143.53
Real household consumption 0.0028 0.0059 0.0008 0.0067 Total loss
Real government 0.0033 0.0094 0.0304 0.0446 S~
ao Paulo City (SPC) 40.08 94.38 150.34 248.55
consumption - Regional Rest of SPMR (SPMR) 3.35 9.05 12.72 19.66
Regional government 0.0024 0.0062 0.0120 0.0204 Rest of Sa~o Paulo State (RSP) 9.14 21.04 32.49 49.86
consumption - Federal Rest of Brazil (RB) 39.95 93.72 154.91 246.10
Real investment e e e e Brazil 92.52 218.19 350.46 564.17
Interregional export volume 0.0090 0.0223 0.0379 0.0628 Intra-city total impact-damage ratio 2.4 2.2 1.7 1.7
Interregional import volume 0.0005 0.0006 0.0023 0.0008 Economy-wide total impact-damage ratio 5.6 5.0 4.0 3.9
International export volume 0.0264 0.0578 0.0932 0.1493
International import volume 0.0051 0.0109 0.0177 0.0257

Brazil Final remarks

Real GDP 0.0030 0.0071 0.0114 0.0183
Real household consumption 0.0013 0.0033 0.0054 0.0097 Throughout most of the 20th century, the number of days when
Real investment e e e e rainfall exceeded 80 mm/day in Sa ~o Paulo was very small (one per
Real government 0.0022 0.0060 0.0141 0.0229
consumption - Regional
decade since the 1930s). After the early 1970s, however, the num-
Regional government 0.0024 0.0062 0.0120 0.0204 ber of cases increased, reaching nine cases in the first decade of this
consumption - Federal century (Dias et al., 2013). Using a SCGE model, we have ascribed
International export volume 0.0087 0.0188 0.0271 0.0405 output shocks based on GIS simulations in which locations of flood
International import volume 0.0032 0.0067 0.0088 0.0123
points and firms were combined, in projecting the economic im-
~o Paulo in 2008.
pacts of flood events that took place in the city of Sa
It was estimated that floods contributed to reduce city growth and
As expected, the largest negative effects of the floods e besides that residents welfare, as well as to hamper local competitiveness in
on Sa ~o Paulo City e are projected to occur in municipalities with both domestic and international markets. Not only was the city
stronger production and income coefficients linked to the capital economy affected, but also other regional economies in Brazil.
city. The key message is that one needs to consider interactions both
In money values, the total value added impact on the Brazilian inside and outside an urban system to assess the consequences of
economy is estimated to be BRL 218.19 million (100 m-scenario), for apparently local phenomena. Measures relating to the planning and
a direct damage of BRL million 43.54, so that the associated total control of land use should be executed in parallel with engineering
impact-damage ratio is 5.0. Considering only the intra-city impact, projects to improve the urban drainage system and prevent the
the impact-damage ration is equivalent to 2.2 (Table 5). Given the emergence of new risk areas. However, the recognition that the
existing spatial fragmentation observed in Brazil, and the structure economic effects of flooding are not only local poses an additional
of spatial dependence observed in the data (Table 2), the hierarchy governance problem in the context of Brazilian federalism. As
of impacts shows the rest of Brazil as the second most affected policy decisions (and budget controls) are made either at the mu-
region, with potential GRP losses similar in magnitude to those nicipality, state or federal level, coordination problems will be
projected for S~ ao Paulo City. The rest of the State of Sa ~o Paulo, faced. As Biesbroek, Swart, Van der Knaap, (2009) pointed out,
benefitted by the recent process of hollowing-out of the SPMR, there is no unequivocal solution to the challenge of climate change,
presents potential GRP losses greater than the totality of the other which will require a combination of long-term efforts and short-
municipalities in the SPMR. Roughly, the GDP impact in all sce- term actions taken at all governmental levels. In the context of
~o Paulo City and the rest
narios is spatially distributed as follows: Sa the problems of floods in Sa ~o Paulo, a further challenge emerges as,
of the country with similar shares in total losses of around 44%; the in addition to the traditional administrative structures that spatial
rest of Sa~o Paulo State with 9%, and the remaining 4% accruing to planners are accustomed to e which favor the treatment of socio-
the rest of SPMR. economic issues e additional spatial layers come into scene (e.g.

Fig. 6. Potential GRP losses in the RMSP municipalities, 100 m scenario.

 E.A. Haddad, E. Teixeira / Habitat International 45 (2015) 106e113 113

river basins, climate) bringing other dimensions to be integrated in Hewings, G. J. D., Okuyama, Y., & Sonis, M. (2001). Economic interdependence
within the Chicago metropolitan region: a Miyazawa analysis. Journal of
the search for a solution.
Regional Science, 41(2), 195e217.
Huntingford, C., Hemming, D., Gash, J. H. C., Gedney, N., & Nuttall, P. A. (2007).
Acknowledgments Impact of climate change on health: what is required of climate modellers?
Transactions of the Royal Society of Tropical Medicine and Hygiene, 101, 97e103.
Jun, M.-J. (2004). A metropolitan input-output model: multisectoral and multi-
We are grateful to Silvio Ichihara and Renato Vieira who pro- spatial relations of production, income formation, and consumption. Annals of
vided excellent research support. The authors acknowledge finan- Regional Science, 38, 131e147.
cial support provided by Rede CLIMA and INCT-MC. Eduardo A. Konrad, C. P., & Booth, D. B. (2002). Hydrologic trends associated with urban devel-
opment for selected streams in the Puget Sound Basin, Western Washington. U.S.
Haddad also acknowledges financial support from Conselho Geological Survey, Water-Resources Investigations Report 02-4040 Accessed on
Nacional de Desenvolvimento Científico e Tecnolo gico (CNPq) and February 15, 2013 http://pubs.usgs.gov/wri/wri024040/.
Fapesp; he thanks Princeton University and Rutgers University for Linnekamp, F., Koedam, A., & Baud, I. S. A. (2011). Household vulnerability to climate
change: examining perceptions of households of flood risks in Georgetown and
their hospitality. Paramaribo. Habitat International, 35, 447e456.
Marengo, J. A., Jones, R., Alves, L. M., & Valverde, M. C. (2009). Future change of
References temperature and precipitation extremes in South America as derived from the
PRECIS regional climate modeling system. International Journal of Climatology,
~o, J. L. B., Silva, O. F., & Ono, S. (2005). The impact of urban 29, 2241e2255.
Barros, M. T. L., Branda
Nobre, C., Young, A., Saldiva, P., Orsini, J., Nobre, A., Ogura, A., et al. (2011).
sprawl on flood risk areas. In Glenn E. Moglen (Ed.), Managing watersheds for
Vulnerability of Brazilian megacities to climate change: the S~ ao Paulo Metro-
human and natural impacts: Engineering, ecological, and economic challenges.
politan Region (RMSP). In Ronaldo S. Motta, Jorge Hargrave,
Williamsburg, VA: American Society of Civil engineers.
Gustavo Luedemann, & Maria B. S. Gutierrez (Eds.), Climate change in Brazil:
Biesbroek, G. R., Swart, R. j., & Van der Knaap, W. G. M. (2009). The mitigation-
Economic, social and regulatory aspects. Brasília: IPEA.
adaptation dichotomy and the role of spatial planning. Habitat International,
Okuyama, Y. (2007). Economic modeling for disaster impact analysis: past, present,
33, 230e237.
and future. Economic Systems Research, 19(2), 115e124.
Botzen, W. J. W., & Van den Bergh, J. C. J. M. (2008). Insurance against climate
Ostrowsky, M.de S. B. (1991). Urbanizaça ~o e controle de enchentes: o caso de Sa
~o
change and flooding in the Netherlands: present, future, and comparison with
Paulo (Urbanization and flood control: the case of Sa ~o Paulo). Master Thesis.
other countries. Risk Analysis, 28, 413e426.
Escola Polite cnica, Universidade de S~ ao Paulo.
Dias, M. A. F. S., Dias, J., Carvalho, L. M. V., Freitas, E. D., & Dias, P. L. S. (2013).
Rose, A. (2004). Economic principles, issues, and research priorities in hazard loss
Changes in extreme daily rainfall for S~ ao Paulo, Brazil. Climatic Change, 116,
estimation. In Yasuhide Okuyama, & Stephanie E. Chang (Eds.), Modeling spatial
705e722.
and economic impacts of disasters. New York: Springer.
Diniz, C. C. (1994). Polygonized development in Brazil: neither decentralization no
Rose, A., & Guha, G. S. (2004). Computable general equilibrium modeling of electric
continued polarization. International Journal of Urban and Regional Research,
utility lifeline losses from earthquakes. In Yasuhide Okuyama, & Stephanie
18(2), 293e314.
E. Chang (Eds.), Modeling spatial and economic impacts of disasters. New York:
Gu, C., Hu, L., Zhang, X., Wang, X., & Guo, J. (2011). Climate change and urbanization
Springer.
in the Yangtze River Delta. Habitat International, 35, 544e552.
Shepherd, J. M., Pierce, H., & Negri, A. J. (2002). Rainfall modification by major urban
Haddad, E. A. (1999). Regional inequality and structural changes: Lessons from the
areas: observations from spaceborne rain radar on the TRMM satellite. Journal
Brazilian economy. Aldershot: Ashgate.
of Applied Meteorology, 41, 689e701.
Haddad, E. A., & Hewings, G. J. D. (2005). Market imperfections in a spatial econ-
Suarez, P., Anderson, W., Mahal, V., & Lakshmanan, T. R. (2005). Impacts of flooding
omy: some experimental results. The Quarterly Review of Economics and Finance,
and climate change on urban transportation: a systemwide performance
45(2), 476e496.
assessment of the Boston Metro Area. Transportation Research Part D, 10,
Harrison, D. M., Smersh, G. T., & Schwartz, A. L., Jr. (2001). Environmental de-
231e244.
terminants of housing prices: the impact of flood zone status. The Journal of Real
Van der Veen, A., & Logtmeijer, C. (2005). Economic hotspots: visualizing vulner-
Estate Research, 21(1/2), 3e20.
ability to flooding. Natural Hazards, 36, 65e80.
Hewings, G. J. D. (October 1e2, 2012). Urban dynamics: some critical modeling
issues. In Paper presented at the conference on “Future challenges of the new
urban world: What model of development for the Moroccan city?”. Rabat: IRES.