Research on Disaster Response and Recovery
This chapter and the preceding one use the conceptual model presented in Chapter 1 (see Figure 1.1) as a guide to understanding societal response to hazards and disasters. As specified in that model, Chapter 3 discusses three sets of pre-disaster activities that have the potential to reduce disaster losses: hazard mitigation practices, emergency preparedness practices, and pre-disaster planning for post-disaster recovery. This chapter focuses on National Earthquake Hazards Reduction Program (NEHRP) contributions to social science knowledge concerning those dimensions of the model that are related to post-disaster response and recovery activities. As in Chapter 3, discussions are organized around research findings regarding different units of analysis, including individuals, households, groups and organizations, social networks, and communities. The chapter also highlights trends, controversies, and issues that warrant further investigation. The contents of this chapter are linked to key themes discussed elsewhere in this report, including the conceptualization and measurement of societal vulnerability and resilience, the importance of taking diversity into account in understanding both response-related activities and recovery processes and outcomes, and linkages between hazard loss reduction and sustainability. Although this review centers primarily on research on natural disasters and to a lesser degree on technological disasters, research findings are also discussed in terms of their implications for understanding and managing emerging homeland security threats.
The discussions that follow seek to address several interrelated questions: What is currently known about post-disaster response and recovery,
and to what extent is that knowledge traceable to NEHRP-sponsored research activities? What gaps exist in that knowledge? What further research—both disciplinary and interdisciplinary—is needed to fill those gaps?
RESEARCH ON DISASTER RESPONSE
Emergency response encompasses a range of measures aimed at protecting life and property and coping with the social disruption that disasters produce. As noted in Chapter 3, emergency response activities can be categorized usefully as expedient mitigation actions (e.g., clearing debris from channels when floods threaten, containing earthquake-induced fires and hazardous materials releases before they can cause additional harm) and population protection actions (e.g., warning, evacuation and other self-protective actions, search and rescue, the provision of emergency medical care and shelter; Tierney et al., 2001). Another common conceptual distinction in the literature on disaster response (Dynes et al., 1981) contrasts agent-generated demands, or the types of losses and forms of disruption that disasters create, and response-generated demands, such as the need for situation assessment, crisis communication and coordination, and response management. Paralleling preparedness measures, disaster response activities take place at various units of analysis, from individuals and households, to organizations, communities, and intergovernmental systems. This section does not attempt to deal exhaustively with the topic of emergency response activities, which is the most-studied of all phases of hazard and disaster management. Rather, it highlights key themes in the literature, with an emphasis on NEHRP-based findings that are especially relevant in light of newly recognized human-induced threats.
Public Response: Warning Response, Evacuation, and Other Self-Protective Actions
The decision processes and behaviors involved in public responses to disaster warnings are among the best-studied topics in the research literature. Over nearly three decades, NEHRP has been a major sponsor of this body of research. As noted in Chapter 3, warning response research overlaps to some degree with more general risk communication research. For example, both literatures emphasize the importance of considering source, message, channel, and receiver effects on the warning process. While this discussion centers mainly on responses to official warning information, it should be noted that self-protective decision-making processes are also initiated in the absence of formal warnings—for example, in response to cues that people perceive as signaling impending danger and in disasters that occur without warning. Previous research suggests that the basic deci-
sion processes involved in self-protective action are similar across different types of disaster events, although the challenges posed and the problems that may develop can be agent specific.
As in other areas discussed here, empirical studies on warning response and self-protective behavior in different types of disasters and emergencies have led to the development of broadly generalizable explanatory models. One such model, the protective action decision model, developed by Perry, Lindell, and their colleagues (see, for example, Lindell and Perry, 2004), draws heavily on Turner and Killian’s (1987) emergent norm theory of collective behavior. According to that theory, groups faced with the potential need to act under conditions of uncertainty (or potential danger) engage in interaction in an attempt to develop a collective definition of the situation they face and a set of new norms that can guide their subsequent action.1 Thus, when warnings and protective instructions are disseminated, those who receive warnings interact with one another in an effort to determine collectively whether the warning is authentic, whether it applies to them, whether they are indeed personally in danger, whether they can reduce their vulnerability through action, whether action is possible, and when they should act. These collective determinations are shaped in turn by such factors as (1) the characteristics of warning recipients, including their prior experience with the hazard in question or with similar emergencies, as well as their prior preparedness efforts; (2) situational factors, including the presence of perceptual cues signaling danger; and (3) the social contexts in which decisions are made—for example, contacts among family members, coworkers, neighborhood residents, or others present in the setting, as well as the strength of preexisting social ties. Through interaction and under the influence of these kinds of factors, individuals and groups develop new norms that serve as guidelines for action.
Conceptualizing warning response as a form of collective behavior that is guided by emergent norms brings several issues to the fore. One is that far from being automatic or governed by official orders, behavior undertaken in response to warnings is the product of interaction and deliberation among members of affected groups—activities that are typically accompanied by a search for additional confirmatory information. Circumstances that complicate the deliberation process, such as conflicting warning information that individuals and groups may receive, difficulties in getting in touch with others whose views are considered important for the decision-making process, or disagreements among group members about any aspect of the
threat situation, invariably lead to additional efforts to communicate and confirm the information and lengthen the period between when a warning is issued and when groups actually respond.
Another implication of the emergent norm approach to protective action decision making is the recognition that groups may collectively define an emergency situation in ways that are at variance from official views. This is essentially what occurs in the shadow evacuation phenomenon, which has been documented in several emergency situations, including the Three Mile Island nuclear plant accident (Zeigler et al., 1981). While authorities may not issue a warning for a particular geographic area or group of people, or may even tell them they are safe, groups may still collectively decide that they are at risk or that the situation is fluid and confusing enough that they should take self-protective action despite official pronouncements.
The behavior of occupants of the World Trade Center during the September 11, 2001 terrorist attack illustrates the importance of collectively developed definitions. Groups of people in Tower 2 of the World Trade Center decided that they should evacuate the building after seeing and hearing about what was happening in Tower 1 and after speaking with coworkers and loved ones, even when official announcements and other building occupants indicated that they should not do so. Others decided to remain in the tower or, perhaps more accurately, they decided to delay evacuating until receiving additional information clarifying the extent to which they were in danger. Journalistic accounts suggest that decisions were shaped in part by what people could see taking place in Tower 1, conversations with others outside the towers who had additional relevant information, and directives received from those in positions of authority in tenant firms. In that highly confusing and time-constrained situation, emergent norms guiding the behavior of occupants of the second tower meant the difference between life and death when the second plane struck (NIST, 2005).
The large body of research that exists regarding decision making under threat conditions points to the need to consider a wide range of individual, group, situational, and resource-related factors that facilitate and inhibit self-protective action. Qualitatively based decision-tree models developed by Gladwin et al. (2001) demonstrate the complexity of self-protective decisions. As illustrated by their work on hurricane evacuation, a number of different factors contribute to decisions on whether or not to evacuate. Such factors range from perceptions of risk and personal safety with respect to a threatened disaster, to the extent of knowledge about specific areas at risk, to constraining factors such as the presence of pets in the home that require care, lack of a suitable place to go, counterarguments by other family members, fears of looting (shown by the literature to be unjustified; see, for example, Fischer, 1998), and fear that the evacuation process may
be more dangerous than staying home and riding out a hurricane. Warning recipients may decide that they should wait before evacuating, ultimately missing the opportunity to escape, or they may decide to shelter in-place after concluding that their homes are strong enough to resist hurricane forces despite what they are told by authorities.
In their research on Hurricane Andrew, Gladwin and Peacock describe some of the many factors that complicate the evacuation process for endangered populations (1997:54):
Except under extreme circumstances, households cannot be compelled to evacuate or to remain where they are, much less to prepare themselves for the threat. Even under extraordinary conditions many households have to be individually located and assisted or forced to comply. Segments of a population may fail to receive, ignore, or discount official requests and orders. Still others may not have the resources or wherewithal to comply. Much will depend upon the source of the information, the consistency of the message received from multiple sources, the nature of the information conveyed, as well as the household’s ability to perceive the danger, make decisions, and act accordingly. Disputes, competition, and the lack of coordination among local, state, and federal governmental agencies and between those agencies and privately controlled media can add confusion. Businesses and governmental agencies that refuse to release their employees and suspend normal activities can add still further to the confusion and noncompliance.
The normalcy bias adds other complications to the warning response process. While popular notions of crisis response behaviors seem to assume that people react automatically to messages signaling impending danger—for example, by fleeing in panic—the reality is quite different. People typically “normalize” unusual situations and persist in their everyday activities even when urged to act differently. As noted earlier, people will not act on threat information unless they perceive a personal risk to themselves. Simply knowing that a threat exists—even if that threat is described as imminent—is insufficient to motivate self-protective action. Nor can people be expected to act if warning-related guidance is not specific enough to provide them with a blueprint for what to do or if they do not believe they have the resources required to follow the guidance. One practical implication of research on warnings is that rather than being concerned about panicking the public with warning information, or about communicating too much information, authorities should instead be seeking better ways to penetrate the normalcy bias, persuade people that they should be concerned about an impending danger, provide directives that are detailed enough to follow during an emergency, and encourage pre-disaster response planning so that people have thought through what to do prior to being required to act.
Other Important Findings Regarding the Evacuation Process
As noted earlier, evacuation behavior has long been recognized as the reflection of social-level factors and collective deliberation. Decades ago, Drabek (1983) established that households constitute the basic deliberative units for evacuation decision making in community-wide disasters and that the decisions that are ultimately made tend to be consistent with pre-disaster household authority patterns. For example, gender-related concerns often enter into evacuation decision making. Women tend to be more risk-averse and more inclined to want to follow evacuation orders, while males are less inclined to do so (for an extensive discussion of gender differences in vulnerability, risk perception, and responses to disasters, see Fothergill, 1998). In arriving at decisions regarding evacuation, households take official orders into account, but they weigh those orders in light of their own priorities, other information sources, and their past experiences. Information received from media sources and from family and friends, along with confirmatory data actively sought by those at risk, generally has a greater impact on evacuation decisions than information provided by public officials (Dow and Cutter, 1998, 2000).
Recent research also suggests that family evacuation patterns are undergoing change. For example, even though families decide together to evacuate and wish to stay together, they increasingly tend to use more than one vehicle to evacuate—perhaps because they want to take more of their possessions with them, make sure their valuable vehicles are protected, or return to their homes at different times (Dow and Cutter, 2002). Other social influences also play a role. Neighborhood residents may be more willing to evacuate or, conversely, more inclined to delay the decision to evacuate if they see their neighbors doing so. Rather than becoming more vigilant, communities that are struck repeatedly by disasters such as hurricanes and floods may develop “disaster subcultures,” such as groups that see no reason to heed evacuation orders since sheltering in-place has been effective in previous events.
NEHRP-sponsored research has shown that different racial, ethnic, income, and special needs groups respond in different ways to warning information and evacuation orders, in part because of the unique characteristics of these groups, the manner in which they receive information during crises, and their varying responses to different information sources. For example, members of some minority groups tend to have large extended families, making contacting family members and deliberating on alternative courses of action a more complicated process. Lower-income groups, inner-city residents, and elderly persons are more likely to have to rely on public transportation, rather than personal vehicles, in order to evacuate. Lower-income and minority populations, who tend to have larger families, may
also be reluctant to impose on friends and relatives for shelter. Lack of financial resources may leave less-well-off segments of the population less able to afford to take time off from work when disasters threaten, to travel long distances to avoid danger, or to pay for emergency lodging. Socially isolated individuals, such as elderly persons living alone, may lack the social support that is required to carry out self-protective actions. Members of minority groups may find majority spokespersons and official institutions less credible and believable than members of the white majority, turning instead to other sources, such as their informal social networks. Those who rely on non-English-speaking mass media for news may receive less complete warning information, or may receive warnings later than those who are tuned into mainstream media sources (Aguirre et al., 1991; Perry and Lindell, 1991; Lindell and Perry, 1992, 2004; Klinenberg, 2002; for more extensive discussions, see Tierney et al., 2001).
Hurricane Katrina vividly revealed the manner in which social factors such as those discussed above influence evacuation decisions and actions. In many respects, the Katrina experience validated what social science research had already shown with respect to evacuation behavior. Those who stayed behind did so for different reasons—all of which have been discussed in past research. Some at-risk residents lacked resources, such as automobiles and financial resources that would have enabled them to escape the city. Based on their past experiences with hurricanes like Betsey and Camille, others considered themselves not at risk and decided it was not necessary to evacuate. Still others, particularly elderly residents, felt so attached to their homes that they refused to leave even when transportation was offered.
This is not to imply that evacuation-related problems stemmed solely from individual decisions. Katrina also revealed the crucial significance of evacuation planning, effective warnings, and government leadership in facilitating evacuations. Planning efforts in New Orleans were rudimentary at best, clear evacuation orders were given too late, and the hurricane rendered evacuation resources useless once the city began to flood.
With respect to other patterns of evacuation behavior when they do evacuate, most people prefer to stay with relatives or friends, rather than using public shelters. Shelter use is generally limited to people who feel they have no other options—for example, those who have no close friends and relatives to take them in and cannot afford the price of lodging. Many people avoid public shelters or elect to stay in their homes because shelters do not allow pets. Following earthquakes, some victims, particularly Latinos in the United States who have experienced or learned about highly damaging earthquakes in their countries of origin, avoid indoor shelter of all types, preferring instead to sleep outdoors (Tierney, 1988; Phillips, 1993; Simile, 1995).
Disaster warnings involving “near misses,” as well as concerns about the possible impact of elevated color-coded homeland security warnings,
raise the question of whether warnings that do not materialize can induce a “cry-wolf” effect, resulting in lowered attention to and compliance with future warnings. The disaster literature shows little support for the cry-wolf hypothesis. For example, Dow and Cutter (1998) studied South Carolina residents who had been warned of impending hurricanes that ultimately struck North Carolina. Earlier false alarms did not influence residents’ decisions on whether to evacuate; that is, there was little behavioral evidence for a cry-wolf effect. However, false alarms did result in a decrease in confidence in official warning sources, as opposed to other sources of information on which people relied in making evacuation decisions—certainly not the outcome officials would have intended. Studies also suggest that it is advisable to clarify for the public why forecasts and warnings were uncertain or incorrect. Based on an extensive review of the warning literature, Sorensen (2000:121) concluded that “[t]he likelihood of people responding to a warning is not diminished by what has come to be labeled the ‘cry-wolf’ syndrome if the basis for the false alarm is understood [emphasis added].” Along those same lines, Atwood and Major (1998) argue that if officials explain reasons for false alarms, that information can increase public awareness and make people more likely to respond to subsequent hazard advisories.
Dispelling Myths About Crisis-Related Behavior: Panic and Social Breakdown
Numerous individual studies and research syntheses have contrasted commonsense ideas about how people respond during crises with empirical data on actual behavior. Among the most important myths addressed in these analyses is the notion that panic and social disorganization are common responses to imminent threats and to actual disaster events (Quarantelli and Dynes, 1972; Johnson, 1987; Clarke, 2002). True panic, defined as highly individualistic flight behavior that is nonsocial in nature, undertaken without regard to social norms and relationships, is extremely rare prior to and during extreme events of all types. Panic takes place under specific conditions that are almost never present in disaster situations. Panic only occurs when individuals feel completely isolated and when both social bonds and measures to promote safety break down to such a degree that individuals feel totally on their own in seeking safety. Panic results from a breakdown in the ongoing social order—a breakdown that Clarke (2003:128) describes as having moral, network, and cognitive dimensions:
There is a moral failure, so that people pursue their self interest regardless
of rules of duty and obligation to others. There is a network failure, so that the resources that people can normally draw on in times of crisis are no longer there. There is a cognitive failure, in which someone’s understanding of how they are connected to others is cast aside.
Failures on this scale almost never occur during disasters. Panic reactions are rare in part because social bonds remain intact and extremely resilient even under conditions of severe danger (Johnson, 1987; Johnson et al., 1994; Feinberg and Johnson, 2001).
Panic persists in public and media discourses on disasters, in part because those discourses conflate a wide range of other behaviors with panic. Often, people are described as panicking because they experience feelings of intense fear, even though fright and panic are conceptually and behaviorally distinct. Another behavioral pattern that is sometimes labeled panic involves intensified rumors and information seeking, which are common patterns among publics attempting to make sense of confusing and potentially dangerous situations. Under conditions of uncertainty, people make more frequent use of both informal ties and official information sources, as they seek to collectively define threats and decide what actions to take. Such activities are a normal extension of everyday information-seeking practices (Turner, 1994). They are not indicators of panic.
The phenomenon of shadow evacuation, discussed earlier, is also frequently confused with panic. Such evacuations take place because people who are not defined by authorities as in danger nevertheless determine that they are—perhaps because they have received conflicting or confusing information or because they are geographically close to areas considered at risk (Tierney et al., 2001). Collective demands for antibiotics by those considered not at risk for anthrax, “runs” on stores to obtain self-protective items, and the so-called worried-well phenomenon are other forms of collective behavior that reflect the same sociobehavioral processes that drive shadow evacuations: emergent norms that define certain individuals and groups as in danger, even though authorities do not consider them at risk; confusion about the magnitude of the risk; a collectively defined need to act; and in some cases, an unwillingness to rely on official sources for self-protective advice. These types of behaviors, which constitute interesting subjects for research in their own right, are not examples of panic.
Research also indicates that panic and other problematic behaviors are linked in important ways to the manner in which institutions manage risk and disaster. Such behaviors are more likely to emerge when those who are in danger come to believe that crisis management measures are ineffective, suggesting that enhancing public understanding of and trust in preparedness measures and in organizations charged with managing disasters can lessen the likelihood of panic. With respect to homeland security threats, some researchers have argued that the best way to “vaccinate” the public
against the emergence of panic in situations involving weapons of mass destruction is to provide timely and accurate information about impending threats and to actively include the public in pre-crisis preparedness efforts (Glass and Shoch-Spana, 2002).
Blaming the public for panicking during emergencies serves to diffuse responsibility from professionals whose duty it is to protect the public, such as emergency managers, fire and public safety officials, and those responsible for the design, construction, and safe operation of buildings and other structures (Sime, 1999). The empirical record bears out the fact that to the extent panic does occur during emergencies, such behavior can be traced in large measure to environmental factors such as overcrowding, failure to provide adequate egress routes, and breakdowns in communications, rather than to some inherent human impulse to stampede with complete disregard for others. Any potential for panic and other problematic behaviors that may exist can, in other words, be mitigated through appropriate design, regulatory, management, and communications strategies.
As discussed elsewhere in this report, looting and violence are also exceedingly rare in disaster situations. Here again, empirical evidence of what people actually do during and following disasters contradicts what many officials and much of the public believe. Beliefs concerning looting are based not on evidence but rather on assumptions—for example, that social control breaks down during disasters and that lawlessness and violence inevitably result when the social order is disrupted. Such beliefs fail to take into account the fact that powerful norms emerge during disasters that foster prosocial behavior—so much so that lawless behavior actually declines in disaster situations. Signs erected following disasters saying, “We shoot to kill looters” are not so much evidence that looting is occurring as they are evidence that community consensus condemns looting.
The myth of disaster looting can be contrasted with the reality of looting during episodes of civil disorder such as the riots of the 1960s and the 1992 Los Angeles unrest. During episodes of civil unrest, looting is done publicly, in groups, quite often in plain sight of law enforcement officials. Taking goods and damaging businesses are the hallmarks of modern “commodity riots.” New norms also emerge during these types of crises, but unlike the prosocial norms that develop in disasters, norms governing behavior during civil unrest permit and actually encourage lawbreaking. Under these circumstances, otherwise law-abiding citizens allow themselves to take part in looting behavior (Dynes and Quarantelli, 1968; Quarantelli and Dynes, 1970).
Looting and damaging property can also become normative in situations that do not involve civil unrest—for example, in victory celebrations following sports events. Once again, in such cases, norms and traditions governing behavior in crowd celebrations encourage destructive activities
(Rosenfeld, 1997). The behavior of participants in these destructive crowd celebrations again bears no resemblance to that of disaster victims.
In the aftermath of Hurricane Katrina, social scientists had no problem understanding why episodes of looting might have been more widespread in that event than in the vast majority of U.S. disasters. Looting has occurred on a widespread basis following other disasters, although such cases have been rare. Residents of St. Croix engaged in extensive looting behavior following Hurricane Hugo, and this particular episode sheds light on why some Katrina victims might have felt justified in looting. Hurricane Hugo produced massive damage on St. Croix, and government agencies were rendered helpless. Essentially trapped on the island, residents had no idea when help would arrive. Instead, they felt entirely on their own following Hugo. The tourist-based St. Croix economy was characterized by stark social class differences, and crime and corruption had been high prior to the hurricane. Under these circumstances, looting for survival was seen as justified, and patterns of collective behavior developed that were not unlike those seen during episodes of civil unrest. Even law enforcement personnel joined in the looting (Quarantelli, 2006; Rodriguez et al., forthcoming).
Despite their similarities, the parallels between New Orleans and St. Croix should not be overstated. It is now clear that looting and violent behavior were far less common than initially reported and that rumors concerning shootings, rapes, and murders were groundless. The media employed the “looting frame” extensively while downplaying far more numerous examples of selflessness and altruism. In hindsight, it now appears that many reports involving looting and social breakdown were based on stereotyped images of poor minority community residents (Tierney et al., forthcoming).
Extensive research also indicates that despite longstanding evidence, beliefs about disaster-related looting and lawlessness remain quite common, and these beliefs can influence the behavior of both community residents and authorities. For example, those who are at risk may decide not to evacuate and instead stay in their homes to protect their property from looters (Fischer, 1998). Concern regarding looting and lawlessness may cause government officials to make highly questionable and even counterproductive decisions. Following Hurricane Katrina, for example, based largely on rumors and exaggerated media reports, rescue efforts were halted because of fears for the safety of rescue workers, and Louisiana’s governor issued a “shoot-to-kill” order to quash looting. These decisions likely resulted in additional loss of life and also interfered with citizen efforts to aid one another. Interestingly, recent historical accounts indicate that similar decisions were made following other large-scale disasters, such as the 1871 Chicago fire, the 1900 Galveston hurricane, and the 1906 San Francisco earthquake and firestorm. In all three cases, armed force was used to stop
looting, and immigrant groups and the poor were scapegoated for their putative “crimes” (Fradkin, 2005). Along with Katrina, these events caution against making decisions on the basis of mythical beliefs and rumors.
As is the case with the panic myth, attributing the causes of looting behavior to individual motivations and impulses serves to deflect attention from the ways in which institutional failures can create insurmountable problems for disaster victims. When disasters occur, communications, disaster management, and service delivery systems should remain sufficiently robust that victims will not feel isolated and afraid or conclude that needed assistance will never arrive. More to the point, victims of disasters should not be scapegoated when institutions show themselves to be entirely incapable of providing even rudimentary forms of assistance—which was exactly what occurred with respect to Hurricane Katrina.
Patterns of Collective Mobilization in Disaster-Stricken Areas: Prosocial and Helping Behavior
In contrast to the panicky and lawless behavior that is often attributed to disaster-stricken populations, public behavior during earthquakes and other major community emergencies is overwhelmingly adaptive, prosocial, and aimed at promoting the safety of others and the restoration of ongoing community life. The predominance of prosocial behavior (and, conversely, a decline in antisocial behavior) in disaster situations is one of the most longstanding and robust research findings in the disaster literature. Research conducted with NEHRP sponsorship has provided an even better understanding of the processes involved in adaptive collective mobilization during disasters.
Helping Behavior and Disaster Volunteers. Helping behavior in disasters takes various forms, ranging from spontaneous and informal efforts to provide assistance to more organized emergent group activity, and finally to more formalized organizational arrangements. With respect to spontaneously developing and informal helping networks, disaster victims are assisted first by others in the immediate vicinity and surrounding area and only later by official public safety personnel. In a discussion on search and rescue activities following earthquakes, for example, Noji observes (1997:162)
In Southern Italy in 1980, 90 percent of the survivors of an earthquake were extricated by untrained, uninjured survivors who used their bare hands and simple tools such as shovels and axes…. Following the 1976 Tangshan earthquake, about 200,000 to 300,000 entrapped people crawled out of the debris on their own and went on to rescue others…. They became the backbone of the rescue teams, and it was to their credit that more than 80 percent of those buried under the debris were rescued.
Thus, lifesaving efforts in a stricken community rely heavily on the capabilities of relatively uninjured survivors, including untrained volunteers, as well as those of local firefighters and other relevant personnel.
The spontaneous provision of assistance is facilitated by the fact that when crises occur, they take place in the context of ongoing community life and daily routines—that is, they affect not isolated individuals but rather people who are embedded in networks of social relationships. When a massive gasoline explosion destroyed a neighborhood in Guadalajara, Mexico, in 1992, for example, survivors searched for and rescued their loved ones and neighbors. Indeed, they were best suited to do so, because they were the ones who knew who lived in different households and where those individuals probably were at the time of the disaster (Aguirre et al., 1995). Similarly, crowds and gatherings of all types are typically comprised of smaller groupings—couples, families, groups of friends—that become a source of support and aid when emergencies occur.
As the emergency period following a disaster lengthens, unofficial helping behavior begins to take on a more structured form with the development of emergent groups—newly formed entities that become involved in crisis-related activities (Stallings and Quarantelli, 1985; Saunders and Kreps, 1987). Emergent groups perform many different types of activities in disasters, from sandbagging to prevent flooding, to searching for and rescuing victims and providing for other basic needs, to post-disaster cleanup and the informal provision of recovery assistance to victims. Such groupings form both because of the strength of altruistic norms that develop during disasters and because of emerging collective definitions that victims’ needs are not being met—whether official agencies share those views or not. While emergent groups are in many ways essential for the effectiveness of crisis response activities, their activities may be seen as unnecessary or even disruptive by formal crisis response agencies. In the aftermath of the attack on the World Trade Center, for example, numerous groups emerged to offer every conceivable type of assistance to victims and emergency responders. Some were incorporated into official crisis management activities, while others were labeled “rogue volunteers” by official agencies (Halford and Nolan, 2002; Kendra and Wachtendorf, 2002).2
Disaster-related volunteering also takes place within more formalized organizational structures, both in existing organizations that mobilize in response to disasters and through organizations such as the Red Cross,
which has a federal mandate to respond in presidentially declared disasters and relies primarily on volunteers in its provision of disaster services. Some forms of volunteering have been institutionalized in the United States through the development of the National Voluntary Organizations Active in Disaster (NVOAD) organization. NVOAD, a large federation of religious, public service, and other groups, has organizational affiliates in 49 states, the District of Columbia, Puerto Rico, and U.S. territories. National-level NVOAD affiliates include organizations such as the Salvation Army, Church World Service, Church of the Brethren Disaster Response, and dozens of others that provide disaster services. Organizations such as the Red Cross and the NVOAD federation thus provide an infrastructure that can support very extensive volunteer mobilization. That infrastructure will likely form the basis for organized volunteering in future homeland security emergencies, just as it does in major disasters.
Helping behavior is very widespread after disasters, particularly large and damaging ones. For example, NEHRP-sponsored research indicates that in the three weeks following the 1985 earthquake in Mexico City, an estimated 1.7 to 2.1 million residents of that city were involved in providing volunteer aid. Activities in which volunteers engaged after that disaster included searching for and rescuing victims trapped under rubble, donating blood and supplies, inspecting building damage, collecting funds, providing medical care and psychological counseling, and providing food and shelter to victims (Wenger and James, 1994). In other research on post-earthquake volunteering, also funded by NEHRP, O’Brien and Mileti (1992) found that more than half of the population in San Francisco and Santa Cruz counties provided assistance to their fellow victims after the 1989 Loma Prieta earthquake—help that ranged from assisting with search and rescue and debris removal activities to offering food, water, and shelter to those in need. Thus, the volunteer sector responding to disasters typically constitutes a very large proportion of the population of affected regions, as well as volunteers converging from other locations.
Social science research, much of it conducted under NEHRP auspices, highlights a number of other points regarding post-disaster helping behavior. One such insight is that helping behavior in many ways mirrors roles and responsibilities people assume during nondisaster times. For example, when people provide assistance during disasters and other emergencies, their involvement is typically consistent with gender role expectations (Wenger and James, 1994; Feinberg and Johnson, 2001). Research also indicates that mass convergence of volunteers and donations can create significant management problems and undue burdens on disaster-stricken communities. In their eagerness to provide assistance, people may “overrespond” to disaster sites, creating congestion and putting themselves and others at risk or insisting on providing resources that are in fact not needed. After disas-
ters, communities typically experience major difficulties in dealing with unwanted and unneeded donations (Neal, 1990).
Research on public behavior during disasters has major implications for homeland security policies and practices. The research literature provides support for the inclusion of the voluntary sector and community-based organizations in preparedness and response efforts. Initiatives that aim at encouraging public involvement in homeland security efforts of all types are clearly needed. The literature also provides extensive evidence that members of the public are in fact the true “first responders” in major disasters. In using that term to refer to fire, police, and other public safety organizations, current homeland security discourse fails to recognize that community residents themselves constitute the front-line responders in any major emergency
One implication of this line of research is that planning and management models that fail to recognize the role of victims and volunteers in responding to all types of extreme events will leave responders unprepared for what will actually occur during disasters—for example, that, as research consistently shows, community residents will be the first to search for victims, provide emergency aid, and transport victims to health care facilities in emergencies of all types.3 Such plans will also fail to take advantage of the public’s crucial skills, resources, and expertise. For this reason, experts on human-induced threats such as bioterrorism stress the value of public engagement and involvement in planning for homeland security emergencies (Working Group on “Governance Dilemmas” in Bioterrorism Response, 2004).
These research findings have significant policy implications. To date, Department of Homeland Security initiatives have focused almost exclusively on providing equipment and training for uniformed responders, as opposed to community residents. Recently, however, DHS has begun placing more emphasis on its Citizen Corps component, which is designed to mobilize the skills and talents of the public when disasters strike. Public involvement in Citizen Corps and Community Emergency Response Team (CERT) activities have expanded considerably since the terrorist attacks of
9/11—a sign that many community residents around the nation wish to play an active role in responding to future disasters. The need for community-based preparedness and response initiatives is more evident than ever follow-ing the Katrina disaster.
Organizational, Governmental, and Network Responses. The importance of observing disaster response operations while they are ongoing or as soon as possible after disaster impact has long been a hallmark of the disaster research field. The quick-response tradition in disaster research, which has been a part of the field since its inception, developed out of a recognition that data on disaster response activities are perishable and that information collected from organizations after the passage of time is likely to be distorted and incomplete (Quarantelli, 1987, 2002). NEHRP funds, provided through grant supplements, Small Grants for Exploratory Research (SGER) awards, Earthquake Engineering Research Institute (EERI) reconnaissance missions, earthquake center reconnaissance funding, and small grants such as those provided by the Natural Hazards Research and Applications Information Center, have supported the collection of perishable data and enabled social science researchers to mobilize rapidly following major earthquakes and other disasters.
NEHRP provided substantial support for the collection of data on organizational and community responses in a number of earthquake events, including the 1987 Whittier Narrows, 1989 Loma Prieta, and 1994 Northridge earthquakes (see, for example, Tierney, 1988, 1994; EERI, 1995), as well as major earthquakes outside the United States such as the 1985 Mexico City, 1986 San Salvador, and 1988 Armenia events. More recently, NEHRP funds were used to support rapid-response research on the September 11, 2001 terrorist attacks and Hurricanes Katrina and Rita. Many of those studies focused on organizational issues in both the public and private sectors. (For a compilation of NEHRP-sponsored quick-response findings on the events of September 11, see Natural Hazards Research and Applications Information Center, 2003).
In many cases, quick-response research on disaster impacts and organizational and governmental response has led to subsequent in-depth studies on response-related issues identified during the post-impact reconnaissance phase. Following major events such as Loma Prieta, Northridge, and Kobe, insights from initial reconnaissance studies have formed the basis for broader research initiatives. Recent efforts have focused on ways to better take advantage of reconnaissance opportunities and to identify topics for longer-term study. A new plan has been developed to better coordinate and integrate both reconnaissance and longer-term research activities carried out with NEHRP support. That planning activity, outlined in the report The Plan to Coordinate NEHRP Post-earthquake Investigations (Holzer et
al., 2003), encompasses both reconnaissance and more systematic research activities in the earth sciences, engineering, and social sciences.
Through both initial quick-response activities and longer-term studies, NEHRP research has added to the knowledge base on how organizations cope with crises. Studies have focused on a variety of topics. A partial list of those topics includes organizational and group activities associated with the post-disaster search and rescue process (Aguirre et al., 1995); intergovernmental coordination during the response period following major disaster events (Nigg, 1998); expected and improvised organizational forms that characterize the disaster response milieu (Kreps, 1985, 1989b); strategies used by local government organizations to enhance interorganizational coordination following disasters (Drabek, 2003); and response activities undertaken by specific types of organizations, such as those in the volunteer and nonprofit sector (Neal, 1990) and tourism-oriented enterprises (Drabek, 1994).
Focusing specifically at the interorganizational level of analysis, NEHRP research has also highlighted the significance and mix of planned and improvised networks in disaster response. It has long been recognized that post-disaster response activities involve the formation of new (or emergent) networks of organizations. Indeed, one distinguishing feature of major crisis events is the prominence and proliferation of network forms of organization during the response period. Emergent multiorganizational networks (EMON) constitute new organizational interrelationships that reflect collective efforts to manage crisis events. Such networks are typically heterogeneous, consisting of existing organizations with pre-designated crisis management responsibilities, other organizations that may not have been included in prior planning but become involved in crisis response activities because those involved believe they have some contribution to make, and emergent groups. EMONs tend to be very large in major disaster events, encompassing hundreds and even thousands of interacting entities. As crisis conditions change and additional resources converge, EMON structures evolve, new organizations join the network, and new relationships form. What is often incorrectly described as disaster-generated “chaos” is more accurately seen as the understandable confusion that results when mobilization takes place on such a massive scale and when organizations and groups that may be unfamiliar with one another attempt to communicate, negotiate, and coordinate their activities under extreme pressure. (For more detailed discussions on EMONs in disasters, including the 2001 World Trade Center attack, see Drabek, 1985, 2003; Tierney, 2003; Tierney and Trainor, 2004.)
This is not to say that response activities always go smoothly. The disaster literature, organizational after-action reports, and official investigations contain numerous examples of problems that develop as inter-
organizational and intergovernmental networks attempt to address disaster-related challenges. Such problems include the following: failure to recognize the magnitude and seriousness of an event; delayed and insufficient responses; confusion regarding authorities and responsibilities, often resulting in major “turf battles;” resource shortages and misdirection of existing resources; poor organizational, interorganizational, and public communications; failures in intergovernmental coordination; failures in leadership and vision; inequities in the provision of disaster assistance; and organizational practices and cultures that permit and even encourage risky behavior. Hurricane Katrina became a national scandal because of the sheer scale on which these organizational pathologies manifested. However, Katrina was by no means atypical. In one form or another and at varying levels of severity, such pathologies are ever-present in the landscape of disaster response (for examples, see U.S. President’s Commission on the Accident at Three Mile Island, 1979; Perrow, 1984; Shrivastava, 1987; Sagan, 1993; National Academy of Public Administration, 1993; Vaughan, 1996, 1999; Peacock et al., 1997; Klinenberg, 2002; Select Bipartisan Committee to Investigate the Preparations for and Response to Hurricane Katrina, 2006; White House, 2006).
Management Considerations in Disaster Response
U.S. disaster researchers have identified two contrasting approaches to disaster response management, commonly termed the “command-and-control” and the “emergent human resources,” or “problem-solving,” models. The command-and-control model equates preparedness and response activities with military exercises. It assumes that (1) government agencies and other responders must be prepared to take over management and control in disaster situations, both because they are uniquely qualified to do so and because members of the public will be overwhelmed and will likely engage in various types of problematic behavior, such as panic; (2) disaster response activities are best carried out through centralized direction, control, and decision making; and (3) for response activities to be effective, a single person is ideally in charge, and relations among responding entities are arranged hierarchically.
In contrast, the emergent human resources, or problem-solving, model is based on the assumption that communities and societies are resilient and resourceful and that even in areas that are very hard hit by disasters, considerable local response capacity is likely to remain. Another underlying assumption is that preparedness strategies should build on existing community institutions and support systems—for example by pre-identifying existing groups, organizations, and institutions that are capable of assuming leadership when a disaster strikes. Again, this approach argues against
highly specialized approaches that tend to result in “stovepiped” rather than well-integrated preparedness and response efforts. The model also recognizes that when a disaster occurs, responding entities must be flexible if they are to be effective and that flexibility is best achieved through a decentralized response structure that seeks to solve problems as they arise, as opposed to top-down decision making. (For more extensive discussions of these two models and their implications, see Dynes, 1993, 1994; Kreps and Bosworth, forthcoming.)
Empirical research, much of which has been carried out with NEHRP support, finds essentially no support for the command-and-control model either as a heuristic device for conceptualizing the disaster management process or as a strategy employed in actual disasters. Instead, as suggested in the discussion above on EMONs, disaster response activities in the United States correspond much more closely to the emergent resources or problem-solving model. More specifically, such responses are characterized by decentralized, rather than centralized, decision making; by collaborative relationships among organizations and levels of government, rather than hierarchical ones; and, perhaps most important, by considerable emergence—that is, the often rapid appearance of novel and unplanned-for activities, roles, groups, and relationships. Other hallmarks of disaster responses include their fluidity and hence the fast pace at which decisions must be made; the predominance of the EMON as the organizational form most involved in carrying out response activities; the wide array of improvisational strategies that are employed to deal with problems as they manifest themselves; and the importance of local knowledge and situation-specific information in gauging appropriate response strategies. (For empirical research supporting these points, see Drabek et al., 1982; Stallings and Quarantelli, 1985; Kreps, 1985, 1989b; Bosworth and Kreps, 1986; Kreps and Bosworth, 1993; Aguirre et al., 1995; Drabek and McEntire, 2002; Waugh and Sylves, 2002; Webb, 2002; Drabek, 2003; Tierney, 2003; Tierney and Trainor, 2004; Wachtendorf, 2004.)
NEW WAYS OF FRAMING DISASTER MANAGEMENT CHALLENGES: DEALING WITH COMPLEXITY AND ACCOMMODATING EMERGENCE
Advancements brought about through NEHRP research include new frameworks for conceptualizing responses to extreme events. In Shared Risk: Complex Systems in Seismic Response, a NEHRP-supported comparative study of organized responses to 11 different earthquake events, Comfort argues that the major challenge facing response systems is to use information in ways that enhance organizational and interorganizational learning and develop ways of “integrating both technical and organiza-
tional components in a socio-technical system to support timely, informed collective action” (Comfort, 1999:14). Accordingly, effective responses depend on the ability of organizations to simultaneously sustain structure and allow for flexibility in the face of rapidly changing disaster conditions and unexpected demands. Response networks must also be able to accommodate processes of self-organization—that is, organized action by volunteers and emergent groups. This approach again contrasts with command-and-control notions of how major crises are managed (Comfort, 1999:263-264):
A socio-technical approach requires a shift in the conception of response systems as reactive, command-and-control driven systems to one of inquiring systems, activated by processes of inquiry, validation, and creative self-organization…. Combining technical with organizational systems appropriately enables communities to face complex events more effectively by monitoring changing conditions and adapting its performance accordingly, increasing the efficiency of its use of limited resources. It links human capacity to learn with the technical means to support that capacity in complex, dynamic environments [emphasis added].
Similarly, research stressing the importance of EMONs as the predominant organizational form during crisis response periods points to the importance of improving strategies for network management and of developing better methods to take advantage of emergent structures and activities during disasters. Planning and management approaches must, in other words, support rather than interfere with the open and dynamic qualities of disaster response activities. Indicators of improved capacity to manage emergent networks could include the diversity of organizations and community sectors involved in pre-crisis planning; plans and agreements facilitating the incorporation of the voluntary sector and emergent citizen groups into response activities; plans and tools enabling the rapid expansion of crisis communication and information-sharing networks during disasters to include new organizations; and protocols, such as mutual aid agreements, making it possible for new actors to more easily join response networks (Tierney and Trainor, 2004).
In the wake of the Katrina disaster, the need for disaster management by command-and-control-oriented entities has once again achieved prominence. For example, calls have increased for greater involvement on the part of the military in domestic disaster management. Such recommendations are not new. Giving a larger role in disaster management to the military was an idea that was considered—and rejected—following Hurricane Andrew (National Academy of Public Administration, 1993). Post-Katrina debates on needed policy and programmatic changes will likely continue to focus on how to most effectively deploy military assets while ensuring that disaster management remains the responsibility of civilian institutions.
Additional Considerations: Do Responses to Natural, Technological, and Human-Induced Events Differ?
One issue that has come to the fore with the emergence of terrorism as a major threat involves the extent to which findings from the field of disaster research can predict responses to human-induced extreme events. Although some take the position that terrorism and bioterrorism constitute such unique threats that behavioral and organizational responses in such events will differ from what has been documented for other types of extreme events, others contend that this assumption is not borne out by social science disaster research.
The preponderance of evidence seems to suggest that there is more similarity than difference in response behaviors across different types of disaster agents. Regarding the potential for panic, for example, there is no empirical evidence that panic was a problem during the influenza pandemic of 1918, among populations under attack during World War II (Janis, 1951), in catastrophic structure fires and crowd crushes (Johnson, 1987; Johnson et al., 1994; Feinberg and Johnson, 2001), or in the Chernobyl nuclear disaster (Medvedev, 1990). Nor was panic a factor in the 1993 bombing of the World Trade Center (Aguirre et al., 1998), the 1995 Tokyo subway sarin attack (Murakami, 2000), or the terrorist attacks of September 11, 2001 (NIST, 2005; National Commission on Terrorist Attacks upon the United States, 2004). The failure to find significant evidence of panic across a wide range of crisis events is a testimony to the resilience of social relationships and normative practices, even under conditions of extreme peril.
Similarly, as noted earlier, research findings on challenges related to risk communication and warning the public of impending extreme events are also quite consistent across different types of disaster events. For individuals and groups, there are invariably challenges associated with understanding what self-protective actions are required for different types of emergencies, regardless of their origin.
In all types of disasters, organizations must likewise face a common set of challenges associated with situation assessment, the management of primary and secondary impacts, communicating with one another and with the public, and dealing with response-related demands. The need for more effective communication, coordination, planning, and training transcends hazard type. Although recent government initiatives such as the National Response Plan will result in the incorporation of new organizational actors into response systems for extreme events, most of the same local-, state-, and federal-level organizations will still be involved in managing extreme events of all types, employing common management frameworks such as
the Incident Command System and now the National Incident Management System (NIMS).
Social scientific studies on disasters have long shown that general features of extreme events, such as geographic scope and scale, impact severity, and speed of onset, combined with the overall quality of pre-disaster preparedness, have a greater influence on response patterns than do the specific hazard agents that trigger response activities. Regardless of their origins, very large, near-catastrophic, and catastrophic events all place high levels of stress on response systems.
In sum, social science disaster research finds little justification for the notion that individual, group, and community responses to human-induced extreme events, including those triggered by weapons of mass terror, will differ in important ways from those that have been documented in natural and technological disasters. Instead, research highlights the importance of a variety of general factors that affect the quality and effectiveness of responses to disasters, irrespective of the hazard in question. With respect to warning the public and encouraging self-protective action, for example, warning systems must be well designed and warning messages must meet certain criteria for effectiveness, regardless of what type of warning is issued. Members of the public must receive, understand, and personalize warning information; must understand what actions they need to take in order to protect themselves; and must be able to carry out those actions, again regardless of the peril in question. Community residents must feel that they can trust their leaders and community institutions during crises of all types. For organizations, training and exercises and effective mechanisms for interorganizational communication and coordination are critical for community-wide emergencies of all types. When such criteria are not met, response-related problems can be expected regardless of whether the emergency stems from a naturally occurring event, a technological accident, or an intentional act.
Individual and group responses, as well as organizational response challenges, are thus likely to be consistent across different types of crises. At the same time, however, it is clear that there are significant variations in the behavior of responding institutions (as opposed to individuals, groups, and first responders) according to event type. In most technological disasters, along with the need to help those affected, questions of negligence and liability typically come to the fore, and efforts are made to assign blame and make responsible parties accountable. In terrorist events, damaged areas are always treated as crime scenes, and the response involves intense efforts both to care for victims and to identify and capture the perpetrators. Further, although as noted earlier, scapegoating can occur in disasters of all types, the tendency for both institutions and the public to assign blame to
particular groups may be greater in technological and terrorism-related crises than in natural disasters.4
Finally, with respect to responses on the part of the public, even though evidence to the contrary is strong, the idea that some future homeland security emergencies could engender responses different from those observed in past natural, technological, and intentional disasters cannot be ruled out entirely. The concluding section of this chapter highlights the need for further research in this area.
Research on Disaster Recovery
Like hazards and disaster research generally, NERHRP-sponsored research has tended to focus much more on preparedness and response than on either mitigation or disaster recovery. This is especially the case with respect to long-term recovery, a topic that despite its importance has received very little emphasis in the literature. However, even though the topic has not been well studied, NEHRP-funded projects have done a great deal to advance social science understanding of disaster recovery. As discussed later in this section, they have also led to the development of decision tools and guidance that can be used to facilitate the recovery process for affected social units.
It is not an exaggeration to say that prior to NEHRP, relatively little was known about disaster recovery processes and outcomes at different levels of analysis. Researchers had concentrated to some degree on analyzing the impacts of a few earthquakes, such as the 1964 Alaska and 1971 San Fernando events, as well as earthquakes and other major disasters outside the United States. Generally speaking, however, research on recovery was quite sparse. Equally important, earlier research oversimplified the recovery process in a variety of ways. First, there was a tendency to equate recovery, which is a social process, with reconstruction, which involves restoration and replacement of the built environment. Second, there was an assumption that disasters and their impacts proceed in a temporal, stage-like fashion, with “recovery” following once “response” activities have
been concluded.5 Earlier research also underemphasized the extent to which recovery may be experienced differently by different sectors and subpopulations within society. Some of these problems were related to the fact that at a more abstract level, earlier work had not sufficiently explored the concept of recovery itself—for example, whether recovery should be equated with a return to pre-disaster circumstances and social and economic activities, with the creation of a “new normal” that involves some degree of social transformation, or with improvements in community sustainability and long-term disaster loss reduction. Since the inception of NEHRP and in large measure because of NEHRP sponsorship, research has moved in the direction of a more nuanced understanding of recovery processes and outcomes that has not entirely resolved but at least acknowledges many of these issues.
The sections that follow discuss significant contributions to knowledge and practice that have resulted primarily from NEHRP-sponsored work. Those contributions can be seen (somewhat arbitrarily) as falling into four categories: (1) refinements in definitions and conceptions of disaster recovery, along with a critique and reformulation of stage-like models; (2) contributions to the literature on recovery processes and outcomes across different social units; (3) the development of empirically based models to estimate losses, anticipate recovery challenges, and guide decision making; and (4) efforts to link disaster recovery with broader ideas concerning long-term sustainability and environmental management.
Conceptual Clarification. Owing in large measure to NEHRP-sponsored efforts, the disaster field has moved beyond equating recovery with reconstruction or the restoration of the built environment. More usefully, research has moved in the direction of making analytic distinctions among different types of disaster impacts, recovery activities undertaken by and affecting different social units, and recovery outcomes. Although disaster impacts can be positive or negative, research generally tends to focus on various negative impacts occurring at different levels of analysis. As outlined in Chapter 3, these impacts include effects on the physical and built environment, including residential, commercial, and infrastructure damage as well as disaster-induced damage to the environment; other property losses; deaths and injuries; impacts on social and economic activity; effects at the community level, such as impacts on community cohesiveness and urban
form; and psychological, psychosocial, and political impacts. Such impacts can vary in severity and duration, as well as in the extent to which they are addressed effectively during the recovery process. An emphasis on recovery as a multidimensional concept calls attention to the fact that physical and social impacts, recovery trajectories, and short- and longer-term outcomes in chronological and social time can vary considerably across social units.
Recovery activities constitute measures that are intended to remedy negative disaster impacts, restore social units as much as possible to their pre-disaster levels of functioning, enhance resilience, and ideally, realize other objectives such as the mitigation of future disaster losses and improvements in the built environment, quality of life, and long-term sustainability.6 Recovery activities include the provision of temporary and replacement housing; the provision of resources (government aid, insurance payment, private donations) to assist households and businesses with replacement of lost goods and with reconstruction; the provision of various forms of aid and assistance to affected government units; the development and implementation of reconstruction and recovery plans in the aftermath of disasters; coping mechanisms developed by households, businesses, and other affected social units; the provision of mental health and other human services to victims; and other activities designed to overcome negative disaster impacts. In some circumstances, recovery activities can also include the adoption of new policies, legislation, and practices designed to reduce the impacts of future disasters.
Recovery processes are significantly influenced by differential societal and group vulnerability; by variations in the range of recovery aid and support that is available; and by the quality and effectiveness of the help that is provided. The available “mix” of recovery activities and post-disaster coping strategies varies across groups, societies, and different types of disasters. For example, insurance is an important component in the reconstruction and recovery process for some societies, some groups within society, and some types of disasters, but not for others.
Recovery outcomes—or the extent to which the recovery activities are judged, either objectively or subjectively, as “complete” or “successful”—also show wide variation across societies, communities, social units, and disaster events. Outcomes can be assessed in both the short and the longer terms, although, as noted earlier, the literature is weak with respect to empirical studies on the outcomes of longer-term disasters. Additionally,
outcomes consist not only of the intended effects of recovery programs and activities, but also of their unintended consequences. For example, the provision of government assistance or insurance payments to homeowners may make it possible for them to rebuild and continue to live in hazardous areas, even though such an outcome was never intended.
Keeping in mind the multidimensional nature of recovery, post-disaster outcomes can be judged as satisfactory along some dimensions, or at particular points in time, but unsatisfactory along others. Outcomes are perceived and experienced differently, when such factors as level of analysis and specific recovery activities of interest are taken into account. With respect to units of aggregation, for example, while a given disaster may have few discernible long-term effects when analyzed at the community level, the same disaster may well be economically, socially, and psychologically catastrophic for hard-hit households and businesses. A community may be considered “recovered” on the basis of objective social or economic indicators, while constituent social units may not be faring as well, in either objective or subjective terms. The degree to which recovery has taken place is thus very much a matter of perspective and social position.
In a related vein, research has also led to a reconsideration of linear conceptions of the recovery process. Past research tended to see disaster events as progressing from the pre-impact period through post-impact emergency response, and later recovery. In a classic work in this genre—Reconstruction Following Disaster (Haas et al., 1977:xxvi), for example—the authors argued that disaster recovery is “ordered, knowable, and predictable.” Recovery was characterized as consisting of four sequential stages that may overlap to some degree: the emergency period; the restoration period; the replacement reconstruction period; and the commemorative, betterment, and developmental reconstruction period. In this and other studies, the beginning of the recovery phase was generally demarcated by the cessation of immediate life saving and emergency care measures, the resumption of activities of daily life (e.g., opening of schools), and the initiation of rebuilding plans and activities. After a period of time, early recovery activities, such as the provision of temporary housing, would give way to longer-term measures that were meant to be permanent. Kates and Pijawka’s (1977) frequently cited four-phase model begins with the emergency period, lasting for a few days up to a few weeks, and encompassing the period when the emergency operations plan (EOP) is put into operation. Next comes the restoration period—when repairs to utilities are made; debris is removed; evacuees return; and commercial, industrial, and residential structures are repaired. The third phase, the reconstruction replacement period, involves rebuilding capital stocks and getting the economy back to pre-disaster levels. This period can take some years. Finally, there is the development phase, when commemorative structures are built, memo-
rial dates are institutionalized in social time, and attempts are made to improve the community.
In another stage-like model focusing on the community level, Alexander (1993) identified three stages in the process of disaster recovery. First, the rehabilitation stage involves the continuing care of victims and frequently is accompanied by the reemergence of preexisting problems at the household or community level. During the temporary reconstruction stage, prefabricated housing or other temporary structures go up, and temporary bracing may be installed for buildings and bridges. Finally, the permanent reconstruction stage was seen as requiring good administration and management to achieve full community recovery.
Later work sees delineations among disaster phases as much less clear, showing, for example, that decisions and actions that affect recovery may be undertaken as early as the first days or even hours after the disaster’s impact—and, importantly, even before a disaster occurs. The idea that recovery proceeds in an orderly, stage-like, and unitary manner has been replaced by a view that recognizes that the path to recovery is often quite uneven. While the concept of disaster phases may be a useful heuristic device for researchers and practitioners, the concept may also mask both how phases overlap and how recovery proceeds differently for different social groups (Neal, 1997). Recovery does not occur at the same pace for all who are affected by disasters or for all types of impacts. With respect to housing, for example, owing to differences in the availability of services and financing as well as other factors, some groups within a disaster-stricken population may remain in “temporary housing” for a very long time—so long, in fact, that those housing arrangements become permanent—while others may move rapidly into replacement housing (Bolin, 1993a). Put another way, as indicated in Chapters 1 and 3, while stage-like approaches to disasters are framed in terms of chronological time, for those who experience them, disasters unfold in social time.
Researchers studying recovery continue to contend with a legacy of conceptual and measurement difficulties. One such difficulty centers on the question of how the dependent variable should be measured. This problem itself is multifaceted. Should recovery be defined as a return to pre-disaster levels of psychological, social, and economic well-being? As a return to where a community, business, or household would have been were it not for the occurrence of the disaster? The study of disaster recovery also tends to overlap with research on broader processes of social change. Thus, in addition to focusing on what was lost or affected as a consequence of disaster events and on outcomes relative to those impacts, recovery research also focuses on more general post-disaster issues, such as the extent to which disasters influence and interact with ongoing processes of social change, whether disaster impacts can be distinguished from those resulting
from broader social and economic trends, whether disasters simply magnify and accelerate those trends or exert an independent influence, and the extent to which the post-disaster recovery period represents continuity or discontinuity with the past. Seen in this light, the study of recovery can become indistinguishable from the study of longer-term social change affecting communities and societies. While these distinctions are often blurred, it is nevertheless important to differentiate conceptually and empirically between the recovery process, specific recovery outcomes of interest, and the wide range of other changes that might take place following (or as a consequence of) disasters.
Analyzing Impacts and Recovery Across Different Social Units. Following from the discussions above, it is useful to keep in mind several points about research on disaster recovery. First, studies differ in the extent to which they emphasize the objective, physical aspects of recovery—restoration and reconstruction of the built environment—or subjective, psychosocial, and experiential ones. Second, studies generally focus on particular units of analysis and outcomes, such as household, business, economic, or community recovery, rather than on how these different aspects of recovery are interrelated. This is due partly to the fact that researchers tend to specialize in particular types of disaster impacts and aspects of recovery, which has both advantages and disadvantages. While allowing for the development of in-depth research expertise, such specialization has also made it more difficult to formulate more general theories of recovery. Third, the literature is quite uneven. Some aspects of recovery are well understood, while there are others about which very little is known.
Even with these limitations, more general theoretical insights about recovery processes and outcomes have begun to emerge. Key among these is the idea that disaster impacts and recovery can be conceptualized in terms of vulnerability and resilience. As noted in Chapters 2 and 3, vulnerability is a consequence not only of physical location and the “hazardousness of place,” but also of social location and of societal processes that advantage some groups and individuals while marginalizing others. The notion of vulnerability applies both to the likelihood of experiencing negative impacts from disasters, such as being killed or injured or losing one’s home or job, and to the likelihood of experiencing recovery-related difficulties, such as problems with access to services and other forms of support. Social vulnerability is linked to broader trends within society, such as demographic trends (migration to more hazardous areas, the aging of the U.S. population) and population diversity (race, class, income, and linguistic diversity). Similarly, resilience, or the ability to survive and cope with disaster impacts and rebound after those events, is also determined in large measure by social factors. According to Rose (2004), resilience can be conceptualized
as both inherent and adaptive, where the former term refers to resilience that is based on resources and options for action that are typically available during nondisaster times, and the latter refers to the ability to mobilize resources and create new options following disasters.7 As discussed in Chapter 6, resilience stems in part from factors commonly associated with the concept of social capital, such as the extensiveness of social networks, civic engagement, and interpersonal, interorganizational, and institutional trust. (For an influential formulation setting out the vulnerability perspective, see Blaikie et al., 1994). As subsequent discussions show, the concepts of vulnerability and resilience are applicable to individuals, households, groups, organizations, economies, and entire societies affected by disasters. The sections that follow, which are organized according to unit of analysis, discuss psychosocial impacts and recovery; impacts and recovery processes for housing and businesses; economic recovery; and community-level and societal recovery.
Psychological Impacts and Recovery. There is no disagreement among researchers that disasters cause genuine pain and suffering and that they can be deeply distressing for those who experience them. Apart from that consensus, however, there have been many debates and disputes regarding the psychological and psychosocial impacts of disasters. One such debate centers on the extent to which disasters produce clinically significant symptoms of psychological distress and, if so, how long such symptoms last. Researchers have also struggled with the questions of etiology, or the causes of disaster-related psychological reactions. Are such problems the direct result of trauma experienced during disaster, the result of disaster-induced stresses, a reflection of a lack of coping capacity or weak social support networks, a function of preexisting vulnerabilities, or a combination of all these factors? Related concerns center on what constitute appropriate forms of intervention and service delivery strategies for disaster-related psychological problems. Do people who experience problems generally recover on their own, without the need for formally provided assistance, or does such assistance facilitate more rapid and complete recovery? What types of assistance are likely to be most efficacious and for what types of problems?
Research has yielded a wide array of findings on questions involving disaster-related psychological and psychosocial impacts and recovery. Findings tend to differ depending upon disaster type and severity, how disaster victimization is defined and measured, how mental health outcomes are measured, the research methodologies and strategies used (e.g., sampling,
timing, variables of interest), and not inconsequentially, the discipline-based theoretical perspectives employed (Tierney, 2000). With respect to the controversial topic of post-traumatic stress disorder (PTSD), for example, well-designed epidemiological studies have estimated the lifetime prevalence of PTSD at around 5.4 percent in the U.S. population. An important epidemiologic study on the incidence of trauma and the subsequent risk of developing PTSD after various types of traumatic events estimates the risk at about 3.8 percent for natural disasters (Breslau et al., 1998; Kessler and Zhao, 1999). NEHRP-sponsored surveys following recent earthquakes in California found PTSD to be extremely rare among affected populations and not significantly associated with earthquake impacts (Seigel et al., 2000). Other studies show immense variation, with estimates of post-disaster PTSD ranging from very low to greater than 50 percent. Such variations could reflect real differences in the traumatic effects of different events, but it is equally likely that they are the result of methodological, measurement, and theoretical differences among investigators.
One key debate centers on the clinical significance of post-disaster emotional and mental health problems. Research is clear on the point that it is not unusual for disaster victims to experience a series of problems, such as headaches, problems with sleeping and eating, and heightened levels of concern and anxiety, that can vary in severity and duration (Rubonis and Bickman, 1991; Freedy et al., 1994). Perspectives begin to diverge, however, on the extent to which these and other disaster-induced symptoms constitute mental health problems in the clinical sense. In other words, would disaster victims, presenting their symptoms, be considered candidates for mental health counseling or medication if those symptoms were present in a nondisaster context? Do their symptoms correspond to survey based or clinically based measures of what constitutes a “case” for psychiatric diagnostic purposes? Again, as with PTSD, findings differ. While noting that many studies do document a rise in psychological distress following disasters, Shoaf et al. (2004:320) conclude that “those impacts are not of a nature that would significantly increase the rates of diagnosable mental illness.” With respect to severe psychological impacts, these researchers found that suicide rates declined in Los Angeles County following the Northridge earthquake—a continuation of a trend that had already begun before that event. They also note that these findings are consistent with research on suicide following the Kobe earthquake, which showed that the suicide rate in the year following that quake was less than the average rate for the previous 10 years (Shoaf et al., 2004). Yet many researchers and practitioners rightly contend that psychosocial interventions are necessary following disasters, both to address clinically significant symptoms and to prevent more serious psychological sequelae.
There is also the question of whether some types of disasters are more
likely than others to cause negative psychological impacts. Some researchers argue that certain types of technological hazards, such as nuclear threats and chronic exposures to toxic substances, are more pernicious in their effects than natural disasters because they persist longer and create more anxiety among potential victims, and especially because they tend to result in community conflict, causing “corrosive” rather than “therapeutic” communities to develop (Erikson, 1994). Events such as the Oklahoma City bombing, the Columbine school shootings, and the events of September 11, 2001 lead to questions about whether intentional attacks engender psychological reactions that are distinctive and different from those that follow other types of community crisis events. Some studies have suggested that the psychological impacts of terrorist attacks are profound, at least in the short term (North et al., 1999). Other research, focusing specifically on the short-term impacts of the September 11, 2001 terrorist attacks, indicates that the psychological impacts resulting from the events of 9/11 “are consistent with prior estimates of the impact of natural disasters and other terrorist events” (Miller and Heldring, 2004:21). Again, drawing conclusions about the relative influence of agent characteristics—as opposed to other factors—is difficult because studies vary so much in their timing, research designs, methodological approaches, and procedures for defining disaster victimization.
Another set of issues concerns factors associated with risk for poor psychological outcomes. Perilla et al. (2002) suggest that such outcomes can vary as a consequence of both differential exposure and differential vulnerability to extreme events. With respect to differential exposure, factors such as ethnicity and social class can be associated with living in substandard and vulnerable housing, subsequently exposing minorities and poor people to greater losses and disaster-related trauma. Regarding differential vulnerability, minorities and the poor, who are more vulnerable to psychosocial stress during nondisaster times, may also have fewer coping resources upon which to draw following disasters.
In a comprehensive and rigorous review of research on the psychological sequelae of disasters, Fran H. Norris and her colleagues (Norris et al., 2002a,b) carried out a meta-analysis of 20 years of research, based on 160 samples containing more than 60,000 individuals who had experienced 102 different disaster events. These data sets included a range of different types of surveys on both U.S. disaster victims and individuals in other countries, on various subpopulations, and on disasters that differed widely in type and severity. Impacts documented in these studies included symptoms of post-traumatic stress, depression, and anxiety; other forms of nonspecific distress not easily related to specific syndromes such as PTSD; health problems and somatic complaints; problems in living, including secondary stressors such as work-related and financial problems; and “psychosocial resource
loss,” a term that refers to negative effects on coping capacity, self-esteem, feelings of self-efficacy, and other attributes that buffer the effects of stress. According to their interpretation, which was based on accepted methods for rating indicators of psychological distress, the symptoms reported by as many as 39 percent of those studied reached clinically significant levels. However—and this is an important caveat—they found negative psychological effects to be much more prevalent in disasters occurring outside the United States. Generally, symptoms were most severe in the year following disaster events and declined over time.
Norris et al. (2002a, 2002b) classified U.S. disasters as low, moderate, and high in their psychosocial impacts, based on empirical data on post-disaster distress. The Loma Prieta and Northridge earthquakes were seen as having relatively few adverse impacts, and Hurricane Hugo and Three Mile Island were classified as moderate in their effects. Hurricane Andrew, the Exxon oil spill, and the Oklahoma City bombing were classified as severe with respect to their psychological impacts. As these examples suggest, the researchers found no evidence that natural, technological, and human-induced disasters necessarily differ in their effects.
This research review uncovered a number of vulnerability and protective factors that were associated with differential psychological outcomes following disasters. Broadly categorized, those risk factors most consistently shown to be negatively associated with post-disaster psychological well-being include severity of disaster exposure at both the individual and the community levels; being female; being a member of an ethnic minority; low socioeconomic status; experiencing other stressors or chronic stress; having had other mental health problems prior to the disaster; employing inappropriate coping strategies (e.g., withdrawal, avoidance); and reporting problems with both perceived and actual social support.
Overall, these findings are very consistent with perspectives in disaster research that emphasize the relationship between systemically induced vulnerability, negative disaster impacts, lower resilience, and poor recovery outcomes. Recent research situates disasters within the context of other types of stressful events (e.g., death of a loved one or other painful losses) that disproportionately affect those who are most vulnerable and least able to cope. At the same time, studies—many conducted under NEHRP auspices—show how social inequality and vulnerability both amplify the stress that results directly from disasters and complicate the recovery process over the longer term. For example, Fothergill (1996, 1998, 2004) and Enarson and Morrow (1998) have documented the ways in which gender is associated both with the likelihood of becoming a disaster victim and with a variety of subsequent post-disaster stressors. Peacock et al. (1997) and Bolin and Stanford (1998) have shown how pre-disaster conditions such as income disparities and racial and ethnic discrimination contribute both to
disaster losses and to subsequent psychosocial stress and make recovery more difficult for vulnerable groups. Perilla et al. (2002), who studied ethnic differences in post-traumatic stress following Hurricane Andrew, also note that ethnicity can be associated with variations in personality characteristics such as fatalism, which tends to be associated with poor psychosocial outcomes resulting from stressful events, as well as with additional stresses associated with acculturation.8
Hurricane Katrina represents a critical test case for theories and research on psychosocial vulnerability and resilience. If, as Norris and her collaborators indicate, Hurricane Andrew resulted in relatively high levels of psychosocial distress, what will researchers find with respect to Katrina? For many victims, Katrina appears to contain all of the ingredients necessary to produce negative mental health outcomes: massive, catastrophic impacts; high property losses resulting in financial distress; exposure to traumas such as prolonged physical stress and contact with dead and dying victims; disruption of social networks; massive failures in service delivery systems; continual uncertainty about the future; and residential dislocation on a scale never seen in a U.S. disaster. Over time, research will result in important insights regarding the psychosocial dimensions of truly catastrophic disaster events.
Household Impacts and Recovery. Within the disaster recovery area, households and household recovery have been studied most often, with a significant proportion of that work focusing on post-earthquake recovery issues. Although this line of research predates NEHRP, many later studies have been undertaken with NEHRP support. Studies conducted prior to NEHRP include Bolin’s research on household recovery processes following the Managua earthquake and the Rapid City flood, both of which occurred in 1972 (Bolin, 1976). Drabek and Key and their collaborators had also examined disaster impacts on families and the household recover process (Drabek et al., 1975; Drabek and Key, 1976, 1984). With NEHRP support, Bolin and Bolton studied household recovery following tornadoes in Wichita Falls, Vernon, and Paris, Texas; a hurricane in Hawaii; flooding in Salt Lake City; and the Coalinga earthquake (Bolin, 1982; Bolin and Bolton, 1986). Bolin’s monograph Household and Community Recovery after
Earthquakes was based on research on the 1987 Whittier Narrows and 1989 Loma Prieta events (Bolin, 1993b). Households have also been the focus of more recent studies on the impacts of Hurricane Andrew (Peacock et al., 1997) and the 1994 Northridge earthquake (Bolin and Stanford, 1998). Other NEHRP-sponsored work has focused more specifically on issues that are important for household recovery, such as post-disaster sheltering processes (Phillips, 1993, 1998) and housing impacts and recovery (Comerio, 1997, 1998). As Bolin (1993a:13) observes
[d]isasters can have a multiplicity of effects on a household, including physical losses to property, injury and/or death, loss of job or livelihood, disruption of social and personal relations, relocation of some or all members of a family, physical disruption or transformation of community and neighborhood, and increased household indebtedness.
Accordingly, the literature has explored various dimensions of household impacts and recovery, including direct impacts such as those highlighted by Bolin; changes in the quality and cohesiveness of relationships among household members; post-disaster problems such as conflict and domestic violence; stressors that affect households during the recovery process; and coping strategies employed by households, including the use of both formal and informal sources of post-disaster support and recovery aid.
The literature also points to a number of factors that are associated with differences in short- and longer-term household recovery outcomes. Housing supply is one such factor—as indicated, for example, by housing costs, other real estate market characteristics, and rental vacancy rates Temporary housing options are affected by such factors as the proximity of friends and relatives with whom to stay, although use of this housing option is generally only a short-term strategy. Extended family members may not be able to help if they also are victims (Morrow, 1997). Such problems may be more prevalent in lower-income groups that have few alternative resources and when most members of an extended family live in the same affected community.
Availability of temporary and permanent housing generally is limited by their pre-impact supply in and near the impact area. In the U.S., in situations in which there is an insufficient supply of housing for displaced disaster victims, FEMA provides mobile homes, but even this expedient method of expanding the housing stock takes time. Even when houses are only moderately damaged, loss of housing functionality may be a problem if there is massive disruption of infrastructure. In such cases, tent cities may be necessary if undamaged housing is beyond commuting range (e.g., Homestead, Florida after Hurricane Andrew, as discussed in Peacock et al., 1997).
In the longer term, household recovery is influenced by such factors as household financial resources, the ability to obtain assistance from friends and relatives, insurance coverage, and the mix of housing assistance pro-
grams available to households. Typically, access to and adequacy of recovery resources are inversely related to socioeconomic status. Those with higher incomes are more likely to own their own homes, to be adequately insured, and to have savings and other financial resources on which to draw in order to recover—although disasters can also cause even better-off households to take on additional debt. With respect to formal sources of aid, the assistance process generally favors those who are adept at responding to bureaucratic requirements and who are able to invest time and effort to seek out sources of aid. The aid process also favors those living in more conventional, nuclear family living arrangements, as opposed to extended families or multiple households occupying the same dwelling unit (Morrow, 1997). Recovery may be particularly difficult for single-parent households, especially those headed by women (Enarson and Morrow, 1998; Fothergill, 2004).
The picture that emerges from research on household recovery is not that of a predictable and stage-like process that is common to all households, but rather of a multiplicity of recovery trajectories that are shaped not only by the physical impacts of disaster but also by axes of stratification that include income, race, and ethnicity, as well as such factors as the availability of and access to different forms of monetary aid, other types of assistance, and informal social support—which are themselves associated with stratification and diversity. Disaster severity matters, both because disasters that produce major and widespread impacts can limit recovery options for households and because they tend to be more damaging to the social fabric of the community. As Comerio’s extensive research on housing impacts and issues following earthquakes and other disasters in different societal contexts illustrates, household recovery processes are also shaped by societal-level policy and institutional factors—which themselves have differential impacts (Comerio, 1998).9
Large-Scale Comparative Research on Household Recovery. Although there is clearly a need for such research, few studies exist that compare household recovery processes and outcomes across communities and disaster events. With NEHRP funding, Frederick Bates and his colleagues carried out what may well be the largest research efforts of this kind: a multicommunity
longitudinal study on household and community impacts and recovery after the 1976 Guatemala earthquake and a cross-national comparative study on household recovery following six different disaster events. The Guatemala study, designed as a quasi-experiment, included households in 26 communities that were carefully selected to reflect differences in the severity of earthquake impacts, size, population composition, and region of the country. That study focused on a broad spectrum of topics, including changes over time in household composition and characteristics; household economic activity; housing characteristics and standards of living; household experiences with relief and reconstruction assistance; and fertility, health, and nutrition. Never replicated for any other type of disaster, the study provided detailed information on these topics, focusing in particular on how different forms of aid provision either facilitated or hampered household recovery (for detailed discussions, see Bates, 1982; Hoover and Bates, 1985; Bates et al., 1979).
The second study carried out by Bates and his colleagues extended methods developed to assess household recovery following the Guatemala earthquake to measure household recovery in disaster-stricken communities in six different countries. The tool used to measure disaster impacts and household recovery across different events and societies, the Domestic Assets Scale, made possible systematic comparisons with respect to one dimension of household recovery—the restoration of household possessions, tools, and technologies (Bates and Peacock, 1992, 1993).
Vulnerability, Resilience, and Household Recovery. Like the other aspects of recovery discussed here, what happens to households during and after disasters can be conceptualized in terms of vulnerability and resilience. With respect to vulnerability, social location is associated with the severity of disaster impacts for households. Poverty often forces people to live in substandard or highly vulnerable housing—manufactured housing is one example—leaving them more vulnerable to death, injury, and homelessness. As discussed in Chapter 3 with respect to disaster preparedness, factors such as income, education, and homeownership influence the ability of households to mitigate and prepare for disasters. Social-structural factors also affect the extent to which families can accumulate assets in order to achieve higher levels of safety, as well as their recovery options and access to resources after disasters strike—for example the forms of recovery assistance for which they are eligible. Households are thus differentially exposed to disasters, differentially vulnerable during the recovery period, and diverse in terms of both inherent and adaptive resilience.
ECONOMIC AND BUSINESS IMPACTS AND RECOVERY: THE CHALLENGE OF ASSESSING DISASTER LOSSES
As discussed in Chapter 3, assessing how much disasters cost the nation and its communities has proven to be a major challenge. A National Research Council (NRC, 1999c) study concluded that such calculations are difficult in part because different agencies and entities calculate costs and losses differently. Moreover, no universally accepted standards exist for calculating economic impacts resulting from disasters, and there is no single agency responsible for keeping track of disaster losses. For any given disaster event, assessments of economic impacts may vary widely depending on which statistics are used—for example, direct or insured losses versus total losses.
NEHRP-sponsored research has addressed these problems to some degree. For example, as part of the NEHRP-sponsored “Second Assessment of Research on Natural Hazards,” researchers attempted to estimates losses, costs, and other impacts from a wide array of natural and technological hazards.10 For the 20 year period 1975–1994, they estimated that dollar losses from disasters amounted to $.5 billion per week, with climatological hazards accounting for about 80 percent of those losses; since 1989, losses have totaled $1 billion per week (Mileti, 1999a). Through work undertaken as part of the Second Assessment, data on losses from natural hazard events from the mid-1970s to 2000 are now available at the county level in geocoded form for the entire United States through the Spatial Hazard Events and Losses Database for the United States (SHELDUS). This data collection and database development effort has made it possible to analyze different types of losses, at different scales, using different metrics, and to assess locations in terms of their hazard proneness and loss histories. (For discussions of the data used in the SHELDUS database and associated challenges see Cutter, 2001.) What is still lacking is a national program to continue systematically collecting and analyzing impact and loss data.
Studies on economic impacts and recovery from earthquakes and other disasters can be classified according to the units of analysis on which they focus. Most research concerns economic losses and recovery at the community or, more frequently, the regional level. A smaller set of studies has analyzed economic impacts and recovery at the firm or facility level. There is even less research documenting national-level and macroeconomic impacts.
Community-Level and Regional Studies
Studies on the economics of natural disasters at the community and regional levels of analysis differ significantly in methods, topics of interest, and conclusions. Some researchers, such as Rossi et al. (1978) and Friesema et al. (1979) have argued that at least in the United States, natural disasters have no discernible social or economic effects at the community level and that nondisaster-related trends have a far more significant influence on long-term outcomes than disasters themselves. This position has also been argued at the macroeconomic level, with respect to other developed and developing countries (Albala-Bertrand, 1993).11 Dacy and Kunreuther (1969:168) even argued (although more than 30 years ago) that “a disaster may actually turn out to be a blessing in disguise” because disasters create reconstruction booms and allow community improvements to be made rapidly, rather than gradually. However, most research contradicts the idea that disasters constitute economic windfalls, emphasizing instead that economic gains that may be realized at one level (e.g., the community, particular economic sectors) typically constitute losses at another (e.g., the national tax base). One analyst has called the idea that disasters are beneficial economically “one of the most widely held misbeliefs in economics” (DeVoe, 1997:188).
Other researchers take the position that post-disaster economic and social conditions are generally consistent with pre-disaster trends, although disasters may amplify those changes (Bates and Peacock, 1993). Disasters may further marginalize firms and sectors of the economy that were already in decline, or they may speed up processes that were already under way prior to their occurrence. For example, Homestead Air Force Base was already slated for closure before Hurricane Andrew despite ongoing efforts to keep the base opened. When Andrew occurred, the base sustained damage and was closed for good. The closure affected businesses that had depended on the base and helped lead to the exodus of many middle-class families from the area, which in turn affected tax revenues in the impact region. These changes would have taken place eventually, but they were accelerated by Hurricane Andrew.
Related research has analyzed the distributive effects of earthquakes and other disasters. In an early formulation, Cochrane (1975) observed that lower-income groups consistently bear a disproportionate share of disaster losses, relative to higher-income groups. This theme continues to be promi-
nent in the disaster literature; the notion that disasters create economic “winners and losers” has been borne out for both households and businesses (Peacock et al., 1997:Chapter 11; Tierney and Webb, forthcoming).
Another prominent research emphasis at the community and regional levels of analysis has grown out of the need to characterize and quantify the economic impacts of disasters (as well as other impacts) in order to be better able to plan for and mitigate those impacts. A considerable amount of NEHRP research on economic impacts and recovery has been driven by concern about the potentially severe economic consequences of major earthquakes, particularly those that could occur in highly populated urban areas. That concern is reflected in a number of NRC reports (1989, 1992, 1999c) on projected losses and potential economic impacts. Within the private sector, the insurance industry has also committed significant resources in an effort to better anticipate the magnitude of insured losses in future disaster events. (For new developments in research on the management of catastrophic insurance risk, see Grossi et al., 2004.)
Stimulated in large measure by NEHRP funding, new tools have been developed for both pre-disaster estimation of potential losses and post-disaster impact assessments, particularly for earthquakes. HAZUS, the national loss estimation methodology, which was originally developed for earthquakes and which has now been extended to flood and wind hazards, was formulated under FEMA’s supervision with NEHRP funding. NEHRP funds have also supported the development of newer and more sophisticated modeling approaches through research undertaken at earthquake centers sponsored by the National Science Foundation (NSF).
The framework for estimating losses from natural hazards was initially laid out more than 20 years ago in publications such as Petak and Atkisson’s Natural Hazard Risk Assessment and Public Policy (1982) and in applied studies such as the PEPPER (Pre-Earthquake Planning for Post-Earthquake Rebuilding) project (Spangle, 1987), which analyzed potential earthquake impacts and post-disaster recovery strategies for Los Angeles. According to the logic developed in these and other early studies (see, for example, NRC, 1989) and later through extensive NEHRP research, loss estimation consists of the analysis of scenario or probabilistic models that include data on hazards; exposures, or characteristics of the built environment at risk, including buildings and infrastructural systems; fragilities, or estimates of damage likelihood as a function of one or more parameters, such as earthquake shaking intensity; direct losses, such as deaths, injuries, and costs associated with damage; and indirect losses and ripple effects that result from disasters. Within this framework, recent research has focused on further refining loss models and reducing uncertainties associated with both the components of loss estimation models and their interrelationships (for
representative work, see theme issue in Earthquake Spectra, 1997; Tierney et al., 1999; Okuyama and Chang, 2004).
This line of research has led both to advances in basic science knowledge and to a wide range of research applications. At the basic science level, loss modeling research—particularly studies supported through NEHRP—has helped distinguish and clarify relationships among such factors as physical damage, direct economic loss, business interruption effects, and indirect losses and ripple effects. For example, it is now more possible than ever before to disaggregate and analyze separately different types of economic effects and to understand how particular types of damage (e.g., damage to electrical power or transportation systems) contribute to overall economic losses. This research has shed light on factors that contribute to the resilience of regional economies, both during normal times and in response to sudden shocks. It has also shown how the application of newer economic modeling techniques, such as computable general equilibrium modeling and agent-based modeling, constitute improvements over more traditional input-output modeling, particularly for the study of extreme events (for discussions, see Rose et al., 2004; Chang, 2005; Rose and Liao, 2005). Econometric modeling provides another promising approach at both the micro and the regional levels (see West and Lenze, 1994), but this potential remains largely untapped.
At the applications level, loss estimation tools and products have proven useful for raising public awareness of the likely impacts of disaster events and for enhancing community preparedness efforts and mitigation programs. They have also made it possible to assess mitigation alternatives, not only in light of the extent to which those measures reduce damage, but also in terms of their economic costs and benefits. When applied in the disaster context, rapid economic loss estimates have also formed the basis for requests for federal disaster assistance. For the insurance industry, loss models provide important tools to improve risk management decision making, particularly with regard to catastrophic risks.
As noted earlier, loss modeling originally was driven by the need to better understand the economic impacts of earthquakes. In addition to economic losses, earthquake loss models are increasingly taking into account other societal impacts such as deaths, injuries, and residential displacement, as well as secondary effects such as earthquake-induced fires. The methodological approach developed to study earthquakes was first extended to other natural hazards and is now being used increasingly to assess potential impacts from terrorism. The nation is now better able to address the issue of terrorism-related losses because of the investments that had been made earlier for earthquakes and other natural hazards. Significantly, when the Department of Homeland Security decided in 2003 to begin funding
university-based “centers of excellence” for terrorism research, the first topic that was selected for funding was risk and economic modeling for terrorist attacks in the United States.12 Many of the investigators associated with that center had previously worked on loss modeling for earthquakes.
Business and Facility-Level Impacts and Recovery. Most research on recovery processes and outcomes has focused on households and communities. Prior to the 1990s, most research on the economic aspects of disasters focused not on individual businesses but rather on community-wide and regional impacts. Almost nothing was known about how private sector organizations are affected by and recover from disasters. Since then, a small number of studies have focused on business firms or, in some cases, commercial facilities, as units of analysis. Much of this work, including studies on large, representative samples of businesses, has been carried out with NEHRP support. Business impacts and recovery have been assessed following the Whittier Narrows, Loma Prieta, Northridge, and Kobe earthquakes; the 1993 Midwest floods; Hurricane Andrew; and other flood and hurricane events (for representative studies and findings, see Dahlhamer, 1998; Chang, 2000; Webb et al., 2000; Alesch et al., 2001). Long-term business recovery has been studied in the context of only two disaster events—the Loma Prieta earthquake and Hurricane Andrew (Webb et al., 2003).
These studies have shown that disasters disrupt business operations through a variety of mechanisms. Direct physical damage to buildings, equipment, vehicles, and inventories has obvious effects on business operation. It might be less obvious that disruption of infrastructure such as water/sewer, electric power, fuel (i.e., natural gas), transportation, and telecommunications frequently forces businesses to shut down in the aftermath of a disaster (Alesch et al., 1993; Tierney and Nigg, 1995; Tierney, 1997a, b; Webb et al., 2000). For example, Tierney (1997b) reported that extensive electrical power service interruption after the 1993 Midwest floods caused a large number of business closures in Des Moines, Iowa, even though the physical damage was confined to a relatively small area.
Other negative disaster effects include population dislocation, losses in discretionary income among those victims who remain in the impact area—which can weaken market demand for many products and services—and competitive pressure from large outside businesses. These kinds of impacts can cause small local businesses to experience major difficulties recovering from the aftermath of a disaster (Alesch et al., 2001). Indeed, such factors
can produce business failures long after the precipitating event, especially if the community was already in economic decline before the disaster occurred (Bates and Peacock, 1993; Webb et al., 2003).
It is difficult to generalize on the basis of so few studies, particularly when the issues involved and the methodological challenges are so complex. However, studies to date have uncovered a few consistent patterns with respect to business impacts and recovery. First, studies show that most businesses do recover, and do so relatively quickly. In other words, typical businesses affected by disasters show a good deal of resilience in the face of major disruption.
Second, some businesses do tend to fare worse than others in the aftermath of disasters; clearly, not all businesses are equally vulnerable or equally resilient. Although findings from individual studies differ, the factors that seem to contribute most to vulnerability include small size; poor pre-disaster financial condition; business type, with wholesale and retail trade appearing to be especially vulnerable, while manufacturing and construction businesses stand to benefit most from disasters; and severity of disaster impacts. With regard to this last-mentioned factor, studies show that negative impacts on businesses include not only direct physical damage, lifeline-related problems, and business interruption, but also more long-lasting operational problems that businesses may experience following disasters, such as employee absenteeism and loss of productivity, earthquake-induced declines in demands for goods and services, and difficulties with shipping or receiving products and supplies.
Third, business recovery is affected by many factors that are outside the control of the individual business owner. For example, businesses located in highly damaged areas may experience recovery difficulties independent of whether or not they experience losses. In this case, recovery is complicated by the fact that disasters disrupt local ecologies on which individual businesses depend. Business recovery processes and outcomes are also linked to community-level decision making. After the Loma Prieta earthquake, for example, the City of Santa Cruz offered extensive support to businesses and used the earthquake as an opportunity to reinvent itself and to revitalize a business district that had fallen short of realizing its potential prior to the disaster (Arnold, 1998). Actions that communities take with respect to land-use, structural mitigation, infrastructure protection, community education, and emergency response planning also affect how businesses and business districts fare during and after disasters.
Fourth, recovery outcomes following disasters are linked to pre-disaster trends and broader market forces. For example, focusing on an important transport facility, the Port of Kobe, Chang (2000) showed that the port’s inability to recover fully after the 1995 earthquake was due in part to losses in one part of the port’s business—trans-shipment cargo—that had already
been declining before the earthquake owing to severe competition from other ports in the region. Similarly, Dahlhamer (1998) found that businesses in the wholesale and retail trade sectors were more vulnerable to experiencing negative economic outcomes following the Northridge earthquake, perhaps because they constitute crowded and highly competitive economic niches and because turnover is high in those sectors during normal times. He also found that firms in industries that had been experiencing growth in the two-year period just before the earthquake were less likely than firms in declining industries to report being worse off following the Northridge event. Such findings are consistent with a more general theme in recovery research discussed earlier—that disasters do not generate change in and of themselves, but rather intensify or accelerate preexisting patterns.
Community Recovery. Although the topic of community recovery is still not well studied, significant progress has been made in understanding both recovery processes and factors that are associated with recovery outcomes for communities. Earlier research indicated that communities rebound well from disasters and that, at the aggregate level and net of other factors, the impacts of disasters are negligible (Friesema et al., 1979; Wright et al., 1979). However, other more recent research suggests that such findings paint an overly simplified and perhaps overly optimistic picture of post-disaster recovery. This may have been due to methodological shortcomings—for example, the tendency to aggregate data and to group together both more damaging disasters and those that did comparatively little damage—or because such studies were based on “typical” disasters in the United States, rather than catastrophic or near-catastrophic ones.13 In contrast, in a methodologically sophisticated study focusing on a much more severe disaster, the 1995 Kobe event, Chang (2001) analyzed a number of recovery indicators, including measures of economic activity, employment in manufacturing, changes in the spatial distribution of work activities, and differences in recovery indicators among different districts within the city. She found that the earthquake did have lasting and significant negative effects on the City of Kobe. Equally important, poor recovery outcomes were more pronounced in some parts of the city than in others—specifically those areas that had already been experiencing declines. This study provides yet another illustration of how disasters exploit existing vulnerabilities. It also cautions against making blanket statements about disaster impacts and recovery.
Another limitation of earlier work on community recovery was that it provided too little information on what actually happens in communities during the recovery process or what communities can do to ensure more rapid and satisfactory recovery outcomes. Later research, much of which has been undertaken with NEHRP support, has addressed these issues. For example, in Community Recovery from a Major Natural Disaster, Rubin et al. (1985) developed a set of propositions regarding factors that affect community recovery outcomes. That monograph, which was based on case study analyses of recovery following 14 disasters that occurred in the early 1980s, emphasized the importance of three general constructs—personal leadership, knowledge of appropriate recovery actions, and ability to act—as well as the influence of intergovernmental (state and federal) policies and programs. This work highlighted the effects of both government decision making and broader societal policies on community recovery.
Some more recent research has more explicitly incorporated community and population vulnerability as factors affecting community-level recovery. Bolin and Stanford (1998) traced how the post-Northridge recovery experiences of Los Angeles and smaller outlying towns differed as a function of such factors as political expertise and influence, preexisting plans, institutional capacity, involvement of community organizations, and interest group competition. In these diverse communities, the needs of more vulnerable and marginalized groups were sometimes addressed during the recovery process. However, recovery programs ultimately did little to improve the safety of those groups, because they failed to address the root causes of vulnerability (Bolin and Stanford, 1998:216):
[s]ince vulnerability derives from political, economic, and social processes that deny certain people and groups access or entitlements to incomes, housing, health care, political rights, and, in some cases, even food, then post-disaster rebuilding by itself will have little effect on vulnerability.
Societal-Level and Comparative Research on Disaster Recovery. International research on disasters is discussed in greater detail in Chapter 6. This chapter focuses in a more limited way on what little research exists on disaster impacts and post-disaster change at the societal level. Regarding long-term societal impacts, researchers have generally found that disasters, even very large ones, typically do not in and of themselves result in significant change in the societies they affect. Instead, the broad consensus has been that to the extent disasters do have lasting effects, it is because they interact with other factors to accelerate changes that were already under way. Albala-Bertrand, for example has argued that while disasters can highlight preexisting political conflicts, whether such effects are sustained over time “has little to do with the disaster itself, but with preexisting economic and sociopolitical
conditions” (1993:197). This research found that the potential for such changes was generally greater in developing countries than developed ones, although not great in any case.
With respect to the political impacts of disasters at the societal level, comparing very large disasters that occurred between 1966 and 1980, political scientist Richard Olson found that that major disasters can result in higher levels of political unrest, particularly in developing countries that are already politically unstable (Olson and Drury, 1997). In other research, Olson argues that under certain (and rare) circumstances, disasters can constitute “critical junctures,” or crises that leave distinctive legacies within those societies. The 1972 earthquake in Managua, Nicaragua, was one such case. Following that devastating event, the corrupt and dictatorial Somoza regime took a large share of post-disaster aid for itself and mismanaged the recovery, in the process alienating Nicaraguan elites, the business establishment, and finally the middle class, and paving the way for the Sandanistas to assume power in 1979. The 1985 Mexico City earthquake also affected the political system of that nation by, among other things, helping to weaken the hegemony of the Institutional Revolutionary Party. However, rather than having a direct and independent influence on subsequent political changes, that earthquake interacted with factors and trends that were already beginning to affect Mexican society before it occurred. That disaster, which was not well managed by the ruling government, provided the Mexican people with a sharp contrast between the vibrancy and the capability of civil society and the government’s lack of preparedness. Grass-roots response and recovery efforts also facilitated broader mobilization by groups that had been pressing for change. Although not a “critical juncture” in its own right, the earthquake did play a role in moving the political system in the direction of greater pluralism and strengthened the power of civil society institutions vis-à-vis the state (Olson and Gawronski, 2003).
Such findings assume particular significance in the aftermath of the December 2004 Indian Ocean earthquake and tsunami. The impacts of that catastrophe span at least 12 different nations and a number of semi-autonomous subnational units, each with its own distinctive history, mode of political organization, internal cleavages, and preexisting problems. Research is needed to better understand both recovery processes and outcomes and the longer-term societal effects of this devastating event.
OTHER DISASTER RECOVERY-RELATED ISSUES
Disaster Experience and the Mitigation of Future Hazards
Social science research has also focused in various ways on the question of whether the positive informational effects of disasters constitute learning
experiences for affected social units by encouraging the adoption of mitigation measures and stimulating preparedness activity. While this idea seems intuitively appealing, the literature is in fact quite equivocal with regard to the extent to which disasters actually promote higher levels of safety. On the one hand, at the community and societal levels, there is considerable evidence to suggest that disasters constitute “windows of opportunity” for those seeking to enact loss reduction programs, making it possible to achieve policy victories that would not have been possible prior to those events (Alesch and Petak, 1986). Disasters have the potential to become “focusing events” (Birkland, 1997) that can alter policy agendas through highlighting areas in which current policy has failed, energizing advocates, and raising public awareness. On the other hand, many disasters fail to become focusing events and have no discernible impacts on the adoption and implementation of loss-reduction measures. For example, Burby et al., (1997), who studied communities in five different states, found no relationship between disaster experience and adoption of mitigation measures. Birkland (1997) suggests that these differences are related in part to the extent to which advocacy coalitions exist, are able to turn disaster events to their advantage, and are able to formulate appropriate policy responses.
Further complicating matters, policies adopted in the aftermath of disasters, like other policies, may meet with resistance and be only partially implemented—or implemented in ways that were never intended. While it is possible to point to examples of successful policy adoption and implementation in the aftermath of disasters, such outcomes are by no means inevitable, and when they do occur, they are typically traceable to other factors, not just to disaster events themselves.
Research does suggest that households, businesses, and other entities affected by disasters learn from their experiences and take action to protect themselves from future events. Those who have experienced disasters may, for example, step up their preparedness for future events or be more likely to heed subsequent disaster warnings. At the same time, it is also clear that there is considerable variability in the relationship between experience and behavioral change. While some studies document the positive informational effects of experience, others show no significant impact, and some research even indicates that repeated experiences engender complacency and lack of action (for a review of the literature, see Tierney et al., 2001).
Role of Prices and Markets
Mainstream economic theory, models, and analytical tools (e.g., benefit-cost analysis) assume that markets generally function efficiently and equilibrate. Barring various situations of market failure, prices serve a key role as signals of resource scarcity. In this context, two broad areas of research
needs can be identified. One is the role of prices and markets in pre-disaster mitigation (see also Chapter 3). Market-based approaches to reducing disaster risk involve such questions as how prices can serve as better signals of risk taking and risk protection, and the potential for new approaches to risk sharing (e.g., catastrophe bonds). At the same time, better understanding is also needed of market failures in mitigation (e.g., externalities in risk taking and risk protection). The second broad research need concerns markets in post-disaster loss and recovery. Little empirical research has been conducted on the degree to which assumptions of efficient markets actually hold in disasters, especially those having catastrophic impacts, and the degree to which markets are resilient in the face of disasters. Research is also needed on how economic models can capture the adjustment processes and disequilibria that are important as economies recover from disasters, and how economic recovery policies can influence recovery trajectories.
Disaster Recovery and Sustainability
As discussed in more detail in Chapter 6, which focuses on international research, disaster theory and research have increasingly emphasized the extent to which vulnerability to disasters can be linked to unsustainable development practices. Indeed, the connection between disaster loss reduction and sustainability was a key organizing principle of the NEHRP-sponsored Second Assessment of Research on Natural Hazards. The title of the summary volume for the Second Assessment, Disasters by Design (Mileti, 1999b), was chosen to emphasize the idea that the impacts produced by disasters are the consequence of prior decisions that put people and property at risk. A key organizing assumption for the Second Assessment was the notion that societies and communities “design” the disasters of the future by failing to take hazards into account in development decisions; pursuing other values, such as rapid economic growth, at the expense of safety; failing to take decisive action to mitigate risks to the built environment; and ignoring opportunities to enhance social and economic resilience in the face of disasters. Conversely, communities and societies also have the ability to design safer futures by better integrating hazard reduction into their ongoing policies and practices in areas such as land-use and development planning, building codes and code enforcement, and quality-of-life initiatives.
Just as disasters dramatically highlight failures to address sources of vulnerability, the post-disaster recovery period gives affected communities and societies an opportunity to reassess pre-disaster plans, policies, and programs, remedy their shortcomings, and design a safer future (Berke et al., 1993). The federal government seeks to promote post-disaster mitigation through FEMA’s Hazard Mitigation Grant Program, as well as programs
that seek to reduce repetitive flood losses through relocating flood-prone properties. The need to weave a concern with disaster loss reduction into the fabric of ongoing community life has also guided federal initiatives such as Project Impact, FEMA’s Disaster Resistant Communities program.
Yet the research record suggests that those opportunities are often missed. While it is clear that some disaster-stricken communities do act decisively to reduce future losses, for others the recovery period brings about a return to the status quo ante, marked at most by gains in safety afforded by reconstruction to more stringent building codes. The section above noted that disasters create “windows of opportunity” for loss reduction advocates, in part by highlighting policy failures and temporarily silencing opponents. At the same time, however, research evidence suggests that even under those circumstances, it is extremely difficult to advance sustainability goals in the aftermath of disasters. Changes in land use are particularly difficult to enact, both during nondisaster times and after disasters, despite the fact that such changes can significantly reduce vulnerability. Land use decision making generally occurs at the local level, but local jurisdictions have great difficulty enacting controls on development in the absence of enabling legislation from higher levels of government. Even when land-use and zoning changes and other mitigation measures are seen as desirable following disasters, community leaders may lack the political will to promote such efforts over the long term, allowing opponents to regroup and old patterns to reassert themselves (see, for example, Reddy, 2000; for more detailed discussions on land-use and hazards, see Burby, 1998). Assessing reconstruction following recent U.S. disasters, Platt (1998:51) observed that “[d]espite all the emphasis on mitigation of multiple hazards in recent years, political, social and economic forces conspire to promote rebuilding patterns that set the stage for future catastrophe.” Overall, the research record suggests that while the recovery period should ideally be a time when communities take stock of their loss reduction policies and enact new ones, post-disaster change tends to be incremental at best and post-disaster efforts to promote sustainability are rare.
This chapter closes by making recommendations for future research on disaster response and recovery. As the foregoing discussions have indicated, existing research has raised numerous questions that need to be addressed through future research. This concluding section highlights general areas in which new research is clearly needed, both to test the limits of current social science knowledge and to take into account broad societal changes and issues of disaster severity and scale.
Recommendation 4.1: Future research should focus on further empirical explorations of societal vulnerability and resilience to natural, technological, and willfully caused hazards and disasters.
Discussions of factors associated with differential vulnerability and resilience in the face of disasters appear in many places in this report. What these discussions reveal is that researchers have only begun to explore these two concepts and much work remains to be done. It is clear that vulnerability is produced by a constellation of psychological, attitudinal, physical, social, and economic factors. However, the manner in which these factors operate and interact in the context of disasters is only partially understood. For example, while sufficient evidence exists to indicate that race, gender, and ethnicity are important predictors of hazard vulnerability and disaster-related behavior, research has yet to fully explore such factors, their correlates, and their interactions across different hazard and disaster contexts. In many cases age is associated with vulnerability to disasters (see Ngo, 2001; Anderson, 2005), but other factors such as ethnicity and socioeconomic status have differential effects within particular age groups (Bolin and Klenow, 1988), and the vulnerability of elderly persons may be related not only to age but also to other factors that are correlated with age, such as social isolation, which can cut off older adults from sources of lifesaving aid under disaster conditions (Klinenberg, 2002).
Even less is known about how to conceptualize, measure, and enhance resilience in the face of disasters—whether that concept is applied to the psychological resilience of individuals or to the resilience of households, communities, local and regional economies, or other units of analysis. Resilience can be conceptualized as the ability to survive disasters without significant loss, disruption, and stress, combined with the ability to cope with the consequences of disasters, replace and restore what has been lost, and resume social and economic activity in a timely manner (Bruneau et al., 2003). Other dimensions of resilience include the ability to learn from disaster experience and change accordingly.
The large volume of literature on psychological resilience and coping offers insights into factors that facilitate resilient responses by individual disaster victims. Other work, such as research on “high-reliability organizations,” organizational adaptation and learning under crisis conditions, and organizational effectiveness (Roberts, 1989; La Porte and Consolini, 1998; Comfort, 1999; Drabek, 2003) also offers insights into correlates of resilience at the organizational and interorganizational levels. As suggested in Chapter 6, the social capital construct and related concepts such as civic engagement and effective collective action are also related to resilience. The challenge is to continue research on the resilience concept while synthesizing theoretical insights from these disparate literatures, with the ultimate objective of developing an empirically grounded
theory of resilience that is generalizable both across different social units and across different types of extreme events.
Recommendation 4.2: Future research should focus on the special requirements associated with responding to and recovering from willful attacks and disease outbreaks.
A better understanding is needed of likely individual, group, and public responses to intentional acts of terrorism, as well as disease outbreaks and epidemics. As indicated in this chapter, there appears to be no strong a priori reason for assuming that responses to natural, technological, or intentionally caused disasters and willful or naturally occurring disease outbreaks will differ. However, research on hazards and disasters also calls attention to factors that could well prove to be important predictors of responses to such occurrences, particularly those involving unique hazards such as chemical, biological, nuclear, and radiological agents. Research on individual and group responses to different types of disasters has highlighted the importance of such factors as familiarity, experience, and perceptual cues; perceptions about the characteristics of hazards (e.g., their dread nature, lethality and other harms); the content, clarity, and consistency of crisis communications; knowledge of appropriate self-protective actions; and feelings of efficacy with respect to carrying out those measures (see, for example, classic work on risk perception, discussed in Slovic, 2000, as well as Lindell and Perry, 2004).
Recent research has also highlighted the importance of emotions in shaping perceptions of risk. Hazards that trigger vivid images of danger and strong emotions may be seen as more likely to occur, and more likely to produce harm, even if their probability is low (Slovic et al., 2004). If willful acts engender powerful emotions, they could potentially also engender unusual responses among threatened populations.
The potential for ambiguity and confusion with respect to public communications may also be greater for homeland security threats and public health hazards such as avian flu than for other hazards. For example, warning systems and protocols are more institutionalized and more widely understood for natural hazards than for homeland security and public health threats. While it is generally recognized that organizations such as the National Hurricane Center and the U.S. Geological Survey constitute reliable sources of information on hurricanes and earthquakes, respectively, members of the public may be less clear regarding responsibilities and authorities with respect to other risks, particularly since such threats and the expertise needed to assess them are so diverse.
These kinds of differences could translate into differences in public perceptions and subsequent responses. Research is needed on the manner in which the distinctive features of particular homeland security and public
health threats, such as those highlighted here, as well as official plans and management strategies, could affect responses during homeland security emergencies.
Recommendation 4.3: Future research should focus on the societal consequences of changes in government organization and in emergency management legislation, authorities, policies, and plans that have occurred as a result of the terrorist attacks of September 11, 2001, as well as on changes that will almost certainly occur as a result of Hurricane Katrina.
The period since the 2001 terrorist attacks has been marked by major changes in the nation’s emergency management system and its plans and programs. Those changes include the massive government reorganization that accompanied the creation of the Department of Homeland Security (DHS); the transfer of FEMA, formerly an independent agency, into DHS; the shifting of many duties and responsibilities formerly undertaken by FEMA to DHS’s Office of Domestic Preparedness, which was formerly a part of the Justice Department; the development of the National Response Plan, which supercedes the Federal Response Plan; Presidential Homeland Security Directives 5 and 8, which make the use of the National Incident Management System (NIMS) mandatory for all agencies and organizations involved in responding to disasters and also mandate the establishment of new national preparedness goals; and increases in funding for special homeland security-related initiatives, particularly those involving “first responders.” Other changes include a greater emphasis on regionalized approaches to preparedness and response and the growth at the federal, state, and local levels of offices and departments focusing specifically on homeland security issues—entities that in many cases exist alongside “traditional” emergency management agencies. While officially stressing the need for an “all-hazards” approach, government initiatives are concentrating increasingly on preparedness, response, and recovery in the context of willful attacks. These changes, all of which have taken place within a relatively short period of time, represent the largest realignment of emergency management policies and programs in U.S. history.
What is not known at this time—and what warrants significant research—is how these changes will affect the manner in which organizations and government jurisdictions respond during future extreme events. Is the system that is evolving more centralized and more command-and-control oriented than before September 11? If so, what consequences will that have for the way organizations and governmental entities respond? What role will the general public and emergent groups play in such a system? How will NIMS be implemented in future disasters, and to what effect? What new forms will emergent multiorganizational networks assume in future
disasters? Which agencies and levels of government will be most central, and how will shifts in authority and responsibility affect response and recovery efforts? Will the investment in homeland security preparedness translate into more rapid, appropriate, and effective responses to natural and technological disasters, or will the new focus on homeland security lead to an erosion in the competencies required to manage other types of emergencies? A major research initiative is needed to analyze the intended and unintended consequences in social time and space of the massive changes that have taken place in the nation’s emergency management system since September 11, 2001.
These concerns loom even larger in the aftermath of Hurricane Katrina. That disaster revealed significant problems in virtually every aspect of intergovernmental preparedness and response. The inept management of the Katrina disaster was at least in part a consequence of the myopic institutional focus on terrorism that developed in the wake of the September 11, 2001 attacks—a focus that included marginalizing and underfunding FEMA and downplaying the challenges associated with responding to large-scale natural disasters (Tierney, 2006, forthcoming). Katrina is certain to bring about further efforts at reorganizing the nation’s response system, particularly at the federal level. These reorganizations and their consequences merit special attention.
Recommendation 4.4: Research is needed to update current theories and findings on disaster response and recovery in light of changing demographic, economic, technological, and social trends such as those highlighted in Chapter 2and elsewhere in this report.
It is essential to keep knowledge about disaster response and recovery current. The paragraphs above highlight the need for new research on homeland security threats and institutional responses to those threats. Research is also needed to update what is known about disaster response and recovery in light of other forms of social change and to reassess existing theories. Technological change is a case in point. Focusing on only one issue—disaster warnings—the bulk of the research that has been conducted on warning systems and warning responses was carried out prior to the information technology and communications revolutions. With the rise of the Internet and interactive Web-based communication, the proliferation of cellular and other wireless media, and the growing potential for ubiquitous communications, questions arise regarding the applicability of earlier research findings on how members of the public receive, interpret, and act on warnings. Changes in the mass media, including the rise of the 24-hour news cycle and the trend toward “narrowcasting” and now “podcasting” for increasingly specialized audiences, also have implications for the ways in which the public learns about hazards and receives warning-related
information. In many respects, warning systems reflect a preference for “push-oriented” information dissemination approaches. However, current information collection practices are strongly “pull oriented.” These and other trends in communications technology introduce additional complexity into already complex processes associated with issuing and receiving warnings, decision making under uncertainty, and crisis-related collective behavior. New research is needed both to improve theories and models and to serve as the basis for practical guidance.
Much the same can be said with respect to organizations charged with responding during disaster events. Along with being affected by policy and programmatic changes such as those discussed above, crisis-relevant agencies are also being influenced by the digital and communications revolution and by the diffusion of technology in areas such as remote sensing, geographic information science, data fusion, decision support systems, and visualization. In the more than 15 years since Drabek (1991b) wrote Microcomputers and Emergency Management, which focused on the ways in which computers were affecting the work of local emergency management agencies, technological change has been rapid and massive. How such changes are affecting organizational performance and effectiveness in disasters is not well understood and warrants extensive systematic study.
Recommendation 4.5: More research is needed on response and recovery for near-catastrophic and catastrophic disaster events.
Chapter 1 discusses issues of determining thresholds of disastrous conditions. NEHRP-sponsored social science research indicates that, in the main, U.S. communities have shown considerable resilience even in the face of major disasters. Similarly, at the individual level, U.S. disasters have produced a range of negative psychosocial impacts, but such impacts appear to have been neither severe nor long-lasting. While recognizing that disasters disproportionately affect the most vulnerable in U.S. society and acknowledging that recovery is extremely difficult for many, disasters have been less devastating in the United States and other developed societies than in the developing world. Disaster-related death tolls have also been lower by orders of magnitude, and economic losses, although often large in absolute terms, have also been lower relative to the size of the U.S. economy. At least that was the case until Hurricane Katrina, a catastrophic event that has more in common with disasters in the developing world than with the typical U.S. disaster.
The vast majority of empirical studies on which such generalizations are based have not focused on truly catastrophic disasters, and therefore research results may not be “scalable” to such events. Katrina clearly demonstrates that the nation is at risk for events that are so large that they overwhelm response systems and produce almost insurmountable post-
disaster recovery challenges. What kinds of social and economic impacts and outcomes would result from a large earthquake under downtown Los Angeles, a 7.0 earthquake event on the Hayward Fault in the San Francisco Bay area, a repeat of Hurricane Andrew directly striking Miami, or another hurricane landfall in the already devastated Gulf Coast region? What about situations involving multiple disaster impacts, such as the 2004 hurricane season in Florida and multiple disaster events that produce protracted impacts over time, such as the large aftershocks that are now occurring after the Indian Ocean earthquake and tsunami? To move into the realm of worst cases, what about an attack involving weapons of mass destruction, or simultaneous terrorist attacks in different cities around the United States? Such events are not outside the realm of possibility. There is a need to envision the potential social and economic effects of very large disasters, to learn from catastrophic events such as Hurricane Katrina, and to analyze historical and comparative cases for the insights they can provide.
Recommendation 4.6: More cross-societal research is needed on natural, technological, and willfully caused hazards and disasters.
Most of the research discussed in this chapter has focused on studies conducted within the United States, but it is important to recognize that findings from U.S. research cannot be overgeneralized to other societies. Disaster response and recovery challenges are greater by many orders of magnitude in smaller and less developed societies than in larger and more developed ones.
Disaster impacts, disaster responses, and recovery processes and outcomes clearly vary across societies. Although the earthquakes that struck Los Angeles in 1994, Kobe in 1995, and Bam, Iran, in 2003 were roughly equivalent in size, they differed in almost every other way: lives lost, injuries, extent of physical damage, economic impacts, and subsequent response and recovery activities. Research suggests that such cross-societal differences are attributable to many factors, including differences in physical and social vulnerability; governmental and institutional capacity; government priorities with respect to loss reduction; and response and recovery policies and programs (see, for example, Davis and Seitz, 1982; Blaikie et al., 1994; Berke and Beatley, 1997; Olson and Gawronski, 2003). NEHRP has made significant contributions to cross-societal research through initiatives such as the U.S.-Japan research program on urban earthquake hazards, which was launched following the Northridge and Kobe earthquakes, as well as a similar initiative that was developed after the 1999 Turkey and Taiwan earthquakes. In some cases, these initiatives have led to longer-term research partnerships; Chapter 6 contains information on one such collaboration, involving the Texas A&M University Hazard Reduction and Recovery Center and the National Center for Hazards Mitigation at the National
Taiwan University. Significantly more cross-national and comparative research is needed to further document and explain cross-societal variations in response and recovery processes and outcomes across different scales and different disaster events. Disasters such as the Indian Ocean earthquake and tsunami merit intensive study because they allow for rich comparisons at various scales (individuals, households, communities, and institutional and societal levels), providing an opportunity to greatly expand existing social science knowledge.
Recommendation 4.7: Taking into account both existing research and future research needs, sustained efforts should be made with respect to data archiving, sharing, and dissemination.
As noted in detail in Chapter 7, attention must be paid to issues related to data standardization, data archiving, and data sharing in hazards and disaster research. NEHRP has been a major driving force in the development of databases on response and recovery issues. However, vast proportions of these data have yet to be fully analyzed. For social scientists to be able to fully exploit the data that currently exist, let alone the volume of data that will be collected in the future, specific steps have to be taken to make available and systematically collect, preserve, and disseminate such data appropriately within the research community. As recommended in Chapter 7, information management strategies must be well coordinated, formally planned, and consistent with federal guidelines governing the protection of information on human subjects. Assuming that these foundations are established, the committee supports the creation of a Disaster Data Archive organized in ways that would encourage broader use of social science data on disaster response and recovery. Contents of this archive would include (but not be limited to) survey instruments; cleaned databases in common formats; code books, coding instructions and other forms of documentation; descriptions of samples and sampling methods; collections of papers containing analyses using those databases; photographs and Internet links (where applicable); and related research materials. Procedures for data archiving and sharing would build on existing protocols set out by organizations such as the Inter-University Consortium for Political and Social Research (e.g., ICPSR, 2005).
The distributed Disaster Data Archive would perform a number of important functions for social science hazards and disaster research and for the nation. The existence of the archive would make it much more likely that existing data sets will be used to their full potential by greatly improving accessibility. The archive would serve as an important tool for undergraduate and graduate education by making data more easily available for course projects, theses, and dissertations. By enabling researchers to access instruments used in previous research and incorporate past survey and
interview items into their own research, the archive should help make social science research on disasters more cumulative and replicable. An archive would also make it easier for newcomers to the field of disaster research to become familiar with existing research and enable researchers to identify gaps in past research and avoid unnecessary duplication. The archive would also serve an important function in preserving data that might otherwise be lost. Finally, such an archive would enable social science disaster research to better respond to agency directives regarding the desirability of data sharing.
For an effort of this kind to succeed, a number of conditions must be met. Funds will be needed to support the development and maintenance of the archive, and researchers must be willing to make their data sets and all relevant documentation available. This second condition is crucial, because the committee is aware of a number of important data sets that are not currently being shared, and the archive cannot succeed without broad researcher support. Challenges related to human subjects review requirements, confidentiality protections, and disclosure risks must be fully explored and addressed. Other issues include challenges associated with the development and enforcement of quality control standards, rules and standards for data sharing, procedures to ensure that proper acknowledgment is given to project sponsors and principal investigators, and questions about long-term management of the archive.
Related to the need for better data archiving, sharing, and dissemination strategies, social scientists must be poised to take advantage of new capabilities for data integration and fusion. Strategies are needed to integrate social science data with other types of data collected by both pervasive in situ and mobile ad hoc sensor networks (Estrin et al., 2003), such as networks that collect data on environmental and ecological changes and disaster impacts. In light of the availability of such a wide array of data, the hazards and disasters research community must recognize that hazards and disaster informatics—the application of information science and technology to disaster research, education, and practice—is an emerging field.
To realize this potential, and with the foundation established through implementing recommendations in Chapter 7, the committee further supports the creation of a Data Center for Social Science Research on Hazards and Disasters. In addition to maintaining the Disaster Data Archive, this center would conduct research on automated information extraction from data, including the development of efficient and effective methods for storing, querying, and maintaining both qualitative and quantitative data from disparate and heterogeneous sources.