Assessing the Vulnerability of Coastal Communities to Extreme Storms: The Case of Revere, Massachusetts, US

Full text

ASSESSING THE VULNERABILITY OF COASTAL COMMUNITIES TO
EXTREME STORMS: THE CASE OF REVERE, MA., USA
GEORGE E. CLARK, SUSANNE C. MOSER1, SAMUEL J. RATICK, KIRSTIN
DOW2, WILLIAM B. MEYER, SRINIVAS EMANI, WEIGEN JIN3, JEANNE X.
KASPERSON4, ROGER E. KASPERSON and HARRY E. SCHWARZ5
Graduate School of Geography and George Perkins Marsh Institute, Clark University, except for:
1Belfer Ctr. for Science & Internatl. Affairs, Kennedy School, Harvard University
2Geography Department, University of South Carolina
3Data Inc., Englewood Cliffs, NJ
4George Perkins Marsh Institute, Clark University
5International Development Program, Clark University
(Received 11 July 1997; in revised form 3 March 1998; accepted 16 March 1998)
Abstract. Climate change may affect the frequency, intensity, and geographic distribution of severe
coastal storms. Concurrent sea-level rise would raise the baseline of flooding during such events.
Meanwhile, social vulnerability factors such as poverty and disability hinder the ability to cope with
storms and storm damage. While physical changes are likely to remain scientifically uncertain into
the foreseeable future, the ability to mitigate potential impacts from coastal flooding may be fostered
by better understanding the interplay of social and physical factors that produce human vulnerability.
This study does so by integrating the classic causal model of hazards with social, environmental, and
spatial dynamics that lead to the differential ability of people to cope with hazards. It uses Census
data, factor analysis, data envelopment analysis, and floodplain maps to understand the compound
social and physical vulnerability of coastal residents in the city of Revere, MA, USA.
Background
The impacts of hazardous events are usually unevenly distributed among and within
nations, regions, communities, and groups of individuals. Vulnerable groups are
those who are likely to suffer a disproportionate share of the effects of hazardous
events. For the purposes of this study, we draw on a growing literature of vulner-
ability studies (reviewed by Dow, 1992, 1993) to define ‘vulnerability’ to hazards
as people’s differential incapacity to deal with hazards, based on the position of
groups and individuals within both the physical and social worlds. Building on
this literature, we see vulnerability as a function of two attributes: 1) exposure
(the risk of experiencing a hazardous event); and 2) coping ability, subdivided into
resistance (the ability to absorb impacts and continue functioning), and resilience
(the ability to recover from losses after an impact). In this paper, we use coping
ability as an antonym of social vulnerability.
Exposure (the risk of experiencing a hazardous event) as a concept is well-
explicated in the ‘technical’ risk literature (see for example Renn, 1992, pp. 58–
Mitigation and Adaptation Strategies for Global Change 3: 59–82, 1998.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

60 GEORGE E. CLARK ET AL.
TABLE I.
Sources of vulnerability themes
Age Bolin, 1982
Bolin and Klenow, 1983
Drabek and Key, 1984
Quarantelli, 1991
Rossi et al., 1983
Disabilities Parr, 1987
Family structure and social networks Bolin and Bolton, 1986
Drabek and Key, 1991
Housing and the built environment Bolin and Bolton, 1986
Bolin and Stanford, 1991
Godschalk, Brower and Beatley, 1989
NRC, 1984
White and Haas, 1975
Income and material resources Bolin and Bolton, 1986
Bolin and Stanford, 1991
Drabek and Key, 1984
Perry and Lindell, 1991
Quarantelli, 1991
Rossi et al., 1983
Lifelines NRC, 1984; Platt, 1991
Occupation Bolin, 1982
Race and ethnicity Bolin and Bolton, 1986
Drabek and Key, 1991
Perry, Green and Mushkatel, 1983
Perry and Lindell, 1991
Rossi et al., 1983
Trainer and Bolin, 1976
61) and is in general the most widely cross-disciplinarily understood dimension of
vulnerability. For this reason, we do not comment further on exposure except to
say that it, too, is partly socially constructed, in that existing land use and daily
commuting patterns, to name but two exposure variables, are social and temporal
phenomena. We use a simplified measure of exposure – a floodplain map – in
order to focus on the more explicitly social variables that together determine coping
ability. Coping ability, the ability to either absorb impacts (e.g., by exiting the scene
or dealing with the hazard in-place) orto recoverfrom them (e.g., though insurance,
cash reserves, or other means), is influenced by a large list of variables identified by

VULNERABILITY OF COASTAL COMMUNITIES 61
sociologists, geographers, political scientists, and other investigators. See Table I
and below under ‘Methods’.
This paper does not pretend to be a review of the vulnerability literature or
a catalog of all the past and future definitions and approaches to vulnerability.
However, There are several authors who should be mentioned for those who wish
to explore the terminology in depth. Timmerman (1981) provided an early look at
the concept and origins of the term ‘vulnerability’ and sparked a need to explore
and clarify the term as well as our understanding of the phenomenon. Liverman
(1990) focused the debate by providing a catalog of the ‘conditions and variables
...important in determining vulnerability to global environmental change.’ Wisner
(1992) looked at vulnerability in the developing world, examining the concept of
marginality and the role of the state in vulnerability. Blaikie et al. (1994) also fo-
cused on the developing world, analyzing the forces that combine to cause a crisis.
Chen (1994) examined vulnerability on a relatively large scale in light of global
environmental change. Dow and Downing (1995) commented on the state of vul-
nerability research in the wake of the Social Science Research Council-sponsored
First Open Meeting of the Human Dimensions of Global Environmental Change
Community, at Duke University in June, 1995. Cutter (1996) broadly reviews the
definitions of vulnerability and the types of studies carried out since 1980. Hewitt
(1997) reviews the vulnerability literature with an emphasis on vulnerability as a
form of powerlessness that is socially reproduced.
Exposure and coping ability as co-determinants of people’s vulnerability to
hazards are of particular interest to hazard managers inclined to address behav-
ioral, managerial, institutional and other human activity-related issues that change
the likelihood of severe impacts from hazards. Short of affecting the environ-
mental hazard itself, hazard managers seek points of intervention in the causal
chain between an hazardous event and the downstream human consequences. In
the hazards literature of the past two decades, this search has been supported
through causal modeling (inspired by the classic work of Hohenemser, Kasperson,
and Kates, 1985), which has since then also found entry into the global change
literature (e.g., Clark, Jager, and van Eijndhoven, forth-coming; Moser, 1997).
Hohenemser, Kasperson, and Kates (1985) illustrate that this view of hazards as
processes ‘strongly implies three possible strategies for hazard control: (1) pre-
vention of hazard events; (2) prevention of hazard consequences once events have
taken place; and (3) mitigation of consequences once these have occurred’.
The parallels to the search for possible responses to global climate change are
obvious in spite of the varigated terminology: (1) prevention of further increases
in the atmospheric concentration of greenhouse gases, (2) mitigation of impacts
(bounding of first order environmental changes), and (3) adaptation to the impacts
(higher order responses to impacts) (Bruce, Lee, and Haites, 1996; Houghton et al.,
1996; Watson et al., 1996). Understanding the causal linkages among the social and
physical processes that interact to produce a hazard on the one hand, and illuminat-
ing the causal linkages across scales from the global to the local on the other, are

62 GEORGE E. CLARK ET AL.
parallel challenges that need to be addressed in order to adequately prepare for the
management of climate change-related environmental hazards (e.g., Cash, 1997;
Global Environmental Assessment Project, 1997; Kotlyakov et al., 1988; Turner et
al., 1990).
In this paper we focus on the former challenge largely because the likely mag-
nitude of local consequences from global climatic changes are far from predictable
and the causal mechanisms that would link global processes to local impacts are
still inadequately understood. In the absence of such greater ability to predict local
consequences of global climate change, it is still possible and, as a no-regrets strat-
egy, advisable to address human vulnerability to environmental perturbations. We
thus adapt the hazards causal model here by introducing vulnerability via its two
dimensions, exposure and coping ability.
Some attempts have already been made to link vulnerability and the causal
structure of hazards. Watts and Bohle (1993) use a tripartate model of property rela-
tions, political power, and economic power. While this model is useful in exploring
the political economic dimensions of the problem, it addresses neither the spatial
aspects nor the physical environment. Cutter (1996) posits a synthetic ‘hazards
of place’ model which recognizes both the physical and the social in relation to
the causal processes of hazards. (Although Cutter cites Hewitt and Burton (1971)
as the inspiration for her term ‘hazards of place’, it should be noted that Hewitt
and Burton use the term ‘hazardousness of a place’ differently, not to introduce
‘vulnerability’ or to emphasize causal modelling, but simply to mean the inclusion
of multiple hazard types in the analysis of hazards ata given location.) While Cutter
integrates the physical and social into a model of hazard processes, our incorpora-
tion of vulnerability into the hazards causal model of Hohenemser, Kasperson, and
Kates (1985) accomplishes this integrative step while at the same time emphasizing
the spatial distribution of vulnerability and, through the notion of hazard as process,
allowing the identification of points of action in order to combat the effects of the
hazard.
The crux of vulnerability to global environmental change is as follows: people
stand to experience impacts from hazards of global change of varying degrees
that fall along a spectrum from positive to negative, based on their position in the
social and physical worlds. The problem, then, is twofold. (1) How do we begin
to analyze and understand the differential potential for harm in a local context?
(2) How do we incorporate vulnerability into our understanding of hazards as a
type of interplay between social and physical phenomena, in order to have a way
of discussing vulnerability’s implications and of adjudicating between different
policy and management options?
In order to answer the first question, we need to explore the social and spatial
distribution of vulnerability to global change hazards. In order to answer the sec-
ond question, we examine the implications of this research for causal modelling
of hazards and for storm preparation and response. We do so in the context of
a small U.S. coastal city, Revere, MA (see below) as an example of the type of

VULNERABILITY OF COASTAL COMMUNITIES 63
Figure 1. An MBTA (metropolitan Boston) bus driver works to keep his windshield clear during the
January, 1996 ‘northeaster’ in Revere, MA. Access to private automobiles and dependence on public
transportation are among the variables that determine the ability to cope with severe storms.
coastal communities and the types of management challenges faced by coastal
zone and hazard managers in developed countries. Clearly, we make no claim for
general representativeness, especially not in an international comparison. We do
believe, however, that Revere is far from singular with regard to the problems of
on-the-ground storm hazard and sea-level rise management faced along developed
shorelines. Comparative studies of coastal communities in other countries are in-

64 GEORGE E. CLARK ET AL.
vited. Our analysis offers one approach to address vulnerability not at the large
scale at which economists and other workers often assess problems, but at the
scale at which hazards managers commonly need to implement hazard reduction
strategies.
Setting
Revere, Massachusetts (1990 population 42 786), located just north of Boston, is
exposed to flooding and wave damage on three sides, from the Atlantic Ocean to
the east and from tidal rivers to the north and south. We chose this city for our study
because of (1) the availability of extensive detailed economic and topographic
baseline data (Hunt, 1990); (2) the large portion of floodprone residential area
in the city; and (3) the range of economic circumstances, from working-class to
affluent (despite Revere’s local reputation as a blighted community [Vigue, 1997]),
in the floodplain and community as a whole.
The threat of accelerated global sea-level rise, which is projected to be between
13 cm and 94 cm by 2100 (Warrick et al., 1996), portends increased flood and
wave damage. Massachusetts also has a recent historical 2 millimeters of annual
relative sea-level rise due to subsidence (Giese, Aubrey, and Zeeb, 1987), which if
extrapolated over the Warrick et al. study span could amount to an additional 20 cm
by 2100. Wind, snow, and ice damage may also increase, depending on the highly
uncertain effects of global climate change on storm frequency and intensity.
Recent experience underlines the significance of the storm hazard to Massa-
chusetts coastal communities whether climate change and accelerated sea-level
rise occur or not. The area weathered four especially severe coastal storms in the
nineteen-month period between August 1991 and March 1993: Hurricane Bob in
1991 and three severe extratropical storms, regionally known as ‘northeasters’: the
‘Halloween Northeaster’ or ‘No-name Storm’ of 1991, the ‘Blizzard of December
’92’, and the ‘Blizzard of March ’93’, which affected the entire eastern seaboard
of the United States. Most recently, the area endured a major snowstorm in January
1996 that left two feet of new snow on eastern Massachusetts, set records for total
snow depth in the area, and put the region on the way to a new record for seasonal
snowfall (Nealon, and Brelis, 1996; see Figures 1, and 2). The damage from each of
these storms pales in comparison to the flood damage from the Blizzard of 1978,
when federal assistance for coastal Massachusetts totalled $38 million, and Red
Cross spending in Revere alone totalled $400 000 (Corps of Engineers, 1979; Hunt,
1990). Each event also caused serious disruption beyond its monetary toll.

VULNERABILITY OF COASTAL COMMUNITIES 65
TABLE II.
Key themes and possible exposure/resistance/resilience scenarios
Age. Young children entail an extra burden of child care, which may be disrupted
during storm events (exposure). Parents may lose work time if their daycare is
closed down. The same may apply to adults in elder care. Both young and old
populations also may be unable to resist storms or respond on their own, although
the vulnerability of elders is decreased by their wealth of experience.
Disabilities. Disabilities can hinder taking action in any of the phases of
vulnerability.
Family structure and social networks. Large families may be harder to care for
or keep track of in a storm and its aftermath (exposure, resistance, resilience).
Yet if enough are working, large families could be a benefit in sharing of re-
covery costs. Social networks play a role in disaster warning, perception, and
behavior (exposure, resistance). Strong social networks may help bear the cost
of rebuilding as well (resilience).
Housing and the built environment. The spatial distribution of the built envi-
ronment is highly inertial, establishing patterns of peoples’ location, use, and
travel (exposure). Likewise the quality of construction can determine the ability
of individuals or groups to successfully ride out a storm (resistance).
Income and material resources. Income allows spending on prevention items
such as retrofitting the house or relocating furnaces to higher floors in a flood.
Money or vehicles may also enable a fast exit from a hazardous environment.
(exposure, resistance) Of course, money also allows rebuilding to proceed (re-
silience). On the other hand, those with a lot of wealth have a lot to lose
(exposure).
Lifelines (which includes transportation, communication, utilities, emergency
response, and hospitals). Telephones and the media provide advance warning
of storms. Transportation can provide a way out of the hazard for evacuation or
a way in if caught in rush hour (exposure). Utilities of all sorts are necessary for
resistance and resilience. Hospitals and emergency response of course provide
resistance to the storm, and help to those who have already fallen victim (re-
silience). Emergency response crews can also lessen exposure by helping people
evacuate.
Occupation. Some occupations, such as fishing, tend to be located in harm’s
way (exposure). Once equipment is ruined, it may be months and a whole fish-
ing season away before insurance or relief payments begin, so opportunity is
lost. Self-employed people often have poor documentation of business receipts
and therefore may have difficulty establishing the record necessary to receive
recovery aid (resilience).
Race and ethnicity. Minorities may encounter discrimination when seeking post-
storm aid (resilience) which may change the ability to prepare for the next
storm (resistance). They may be confined by real estate discrimination to certain
hazard-prone neighborhoods (exposure).

66 GEORGE E. CLARK ET AL.
Figure 2. Temporary pipes snake across a street in the low-lying neighborhood of Roughan’s Point
in Revere, during the January ‘Blizzard of ’96’. City workers pumped the storm drains in order to
prevent damage due to tidal flooding in this physically vulnerable neighborhood.
The problem
Such impacts and the potential for worse ones with climate change indicate the
value of a better understanding of how hazardous events and human populations
interact. Aggregate estimates of damage of the sort just cited, though useful, do

VULNERABILITY OF COASTAL COMMUNITIES 67
not address the point emphasized in vulnerability analysis: that the impacts of haz-
ardous events are unevenly distributed among and within the exposed populations.
Methods
The literature on vulnerability identifies many elements contributing to differential
ability to cope with hazards. Ones frequently mentioned are age, disabilities, family
structure and social networks, housing and the built environment, income and ma-
terial resources, lifelines (including transportation, communication, utilities, and
other services), occupation, and race and ethnicity. Studies addressing each of
these themes are listed in Table I. Less explicitly dealt with in the vulnerability
literature were transience, immigration, and education levels, but the significance
of all three can be inferred from discussions of the importance of hazard perception
and experience (e.g. Mitchell, 1984). Table II illustrates how various attributes may
influence the ability to deal with and recover from storms.
We chose Census data to represent these attributes because they are widely
available and familiar to local managers. Data from the 1990 Census on 34 vari-
ables reflecting themes from the literature were assembled at the block group
level (Table III). The block group is a Census unit containing approximately 1000
people; this unit is chosen for analysis because it is the smallest for which rela-
tively complete socioeconomic data is available. We acknowledge that vulnerabil-
ity varies on small scales and even at the household level.∗Nevertheless, the block
group is a practical unit in advising local officials on the allocation of resources.
We used factor analysis as an objective way to simplify our multivariate data
set. Factor analysis allows researchers to identify and cluster variables that measure
essentially the same underlying theme. Factorial ecology in the 1960’s and 1970’s,
for example, showed that many descriptive attributes of urban populations could be
grouped into a few factors, but that these factors were not further reducible. These
factors displayed differing and characteristic spatial patterns such that city form
was not simply a matter of overall ring or sector patterns of all phenomena, but a
mosaic resulting from the superimposition of socioeconomic status, family status,
and ethnicity. A factor analysis of coping abilities shows how far, and in what
combinations, the many variables suggested by the literature measure separate or
similar characteristics in the chosen block groups.
We identified five most important factors from the analysis (Table IV) and
gave each an appropriate name. The factor that accounted for the largest amount
∗Revere has 44 block groups; we extracted and analyzed the data for a larger area consisting of
Revere and the adjoining municipalities of Malden, Lynn, and Saugus, with 225 block groups in all.
The larger scope allows us to have robustness of analysis as well as a sense of how Revere (with the
most floodprone residential area) compares to other towns in the watershed. At the same time, our
focus on Revere allows us to keep the focus on social variation within one community rather than
institutional differences between neighboring jurisdictions.

68 GEORGE E. CLARK ET AL.
TABLE III.
Census variables analyzed
WRKPRNT Children 17 and under whose resident parents or guardians all
work, expressed as a percentage of the population.
AMERINPC Percent of people who are American Indian, Eskimo, or Aleutian.
ASIAPIPC Percent of people who are Asian or Pacific Islander.
BIGFAM Households with 7 people per occupied housing unit.
BLACKPC Percent of population which is Black.
BLD.39PB Percent of housing units which were built prior to 1939.
BLIZIMIG Percent of the population which was foreign-born and came to the
U.S. between 1980 and 1990 (approximation of post-
Blizzard of ’78).
CARPLPC Percent of people who carpool to work.
CARSPCAP Cars per person.
CHILD5PC Percent of people age 5 and under.
COMMUTE Average travel time to work (minutes) not including those who
work at home nor those who do not work.
DISABLPC Work-disabled people per capita.
FEMALEPC Percent of population which is female.
FISHERPC Percent of population employed in fishing, agriculture, or forestry.
HISPNCPC Percent of the population which is Hispanic.
HOMELESS Percent of population counted in shelters or on the street.
HOMEVALU Median value (dollars) of owner-occupied homes.
IMMOBLPC Percent of the population which is physically immobile.
LANGISOL Percent of the population which lives in a household without at
least one English-speaking adult or older child.
LOCAREPC Percent of the population which has a low capacity for self-care.
MTG35.PH Percent of owner-occupied households with mortgages 35 percent
or more of household income.
NEWCMRPC Percent of the population which moved in from 1989 to 1990.
NEWIMMIG Percent of the population which is foreign-born and entered the
U.S. between 1987 and 1990.
NODIPL18 Adults with educational levels less than a high school diploma.
NURSHMPC Percent of people who live in nursing homes.
NWCMSTRM Percent of the population which moved in between 1980 and 1990
(approximation of post-Blizzard of ’78).
OLD65.UP Percent of people age 65 and up.
PCINCOME Per capita income in 1989.
PHONEPH Percent of occupied housing units with no telephone.
POVRTYPC Percent of people with incomes below the federally-defined poverty
line.
PUBTRANS Percent of the population that travels to work on public transporta-
tion.
RACEOTPC Percent of race not White, Black, Native American, Asian, or
Hispanic.
RENTINC Median gross rent as a percent of median household income.
SLFMPLPC Percent of people who are self-employed.

VULNERABILITY OF COASTAL COMMUNITIES 69
TABLE IV.
Groupings of vulnerability factors
Literature Themes Census Variables Factor Name
Factor Grouping 1
Income/Resources Low Income ‘Poverty’
High Fed. Poverty Variance explained: 25%
Race/Ethnicity Hispanic
‘Other’ Race
Black
Education Few H.S. Diplomas
Lifelines Fewer Cars
Factor Grouping 2
Transience Newcomer Since ’80 ‘Transience’
Recent Newcomers Cumulative variance explained: 24% [postpublication correction: 34%]
Built environment housing Few High Mortgages
Factor Grouping 3
Disabilities Immobile ‘Disabilities’
Low Self-Care Cumulative variance explained: 42%
Disabled
Factor Grouping 4
Race/Ethnicity Asian & Pacific ‘Immigrants’
Cumulative variance explained: 49%
Lifelines Public Transport
Immigrants Immigrants Since ’80
Recent Immigrants
Factor Grouping 5
Age Few Elders ‘Young Families’
Cumulative variance explained: 55%
Family Structure Many Children <5
Working Parents
xx

70 GEORGE E. CLARK ET AL.
Figure 3. Factor scores on the ‘poverty’ factor, Revere, MA. The coastline is to the east, with tidal
rivers to the north and south. Boston lies directly south along the coast. In this and following maps,
north is at the top of the page.
of variance included issues of income and material resources, race and ethnicity,
education, and lifelines (transportation, utilities, and emergency response–see Platt,
1991). For want of a more inclusive term, and because per capita income had the
strongest association of all the variables, we named this factor ‘poverty’.
The remaining factors, in order of variance explained, were ‘transience’, as-
sociated most with the people new to the area since a record-setting blizzard in
1978; ‘disabilities’, which included measures of immobility, ability to care for
oneself, and work disability; ‘immigrants’, which reflected the concentration of
recent Asian immigrants in certain parts of the study area; and ‘young families’,
which represented families with young children without a parent at home, high
numbers of young children, and a low proportion of elderly residents.
Four more factors, which showed lesser significance, but had eigenvalues of
greater than one, included the following groups of census variables: few homeless
and high percentage of women; fishing occupations, nursing homes, self employ-

VULNERABILITY OF COASTAL COMMUNITIES 71
ment, and Native Americans; few telephones; and rent as a high percentage of
income. For lack of a compelling reason to exclude them, they were incorporated
in the composite vulnerability indices described below.
The factor analysis thus underlines the complex and multidimensional character
of hazard vulnerability. Low coping ability (high social vulnerability) cannot be
reduced to a single variable such as income or unfamiliarity with the area and
its hazards. At the same time, the results of the factor analysis suggest that the
multidimensional complexity of coping ability may be represented by far fewer
factors than the original list of thirty-four variables.
Results
We mapped the factor scores for each individual factor onto a base map of block
groups for the town of Revere. Figure 3 is a map of factor loadings for the ‘poverty’
factor. Remember that ‘poverty’ as used here includes a complex of several Census
variables. Similar maps were created for each factor.
Mapping each of the factors independently provides valuable information; how-
ever, it is also useful to combine the multidimensional factors into a single scalar
measure of social vulnerability to provide an overall assessment. We use two meth-
ods for obtaining the scalar index of coping ability, (1) averaging, which provides
an absolute index, and (2) data envelopment analysis (DEA), which produces a rel-
ative measure. The method which practitioners would choose to use in a study de-
pends on a number of circumstances, including confidence in weighting of factors
and the end use of the index.
The most common way to combine factors would be to create an index based
upon a weighted average, where the weights reflect the importance of each of
the factors to the activities or decisions that need to be made. One difficulty in
implementing this method centers on the way in which the weights are obtained,
which often requires subjective assessments of importance. Another characteristic
of this technique is that averaging may obscure high values on one of the factors
when it is combined with other factors whose scores are low. For some end uses of
an index, e.g., emergency response, this may not be desirable since extreme values
may indicate where resources are needed most.
Creating a scalar index using DEA (See Haynes et al., 1993; and Charnes,
Cooper, and Rhodes, 1979) has the same mathematical structure as the weighted
average except that the weights are obtained for each block group objectively
through the use of an optimization model. In this method, no subjective apriori
evaluations need to be made about the weights. In addition, when DEA is used,
block groups that have high values on only one factor may still be identified as
vulnerable. The DEA index technique provides a relative measure of vulnerability,
ranking block groups on the basis of their comparative degrees of social vulnera-
bility. Note that the DEA measure can change significantly if new block groups are

72 GEORGE E. CLARK ET AL.
Figure 4. Average vs. DEA overall social vulnerability, Revere, MA. Composite social vulnerability
using the average method is on the left, and using the Data Envelopment Analysis (DEA) method is
on the right. The block group on the coast at ‘A’ appears more vulnerable using the DEA method be-
cause its inhabitants are highly transient. The block group on the coast at ‘B’ appears more vulnerable
using the DEA method because it contains a large number of disabled inhabitants.
added to the analysis, making this technique sensitive to the set of block groups
chosen.
Figure 4 shows the average social vulnerability for Revere on the left (in the
absence of other information all factors were equally weighted) and the DEA social
vulnerability index on the right. Darker areas are calculated to have a higher degree
of social vulnerability. As expected, there are several blockgroups that appear more
vulnerable in the DEA map due to the dampening of high values in the weighted
average. For example, one block group (A)is flagged because of a high factor score
on transience and appears more vulnerable in the DEA map. Block group (B) has a
high factor score for disabilities, and so it also is more vulnerable in the DEA map.
For the remainder of the paper, we focus on the DEA map, using a three-category
(or low, medium, and high vulnerability) version of the DEA map when discussing
overall social vulnerability.

VULNERABILITY OF COASTAL COMMUNITIES 73
Figure 5. Physical flood exposure, Revere, MA.
In order to understand how social vulnerability interacts with physical exposure
to the hazard, we adapted a FEMA Insurance Rate Map of flood zones for Revere
(Figure 5). This adapted map has three different risk zones: one of no risk (light
gray), one 500 y flood risk zone (medium gray), and one zone subject to both 500-
year flood risk and wave action (dark gray). (In our study area, the 100- and 500 y
floodplains are coincident for all practical purposes. Either term could be used.)
It is important to note that, though a smaller risk is assumed for people who do
not live in the floodplain, they may also suffer the effects of severe storms. They
may be affected by wind, rain, snow, and ice damage, and their routes of travel,
day care centers, other lifelines, and places of business may also be affected by
flooding.
Analysis
Finding all those areas that are both physically high-risk and socioeconomically
in less of a position to cope with the hazard allows us to display the interaction of

74 GEORGE E. CLARK ET AL.
Figure 6. Overall social (DEA) and physical vulnerability, Revere, MA. Note that the upper left
legend category, the combination of highest physical exposure and lowest DEA socioeconomic
vulnerability, does not occur in Revere.
physical risk with socioeconomic dimensions of resistance and resilience. Analysis
of how each block group of Revere scores on the physical and socioeconomic
scales lets us create a new, overall vulnerability map of Revere (Figure 6). The
legend shows increasing physical vulnerability in darker shades of gray on the
Y-axis and increasing social vulnerability in denser crosshatching on the X-axis.
The immediate conclusion that can be drawn from these maps that show how
vulnerability varies spatially is that the threat of physical vulnerability is differen-
tially compounded by social vulnerability. Maps at this level of generality are quite
useful in pointing out areas that need more in-depth attention. For example, coastal
zone and emergency managers have to focus their finite time, personnel, and fiscal
resources when deciding on hazard mitigation projects. This choice may be guided
by physical vulnerability only (e.g., 100 y flood zones on FEMA maps) or they may
be guided, as suggested here, by a combination of physical and social vulnerability
(e.g., choosing all combinations of medium to high physical and medium to high

VULNERABILITY OF COASTAL COMMUNITIES 75
Figure 7. Poverty factor scores and physical vulnerability, Revere, MA. Note that the upper right
legend category, the combination of highest poverty factor scores and highest physical exposure,
does not occur in Revere.
social vulnerability; or choosing only the areas of greatest physical AND social
vulnerability, and so on).
In order to then find out what makes an area socially vulnerable, we have to
go back to the factor level, and perhaps ultimately even the variable level. The
underlying social situation that makes a particular area vulnerable can then inform
the strategy and choice of mitigation effort. For example, if an area scores high
on the ‘disability’ factor, emergency evacuation plans could be adjusted to assure
earlier warning and evacuation, and the availability of transportation for the phys-
ically less able or mobile. In other words, the type and degree of intervention can
be adjusted to the special needs in specified floodprone coastal areas. It should be
apparent that the scale of analysis is important in determining the use of a map, and
that a gradation of scales can be used as a sifting tool, informing where to invest
time and resources as more focused studies and actions are planned.
In our study, we go one step back down the scale to the factor level. Figure 7
is a simplified (three-category) version of poverty factor map crosstabulated with

76 GEORGE E. CLARK ET AL.
Figure 8. Vulnerability and the hazards causal model.
the FEMA flood zone map. Figure 7 shows why in a particular area people are so
vulnerable and how they intersect with the three levels of physical risk.
Implications for causal modelling of hazards
More succinctly describing the components of vulnerability through factor analysis
allows us to add the concept of vulnerability to a model of how hazards happen. The
causal model proposed in Perilous Progress by Hohenemser, Kasperson, and Kates
(1985) outlines the circumstances that must coincide at the intersection of the hu-
man and physical systems in order for a hazardous event to take place. Furthermore,
their model is constructed in a way that highlights possible points of appropriate
and timely intervention in order to avert the consequences–opportunities to expand
the ‘practical range of choice’ (Wescoat, 1987; White, 1961).
Recognizing and incorporating into the causal model the spatial distribution
of the physical and social components of vulnerability (in this case the factors of
poverty, transience, immigrants, disability, and young families) will help interested

VULNERABILITY OF COASTAL COMMUNITIES 77
parties focus attention on problems of social justice, a major issue in emerging
hazards and global environmental change research (Wescoat, 1993).
According to the causal model, hazards emerge through a sequence of choices
and occurrences: human needs, human wants, choice of technology, initiating events,
outcomes, exposure, and consequences. A reinterpretation of the hazards causal
model to incorporate vulnerability based on this research appears in Figure 8.
In our adaptation of the hazards causal model from technical to environmental
hazards, we substitute ‘land use’ for ‘choice of technology’. In doing so, we do not
intend to imply that land use or social situation is a ‘choice’ for everyone, merely
that landscapes are socially adapted phenomena and need to be thought of as such
rather than as tabulae rasae on which environmental hazards play themselves out.
In our model, exposure, resistance, and resilience each function as a filter or
amplifier for the impacts of the hazardous event at different stages. Where these
components of vulnerability are seen to worsen the hazard impact, they identify
places where progress could be made in hazard management.
Exposure is determined by both existing land use and by the characteristics of
the storm event, such as timing, duration, and intensity. At the same time, the storm
as a hazardous event implies an interaction with human systems, so the arrow goes
both ways between exposure and storm event, producing an intersection, known
as an outcome. For example, flood waters rise to a given height, which potentially
affects a given number of houses, businesses, cars, people, etc. The consequences
(i.e., the damages to life, health, well-being, and property) of an outcome are then
determined by peoples’ resistance to the hazard. Finally, the ability to respond (i.e.,
recover from the event and prepare for the next) is determined by the resilience of
the population or individual.
Another aspect of this model is that a map of each factor can be used to inform
each stage of vulnerability (exposure, resistance, and resilience). To use the poverty
example, poor people may be constrained by the real estate market to a given
neighborhood which has poorly maintained drainage or physical protection from
the hazard (exposure). In terms of resistance, poor people may lack the money
or equipment to buy or make stormproofing adjustments to their homes. For re-
silience, they may not have been able to purchase insurance to let them rebuild. To
continue the filter/amplifier metaphor of the causal model, ‘turning down’ a vul-
nerability factor such as poverty in a given area makes the outcomes of hazardous
events become more favorable.
Structuralist practitioners, such as Hewitt (1997, 1983, and especially chapters
13 and 14 in Hewitt, 1983: Watts, 1983: Susman, O’Keefe, and Wisner, 1983);
Blaikie (1985); and Wisner (1992) might argue (somewhat correctly) that the causal
model addresses only the proximate causes of vulnerability, as opposed to systemic
root causes such as global capitalism, colonialism, racism, and so on. They might
further argue that we merely point out poverty rather than question why poverty
exists here and elsewhere in the first place.

78 GEORGE E. CLARK ET AL.
We agree that these are necessary efforts. However, we would like to emphasize
intervention points in the causal chain that can be addressed within the scope of
coastal policy-makers, managers, and community organizers. We also argue that
we too attempt to promote a somewhat sophisticated view of poverty as including
elements of economics, racism, education, and transportation (see Table IV). Fur-
ther, we encourage the identification and rectification of more upstream causes of
vulnerability as local analysis is performed.
Concerted interagency and community efforts to forge better links between
emergency management, public works, labor and employment agencies, and pri-
vate organizations are useful pathways of action at the community level even if
they do not fix, for example, the exploitation of labor through global capitalism.
By speaking to the scale, areas, and processes over which local leaders have some
control, we hope to leave room for action and even optimism on a grassroots level.
Implications for coastal hazard management
Such practical damage reduction programs might include contingency plans for
weather-robust public transportation to and from shelters, insurance centers, places
of employment, and regional rail networks, since transportation access issues score
highly on each of the poverty, disability, and immigrant factors. Interpreters and
victim advocates might be used to help those covered by the immigrants and tran-
sience factors to communicate with or to be made aware of available community re-
sources. Many potential initiatives can be identified through vulnerability analysis,
and the more ‘proactive’ or ‘upstream’ the step taken, the greater the downstream
benefit. Reduced exposure through improvement or elevation of housing stock
means fewer people needing to resist or overcome effects of storms. Increased
resistance means less need for recovery, and so on.
It is no accident that maps of vulnerability feature prominently in this frame-
work. Based on our factor analysis, they represent a succinct understanding of the
distribution of a large number of physical and social variables which is crucial to
a more complete picture of the hazards process. On a practical basis, vulnerability
maps similar to the ones produced in this study have potential benefits for both cri-
sis management and planning for future contingencies, including potential impacts
of global climate change. For example, disaster-management resources can be
more easily and swiftly directed to the locales of greatest need, and resources can
be applied according to the specific factors that render a certain population socially
vulnerable. Vulnerability maps thus allow for the focusing of limited resources on
areas of highest priority or of potentially greatest improvement.
Likewise these vulnerability maps should be very useful for generating scenar-
ios. One can model a ‘100 y coastal storm’ (or any other hazardous event) and see
who is affected in various ways. Thus, vulnerability maps could be used not only as
a response tool to intervene in an ongoing event, but also as an impact assessment

VULNERABILITY OF COASTAL COMMUNITIES 79
tool that is more sophisticated than most current global change impact assessment
models, and hence as a planning tool that can be employed even in spite of lacking
localized information about changing sea level and storm climate.
By focusing attention and action on communities’ current ability to deal with
coastal hazards on a local scale, this study suggests that there are many opportu-
nities to improve on hazard management that are beneficial in the present and that
leave vulnerable areas in a better position for the future if hazards become exacer-
bated by climate change. Localities using this type of ‘no regrets’ approach would
be enabled to take action against hazards in spite of a high level of uncertainty
about the future.
Acknowledgements
This research has been supported by the National Oceanographic and Atmospheric
Administration (NOAA) Climate and Global Change Program, Human Dimen-
sions Section, and by the Department of Energy through the Northeast Regional
Center of the National Institutes for Global Environmental Change (NE NIGEC).
An earlier version of this paper was presented at the First Open Meeting of the
Human Dimensions of Global Environmental Change Community at Duke Uni-
versity, June 3, 1995. Opinions expressed herein are those of the authors and not
necessarily of the supporting agencies, authors’ employers, or other organizations.
This article is dedicated to the memory of Harry E. Schwarz, colleague, teacher,
mentor, and friend. He is greatly missed.
References
Bijlsma, L. et al.: 1996, Coastal zones and small islands. In: Watson, Robert T. et al. (eds.),
Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: Scientific-
Technical Analyses. Contribution of Working Group II to the Second Assessment Report of the
Intergovernmental Panel on Climate Change, 299–324. Cambridge, UK: Cambridge University
Press.
Blaikie, P.: 1985, The Political Economy of Soil Erosion in Developing Countries. Harlow, England:
Longman.
Blaikie, P. M., Cannon, T., Davis, I. and Wisner, B.: 1994, At Risk: Natural Hazards, People’s
Vulnerability, and Disasters. London: Routledge.
Bohle, H. G., Downing, T. E. and Watts, M. J.: 1994, Climate change and social vulnerability:
towards a sociology and geography of food insecurity. Global Environmental Change 4:37–48.
Bolin, R.: 1982, Long-Term Family Recovery from Disaster. Monograph # 36. Boulder, CO.: Institute
of Behavioral Science, University of Colorado.
Bolin, R. and Bolton, P.: 1986, Race, Religion, and Ethnicity in Disaster Recovery. Program on
Environment and Behavior Monograph # 42. Boulder, CO.: Institute of Behavioral Science,
University of Colorado.
Bolin, R. and Klenow, D.: 1983, Response of the elderly to disaster: an age-stratified analysis.
International Journal of Aging and Human Development 16: 283–296.

80 GEORGE E. CLARK ET AL.
Bolin, R. and Stanford, L.: 1991, Shelter, housing, and recovery: A comparison of U.S. disasters.
Disasters 15: 24–34.
Bruce, J. P., Lee, H. and Haites, E. F., (eds.): 1996, Climate Change 1995: Economic and Social
Dimensions of Climate Change. Contributions of Working Group III to the Second Assess-
ment Report of the Intergovernmental Panel on Climate Change. New York, NY: Cambridge
University Press.
Cash, D.: 1997, Local response to global change: Creating effective bridges between science, policy
and action. Paper prepared for the workshop on Regional Climate Change Impacts on Great
Plains Ecosystems, May 27–29, 1997, Fort Collins, CO.
Charnes, A., Cooper, W. W. and Rhodes, E.: 1979, Measuring the efficiency for decisionmaking
units. European Journal of Operational Research 2.
Chen, R. S.: 1994, The human dimensions of vulnerability. In R. H. Socolow, C. Andrews, F.
Berkhout, and V. Thomas (eds.), Industrial Ecology and Global Change. Cambridge: Cambridge
University Press, 85–105.
Clark, W. C., Jager, J. and van Eijndhoven, J. (eds.): Forthcoming. Learning to Manage Global
Environmental Risks: A Comparative History of Social Responses to Climate Change, Ozone
Depletion, and Acid Rain. Cambridge, MA: MIT Press.
Corps of Engineers: 1979, A Report on the Assessment of Flood Damages Resulting from the Storm
of 6–7 February 1978 along the Coastline from Orleans, Massachusetts to New Castle, New
Hampshire. February. Waltham, MA.: U. S. Army Corps of Engineers, New England Division.
Cutter, S. L.: 1996, Vulnerability to environmental hazards. Progress in Human Geography 20: 529–
539.
Dow, K.: 1992, Exploring differences in our common future(s): The meaning of vulnerability to
global environmental change. Geoforum 23: 417–436.
Dow, K.: 1993, Unpublished literature review on the ‘concept of vulnerability’ and the ‘factors
contributing to vulnerability’. Worcester: George Perkins Marsh Institute,Clark University.
Dow, K. and Downing, T. E.: 1995, Vulnerability research: where things stand. Human Dimensions
Quarterly 1:3–5.
Drabek, T. E., and Key, W. H.: 1984, Conquering Disaster: Family Recovery and Long-Term
Consequences. New York: Irvington.
Global Environmental Assessment Project: 1997, A Critical Evaluation of Global Environmental
Assessments: the Climate Experience. Calverton, MD: CARE.
Godschalk, D. R., Brower, D. J. and Beatley, T.: 1989, Catastrophic Coastal Storms. Durham, NC.:
Duke University Press.
Haas, J. E., Kates, R. W. and Bowden, M. J.: 1977, Reconstruction Following Disaster. Cambridge,
MA.: MIT Press.
Haynes, K., Ratick, S., Bowen, W. and Cummings-Saxton, J.: 1993, Environmental decision models:
U.S. experiences and new approaches to pollution management. Environment International 19:
200–220.
Hewitt, K., (ed.): 1983, Interpretations of Calamity. Boston: Allen & Unwin, Inc.
Hewitt, K.: 1997, Regions of Risk: A Geographical Introduction to Disasters. London: Longman.
Hewitt, K., and Burton, I.: 1971, The Hazardousness of a Place: A Regional Ecology of Damaging
Events. Department of Geography Research Publications. Toronto: University of Toronto Press.
Hohenemser, C., Kasperson, R. E. and Kates, R. W.: 1985, Causal structure. In R. W. Kates, C.
Hohenemser, and J. X. Kasperson (ed.), Perilous Progress: Managing the Hazards of Technology,
43–66. Boulder, CO.: Westview Press.
Hunt, R.: 1990, Saugus River and Tributaries Flood Damage Reduction Study, Lynn, Malden, Revere,
and Saugus, Massachusetts. April. Waltham, MA.: U. S. Army Corps of Engineers, New England
Division.
Kotlyakov, V. M. et al.: 1988, Global change: Geographical approaches (a review). Proceedings of
the National Academy of Sciences USA, 85 (August): 5986–5991.

VULNERABILITY OF COASTAL COMMUNITIES 81
Liverman, D. M.: 1990, Vulnerability to global environmental change. In R. E. Kasperson, K. Dow,
D. Golding, and J. X. Kasperson (eds.), Understanding Global Environmental Change: The
Contributions of Risk Analysis and Management. Worcester, MA.: Earth Transformed Program,
Clark University.
Mitchell, J. K.: 1984, Hazard perception studies: Convergent concerns and divergent approaches dur-
ing the past decade. In T. F. Saarinen, D. Seamon, and J. L. Sell, eds., Environmental Perception
and Behavior: An Inventory and Prospect, 33–59. Research Paper No. 209. Chicago: Department
of Geography, University of Chicago.
Moser, S. C.: 1997, Mapping the Territory of Uncertainty and Ignorance: Broadening Current
Assessment and Policy Approaches to Sea-Level Rise. Dissertation. Worcester, MA: Clark
University.
Mitchell, J. K.: 1984, Hazard perception studies: Convergent concerns and divergent approaches dur-
ing the past decade. In T. F. Saarinen, D. Seamon, and J. L. Sell (ed.), Environmental Perception
and Behavior: An Inventory and Prospect, 33–59. Research Paper No. 209. Chicago: Department
of Geography, University of Chicago.
Nealon, P. and Brelis, B.: 1996, The eastcoaster of ’96: N. E. digs out; forecasters say to dig in for
more. The Boston Globe (January 9):1,21.
NRC (National Research Council): 1984, Hurricane Diana, North Carolina, September 10–14, 1984.
Washington, DC.: National Academy Press.
Parr, A. R.: 1987, Disasters and disabled persons: An examination of the safety needs of a neglected
minority. Disasters 11: 148–159.
Perry, R. W., Greene, M. and Mushkatel, A.: 1983, American Minority Citizens in Disaster. Seattle:
Battelle.
Perry, R. W. and Lindell, M. K.: 1991, The effects of ethnicity on evacuation decisionmaking.
International Journal of Mass Emergencies and Disasters 9:47–68.
Platt, R.: 1991, Lifelines: An emergency management priority for the United States in the 1990’s.
Disasters 15: 172–176.
Quarantelli, E. L.: 1991, Patterns of Sheltering and Housing in American Disasters. Preliminary
Paper # 170. Newark, DE.: University of Delaware, Disaster Research Center.
Renn, O.: 1992, Concepts of Risk: A Classification. In S. Krimsky, and D. Golding (eds.), Social
Theories of Risk. Westport, CT.: Praeger, 53–79.
Rossi, P. H., Wright, J. D., Weber-Burdin, E. and Pereira, J.: 1983, Victims of the Environment: Loss
From Natural Hazards in the United States, 1970–1980. New York: Plenum Press.
Susman, P.,O’Keefe, P. and Wisner, B.: 1983, Global disasters: a radical interpretation. In K. Hewitt
(ed.), Interpretations of Calamity. Boston: Allen & Unwin, Inc., 263–283.
Timmerman, P.: 1981, Vulnerability, Resilience, and the Collapse of Society. Toronto: Institute for
Environmental Studies. Environmental Monograph 1.
Timms, D. W. G.: 1971, The Urban Mosaic: Towards a Theory of Residential Differentiation.
Cambridge: Cambridge University Press.
Trainer, P. and Bolin, R.: 1976, Persistent effects of disasters on daily activities: a cross-cultural
comparison. Mass Emergencies 1:279–290.
Turner, B. L., Kasperson, R. E., Meyer, W. B., Dow, K. M., Golding, D., Kasperson, J. X., Mitchell,
R. C. and Ratick, S. J.: 1990, Two types of global environmental change: Definitional and spatial-
scale issues in their human dimensions. Global Environmental Change 1:14–22.
Vigue, D. I.: 1997, Some don’t Revere the name. Boston Globe (March 23), B1, B5.
Warrick, R. A., Le Provost, C., Meier, M. F., Oerlemans, J. and Woodworth, P. L.: 1996, Changes in
sea level. In J. T. Houghton, L. G. Meira Filho, B. A. Callander, N. Harris, A. Kattenberg, and
K. Maskell, eds. Climate Change 1995: The Science of Climate Change. Cambridge: Cambridge
University Press.
Watson, R.T. et al. (eds.): 1996, Climate change 1995: Impacts, adaptations and mitigation of climate
change: Scientific-technical analyses. Contribution of Working Group II to the Second Assess-

82 GEORGE E. CLARK ET AL.
ment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK: Cambridge
University Press.
Watts, M.: 1983, On the poverty of theory: natural hazards research in context. In Hewitt, K., ed.,
Interpretations of Calamity. Boston: Allen & Unwin, Inc., 231–262.
Watts, M. J., and Bohle, H. G.: 1993, The space of vulnerability: the causal structure of hunger and
famine. Progress in Human Geography 17: 43–67.
Wescoat, J. L., Jr.: 1987, The ‘practical range of choice’ in water resources geography. Progress in
Human Geography 11: 41–59.
Wescoat, J. L., Jr.: 1993, Resource management: UNCED, GATT, and global change. Progress in
Human Geography 17: 232–240.
White, G. F. 1961. The choice of use in resources management. Natural Resources Journal 1
(March): 23–40.
White, G. F., and Haas, J. E.: 1975, An Assessment of Research Needs on Natural Hazards.
Cambridge, MA. MIT Press.
Wisner, B.: 1992, Disaster vulnerability. In H.-G. Bohle, ed., Worlds of Pain and Hunger: Geo-
graphical Perspectives on Disaster Vulnerability and Food Security,13–52. Saarbrucken: Verlag
Breitenbach.