2
Direct Losses of Natural Disasters

This committee was requested to identify the cost components that, when combined, would most accurately reflect the total cost of a natural disaster event. The committee thus identified the data it felt should be consistently used in compiling accurate loss estimates.

The committee acknowledged that many of the losses in natural disasters are intangible and difficult to quantify, such as personal anguish, the loss of family treasures, and the disruption of family and work routines. Indeed, these losses may sometimes be greater than the losses of direct physical destruction. Despite the importance of such losses, however, the great difficulties in objectively measuring them make their use in consistent and accurate loss estimations problematic.

The committee's recommendations for those data to be used in compiling accurate loss estimates thus focus on direct losses, as they are easier to objectively measure. This chapter is devoted to direct loss measurement and those direct loss data which should be included in loss estimates. Chapter 3 is devoted to an assessment of indirect losses and ways in which indirect loss estimations might be improved.

As discussed in Chapter 1, it is useful to distinguish between the physical destruction caused by natural disasters to human beings and property, which is the subject of the current chapter, and the consequences of that destruction, considered in the next chapter. In economic terms, physical destruction may be thought of as a loss in asset value (and is often referred to as the direct loss from the event), whereas the consequences of that destruction may be considered to be the loss of income and/or production and impacts on the environment that cannot be readily stated in monetary terms (all of which are included among a disaster's indirect impacts).

So-called direct losses in turn consist of two more refined types of losses. Primary direct losses are those resulting from the immediate destruction caused by the event, such as shake damage from an earthquake or water and wind damage from a hurricane. Secondary direct losses are those additional impacts resulting from follow-on physical destruction, such as fire following an earthquake (due perhaps to breaks in gas lines) or additional water damage to



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 15
--> 2 Direct Losses of Natural Disasters This committee was requested to identify the cost components that, when combined, would most accurately reflect the total cost of a natural disaster event. The committee thus identified the data it felt should be consistently used in compiling accurate loss estimates. The committee acknowledged that many of the losses in natural disasters are intangible and difficult to quantify, such as personal anguish, the loss of family treasures, and the disruption of family and work routines. Indeed, these losses may sometimes be greater than the losses of direct physical destruction. Despite the importance of such losses, however, the great difficulties in objectively measuring them make their use in consistent and accurate loss estimations problematic. The committee's recommendations for those data to be used in compiling accurate loss estimates thus focus on direct losses, as they are easier to objectively measure. This chapter is devoted to direct loss measurement and those direct loss data which should be included in loss estimates. Chapter 3 is devoted to an assessment of indirect losses and ways in which indirect loss estimations might be improved. As discussed in Chapter 1, it is useful to distinguish between the physical destruction caused by natural disasters to human beings and property, which is the subject of the current chapter, and the consequences of that destruction, considered in the next chapter. In economic terms, physical destruction may be thought of as a loss in asset value (and is often referred to as the direct loss from the event), whereas the consequences of that destruction may be considered to be the loss of income and/or production and impacts on the environment that cannot be readily stated in monetary terms (all of which are included among a disaster's indirect impacts). So-called direct losses in turn consist of two more refined types of losses. Primary direct losses are those resulting from the immediate destruction caused by the event, such as shake damage from an earthquake or water and wind damage from a hurricane. Secondary direct losses are those additional impacts resulting from follow-on physical destruction, such as fire following an earthquake (due perhaps to breaks in gas lines) or additional water damage to

OCR for page 15
--> unrepaired structures from rain following a hurricane (as happened after Hurricane Andrew).1 A third important distinction is the difference between reimbursed and unreimbursed losses from natural disasters. Reimbursed losses—referred to in this report as the ''costs" of a natural disaster—include loss claims that are paid by private insurers or local, state, and federal governments. In contrast, unreimbursed losses are the uncompensated impacts that victims must bear. Different types of disasters tend to produce different proportions of reimbursed and unreimbursed losses. For example, a larger fraction of the total losses from earthquakes typically is unreimbursed—primarily because many consumers and businesses choose not to purchase insurance coverage and secondarily because insurance policies for these disasters typically contain large deductibles—than is the case for hurricanes and other wind storms. It cannot be emphasized too strongly that policymakers concerned with devising effective mitigation measures must take account of all losses, whether reimbursed or unreimbursed (and, to the extent possible, estimates of indirect losses, discussed in the next chapter). Some natural disasters trigger the expenditure of additional resources devoted to mitigating future disasters. For example, presidentially declared disasters generally trigger FEMA expenditures on mitigation, provided they are matched to some degree (with some exceptions, as discussed later) by state and/or local governments. In the balance of this chapter, we address three central questions: What information about the destruction caused by natural disasters should ideally be collected and reported? What data about the destructive impacts of natural disasters are actually collected and reported? What steps need to be taken to ensure that all of the relevant information about the destruction due to natural disasters is collected and reported consistently? 1   In some analyses, what we have labeled "direct secondary losses" are treated as indirect losses. We do not do so here because such damage as is caused by fire following an earthquake or additional rain following a hurricane, for example, entails physical destruction, whereas the indirect losses we discuss in the next chapter all result from that destruction.

OCR for page 15
--> Data Relating to Physical Destruction: The Ideal Researchers seem never to have enough data. But collecting information is costly for those who must supply it as well as for those who must compile, organize, and distribute it. The challenge for policymakers interested in obtaining comprehensive and accurate data on the direct costs of natural disasters is to balance the benefits of allowing many constituencies (including the government) access to the information against the costs imposed on others (and the government) in attempting to collect and distribute it. In principle, the desired data should be collected in a disaggregated form across several dimensions. Once collected, the data can be reaggregated and reported in multiple ways, depending on the purposes for which they may be used. Whatever data are compiled should ideally be categorized by type of disaster so that insurers, citizens, and policymakers can be informed of the relative severities and costs of the various events. In our judgment, these should be classified as hurricanes, floods (caused by events other than hurricanes), earthquakes, wildfires, landslides, volcanic eruptions, drought, winter storms, windstorms, hail, tornadoes, and all other events with losses above a certain threshold. Different kinds of disasters entail different types of losses and it is important in both the private and public sectors to be aware of these impacts. Similarly, all cost and loss data should be coded by the state(s) and, if possible, by county and zip code where the losses occur in order to identify where mitigation measures may be most necessary. In addition, state-specific data will be essential if the federal government is authorized to offer disaster related reinsurance to the private market, because it is possible that some coverage may be offered on a state or regional basis. For each major event (or annually by type of event, as the case may be), efforts should be made to collect and report data for both the type of damage caused, and which parties initially bear the losses (we stress the term "initially" because some losses are eventually passed forward to consumers). Natural disasters generally result in the following categories of destruction: to property (structures, contents, and transportation vehicles) with different types of owners (individual residences, businesses, and government-owned infrastructure);2 to agricultural products (crops and livestock); and to people (injuries and death, life insurance payouts, and medical treatment expenses). In addition, disasters require expenditures for response and cleanup and temporary living expenses of displaced people. These losses are absorbed by insurers, governments, 2   It would be useful to further disaggregate losses to structures (by type of building, such as apartment houses and single-family residences); to contents (equipment and inventory/supplies); and to infrastructure (roads, bridges, utilities, and transport facilities).

OCR for page 15
--> businesses, individuals, and nongovernmental organizations (such as charities and relief agencies). As a practical matter, some loss data that are now collected do not (and indeed cannot) distinguish between primary and secondary impacts. But conceptually the distinction is important. As an example, the roofs of homes blown off by hurricanes (a primary impact) will be covered by homeowners' insurance. A well-constructed home, whose roof stays attached during a hurricane, but is damaged by rising water (a secondary impact), is not covered unless the homeowner carries flood insurance. If the roof is blown off, it may be difficult to distinguish which caused the damage first—the roof being dislodged or the rising water. The definition of a "major" natural disaster (for which loss data are to be compiled) should be consistent. One definition would include all events that the president certifies as a "major disaster" under the Stafford Act (and thus whose victims become eligible for disaster assistance provided through FEMA), as well as events in which costs rise above a certain dollar threshold (such as the $25 million now used by PCS).3 The Robert T. Stafford Disaster Relief and Emergency Assistance Act (P.L. 93–288, as amended) is the core statute under which federal emergency management is conducted. The Stafford Act authorizes the president to issue major disaster or emergency declarations, sets broad eligibility criteria, and specifies the type of assistance the president may authorize. There necessarily is some element of judgment required in deciding at what point to cut off the cumulating claims following a specific event or whether to treat a series of events—such as several days of heavy rainfall or an earthquake followed by several aftershocks—as a single event for purposes of satisfying the dollar threshold. Rather than offer specific recommendations for this problem, we suggest that the agencies we recommend be charged with additional data collection responsibilities (described later in this chapter) strive to ensure that the methods used for different types of disasters are consistent.4 Measuring precisely the losses of natural disasters takes time. In the case of earthquakes, many victims may not know for weeks or months the extent of the damage their homes or businesses have suffered. Initial loss estimates may thus understate actual losses, potentially by wide margins. For example, it 3   A more expansive definition could also include events certified as "emergencies" under the Disaster Relief Act of 1974. These are incidents requiring up to $5 million in spending for emergency work that is essential to save lives and protect property. An even broader definition would include the many thousands of smaller disasters that are declared by local and state governments (which the committee believes to be several times the number of presidentially declared disasters). The more expansive the definition, however, the more costly it will be to obtain timely and accurate data. 4   Two approaches exist: (1) the PCS describes a weather system to tie multiple events on consecutive days together; (2) hazard analysts define the single most damaging event and add to it the lesser events that occur before and after.

OCR for page 15
--> was several months after the 1994 Northridge earthquake until many building owners realized that their buildings required major renovation. Ultimately, privately insured damages from that earthquake, initially estimated at $2.5 billion, exceeded $10 billion.5 Any comprehensive loss data base should therefore be held open to reflect revised data (just as economic data are now revised, often years later, to take account of new information and methods of estimation). Ideally, not only should detailed loss data be collected for all major disasters, but an attempt should also be made to gather as much historical data as are available. The more complete the data base, the greater use it will be to all potential users—in government and in the private sector. Direct-Impact Data: The Reality The reality is short of the ideal. The types of loss data that are currently collected, at least by some organizations (private or public), are reviewed below. We do so by identifying the sources of the data, classified by parties that bear those losses. Insurance Claims Disaster-induced losses typically are reimbursed to some extent by insurance companies. Many people who die in a natural disaster may have life insurance; others who are injured may carry medical insurance. In principle, these claims data are kept by the policyholders' insurance companies. As a practical matter, however, it is not clear to what extent these companies identify claims that are disaster related. The situation is clearer for disaster-related property damage. The most comprehensive data base of insurance claims payments for property damage is the one compiled by Property Claims Services for catastrophe-triggered events. PCS has collected these data since 1949, using a dollar cost threshold to determine whether an event qualifies as a major disaster. If so, each event is assigned a number and claims for it are updated as they come in. As mentioned, the PCS threshold was initially $1 million and has twice been revised upward (to $5 million in 1983 and to $25 million in 1997). By these criteria, PCS has collected data on approximately 1,200 catastrophes. If PCS used a common threshold for the entire period since 1949—such as the 1997 cutoff of $25 5   These figures are based on estimates supplied by a representative of the PCS who met with our panel in March 1998. For a more definitive account of the losses from Northridge, see Eguchi et al., 1998, p. 248.

OCR for page 15
--> million, adjusted backward in prior years for inflation—the total number of catastrophes in the PCS data base would be somewhat lower. 6 The PCS data base is the most comprehensive one available for insured losses generated by natural disasters. One useful feature of the data is that they are maintained separately for each major event, by state. A drawback, however, is that the data combine rather than disaggregate losses to property, contents, business interruption and additional living expenses of individuals and families. Another drawback is that the PCS data base does not include damage due to floods, and it combines cost data for wind and hail storms and tornadoes. Separately, the Institute for Business and Home Safety (IBHS) in 1994 began to compile more disaggregated catastrophe claims information, using the PCS dollar cost thresholds. The IBHS data base compiles actual paid losses of large insurers who account for most of the property-casualty market. Estimates are then made (using market shares) of the claims paid by other, smaller insurers. The IBHS data base has the advantage of breaking down damages suffered by businesses and individuals, to buildings and their contents, and by location (state, county, and zip code). Neither PCS nor IBHS maintains data on insured claims paid to individuals for injuries and deaths. In principle, these data are available from medical and life insurers, but to this committee's knowledge, no systematic effort has been made by any organization, public or private, to assemble it. We recommend that the agency charged with assembling the loss data make use of the PCS and IBHS data bases. To the extent the agencies find these sources of information inadequate, we recommend that the federal government work with relevant insurance trade associations and with state insurance commissioners (either with their trade association, the National Association of Insurance Commissioners, or the commissioners of individual states) to obtain data suitable for federal purposes. Since the data sought would be aggregated across insurers, it should not run up against the confidentiality concerns of insurers. The state insurance commissioners in particular should have an interest in cooperating with the federal government in a data collection effort, as they have direct responsibility for overseeing both the rates and the solvency of the insurers who do business in their states. In that capacity, they also have the legal means to compel production of the data. Working with the federal government to standardize data requests would ease the burdens on insurers while providing the government and the public with the information in the most suitable form. One interesting complication with losses to insurers should be noted. In a worst case, some disasters could be large enough that they force the 6   Without being able to examine the PCS data base, we cannot say by how much lower the total number of events would be. Separately, PCS also estimates the net insurance payments likely to be made following major catastrophes. This is a useful service, but the data that are of interest to us in this report are the actual paid insured losses that PCS compiles.

OCR for page 15
--> Flood overtopping Sorlie Bridge in Grand Forks, North Dakota, during the spring of 1997. (Photo courtesy of the U.S. Geological Survey.) bankruptcy of one or more insurers (as was the case with Hurricane Andrew). In that event, claims on insurers would appear to exceed the losses that the industry actually bears. In fact, this is not likely to be the case. States have guaranty funds, which pay claims of failed insurers. These funds typically are financed by post-event assessments on other, surviving insurers doing business in the state. As a result, the insurance industry as a whole is almost certainly likely to bear all of the claims on it even though individual insurers might be forced into insolvency. Losses to Government All branches of government—federal, state, and local and tribal—bear losses associated with natural disasters, which fall into four categories.7 First, the largest costs are disaster payments made to individuals and businesses by the various governments, primarily agencies of the federal government. The main source of federal disaster aid is FEMA, which provides grants to individuals, states, and local governments suffering damage due to presidentially declared disasters. When the president declares an area eligible for 7   This discussion is based in part on a meeting certain members of the committee and the NRC staff had with representatives of federal agencies involved in disaster aid and planning.

OCR for page 15
--> disaster assistance, FEMA makes money available to the states, which normally is conditioned on state matches of 25 percent of the total. The president may waive all or part of the state match, which often occurs in especially large disasters (as was the case in Hurricane Andrew and the Mississippi floods).8 FEMA can distribute, however, only the funds that are made available to it by the Congress through the appropriations process. In recent years, FEMA's disaster payments have significantly exceeded the agency's annual appropriation for such assistance (which, as noted in the first chapter, for the past several years has been $320 million, but in fiscal year 1999 was lowered to $308 million). The additional funds have been provided through supplemental appropriations, which until 1995 were added to the overall federal budget without offsets (reductions in other expenditures) in other federal programs.9 Since then, Congress has required cuts in appropriations of other agencies to pay for supplemental disaster aid. FEMA maintains current and historical data on the assistance it pays, by event, by state, and by type of aid: for infrastructure damage, payments to individuals for property damage, and payments to individuals for adjustment (temporary housing, unemployment, inspections, crisis counseling, and legal services). In addition, FEMA separately maintains data on grants and contracts for mitigation of hazards.10 FEMA does not, however, maintain data on private-sector costs arising from disasters. The Small Business Administration (SBA) is another important source of disaster aid, providing low-interest (between 4–8 percent) loans to credit worthy businesses and individuals (approximately 60 percent of disaster victims who apply) who have suffered property damage from a disaster. The SBA currently compiles its lending data by event and by type of property (and could, if given the resources, aggregate the assistance by type of disaster). The U.S. Department of Agriculture (USDA) also offers lending assistance to farmers and ranchers for losses to crops and livestock due to disasters declared by the president or governors. It is important to note that the true costs of federal loan programs are not measured by the total amount of loans disbursed, but instead by the present value of the interest subsidies on those loans. 8   FEMA can lend money to states to help them meet the required cost share (where it is not waived). In addition, the agency reimburses certain other agencies for their disaster-related expenses, notably the Department of Labor, which provides unemployment compensation to individuals rendered unable to work by a disaster and who are not otherwise covered by the department's regular unemployment insurance program (such as agricultural workers, self-employed individuals, and recent entrants into the labor force). Unemployment insurance costs are part of the indirect losses we discuss in the next chapter. 9   Under the Budget Enforcement Act of 1990, supplemental appropriations for disasters are exempt from the caps on discretionary federal spending if Congress designates the added funds for an emergency. 10   For a list of disasters most costly to FEMA, see Table 1–3.

OCR for page 15
--> Second, governments bear damage-related losses to the buildings and infrastructure they own. To a significant extent, the federal government assumes much of the costs that otherwise would be borne by the states and localities. For example, damaged buildings owned by state and local governments (and certain nonprofit organizations) are eligible for compensation by FEMA if the disasters are large enough to merit a presidential declaration. The Federal Highway Administration (FHA) compensates states for up to 90 percent of the costs associated with the repair of roads if a disaster is declared either by the president or the state's governor. In addition, the Department of Housing and Urban Development (HUD) provides funds for repair of damaged housing and public facilities through its regular Community Development Block Grant program (some portion of whose grants some localities use for disaster recovery) or, in the event of large disasters, through supplemental appropriations granted by Congress. The federal government also bears the losses of damages to property that it owns directly. Obvious examples include damaged federal buildings and facilities. Less known but often significant are damages caused by drought, fires, and floods to federally owned land and forests. For example, the Mississippi floods of 1993 caused $143 million in damages to federal facilities (Changnon, 1996b). Third, all levels of government bear costs in responding to disasters, although for major disasters FEMA compensates local and state governments for their response and cleanup. In addition, FEMA reimburses the Department of Health and Human Services (HHS) for post-disaster counseling (although HHS provides a modest amount of health care assistance out of its own budget). And perhaps the largest response costs are those that fall on the Department of Interior, which typically spends several hundred million dollars a year fighting fires. Finally, the federal government operates two major insurance programs that offer coverage—and thus make payouts—for certain disaster-induced damages. The U.S. Department of Agriculture provides insurance covering crop losses from a range of weather-related impacts (e.g., drought, flood, hail, excess moisture), and losses due to insects. The department collects data on loss payouts by type of crop but not by type of disaster. Meanwhile, the National Flood Insurance Program (NFIP), provides flood insurance to businesses and individuals in flood-prone areas. The NFIP includes three essential components: risk identification, hazard mitigation, and insurance. While the authority for the NFIP rests with FEMA, effective integration of these three components requires cooperation between the federal, state, and local governments and the private property insurance sector (Pasterick, 1998). By definition, the losses from this program are well identified with a single type of disaster (although it is not clear if the data on flood losses can be broken out by type of loss—that is, to structures, personal property, and the like).

OCR for page 15
--> The growth in the costs of natural disasters during the 1990's—both to the private sector and to the government—suggests that federal, state, and local governments should adopt a more systematic approach to data collection for their disaster costs. Such a data base would be useful for several reasons: A comprehensive data base and accompanying report would inform policymakers in both the executive and legislative branches, as well as the broader public, how sizable these losses probably are and clarify which government agencies actually deliver disaster-related aid and services. In addition, this information is critical for developing and implementing cost-effective ways to mitigate the losses of natural disasters. Comprehensive data on the federal government's spending on disasters would assist both the executive and legislative branches in budgeting and planning for disaster-related expenditures. The committee recognizes that members of each branch have incentives not to budget all such expenditures in advance, as there are political gains from appearing to take concrete actions in the immediate aftermath of a disaster, such as proposing and voting on large disaster relief supplementals (additional items in the federal budget). However, the short-run political gains have economic consequences: the supplementals require offsetting cuts in other programs in midyear, as has been the case since the Northridge supplemental considered by Congress in 1995. Offsets randomly interrupt the functioning of other parts of the federal government, delaying the delivery of services and potentially adding to the cost of implementing or developing government programs. The availability of a comprehensive data base, with suitable historical data, might enable policymakers to smooth out disaster-related costs by budgeting them at actuarially appropriate levels, with any surpluses banked in a reserve to be drawn down in years when actual payouts are larger than anticipated. But even small steps toward better planning, short of establishing a single or multiple disaster accounts in the appropriations process, cannot be taken until and unless the information is compiled. Assembling the federal government's disaster cash payouts data can also be usefully compared to the tally of losses absorbed by the private sector (insured and self-insured). Such a comparison would reveal the relative financial burdens of the government and the private sector in connection with disasters. It might also encourage greater use of insurance rather than reliance on taxpayer-supported government expenditures. As it is, the prospect that they might receive some disaster aid reduces the incentive of individuals and businesses to purchase insurance to cover such expenses. What many uninsured individuals and businesses may not realize is that although they may well receive some compensation from the federal government if they are victims of a major disaster, the federal aid is unlikely to be as generous or promptly paid as it would be if the individuals opted to purchase insurance. To the extent individuals and businesses in areas subject to catastrophic hazards are informed

OCR for page 15
--> of the limits of federal aid for disasters, it is possible that some who now are uninsured, perhaps out of ignorance, will buy insurance, modify existing structures to better withstand disaster impacts, or otherwise internalize risks that the government (and taxpayers) now carry.11 Losses to Businesses To the extent businesses have insurance for property damage, their covered losses are included in the losses borne by insurance companies. Nonetheless, because insurance policies carry deductibles, even insured commercial operations suffer some losses, whereas businesses that self-insure for natural disasters absorb all of the losses themselves.12 Business losses are, for all practical purposes, not covered by government disaster programs (though both homeowners and businesses can write off their uninsured losses for tax purposes). To be sure, the Small Business Administration and the U.S. Department of Agriculture provide loans for reconstruction, but a loan is still a loan, not a grant. Even businesses that qualify for federal lending assistance eventually bear the full losses themselves. In 1993 the flooded railroads in the Midwest experienced uninsured losses of $169 million (Changnon, 1996a). In principle, insurers who pay claims made by businesses have the data to compute the total amounts of deductibles that their commercial clients must absorb, but no organization currently compiles that information. In addition, some federal agencies may have data on their disaster-related loans that could permit estimates to be made of the losses incurred by businesses that qualify for federal loans (although these data may not be disaggregated by kind of disaster, location, or type of damage—that is, to structures or contents). To the committee's knowledge, there is no organization that maintains data on disaster-related losses absorbed by businesses that self-insure.13 11   FEMA has been attempting through advertisements in the major media to provide information about flood risks and to encourage individuals to purchase flood insurance against them. 12   In addition, as described in the next chapter, some businesses may suffer lost sales indirectly as a result of a disaster, and these costs are not likely to be covered by insurance (nor are they treated as direct costs for purposes of this report). For example, suppliers to firms with plants or local infrastructure that are damaged or destroyed in a hurricane or earthquake may lose sales for some time until the facilities are repaired (although they may later experience a larger-than-normal level of purchases as firms with repaired plants attempt to satisfy some pent-up demand for their goods). 13   Theoretically, some data on such losses should be reflected in the income tax returns that companies pay. But as a practical matter, this information—even if it is reported separately on tax returns—would not be broken down by event, location, or type of damage. Nor is it likely that the Internal Revenue Service could compile such data as may be reported.

OCR for page 15
--> Residential destruction resulting from the Northridge, California earthquake in 1994. (Photo courtesy of FEMA.) Losses to Individuals Finally, many individuals also self-insure some or all of their losses from disasters, although the extent of reinsurance varies by type of disaster. According to information supplied by a PCS representative, roughly 95 percent of buildings damaged by hurricanes are typically insured, whereas only about 10–15 percent of all California homes are covered by earthquake insurance. Similarly, most potential victims of flood damage do not purchase insurance through the National Flood Insurance Program. For example, in the 1993 Mississippi River flooding these numbers were disturbingly low: ''in the counties and communities affected, it is estimated that no more than 10 percent of insurable properties had flood insurance coverage" (Wright, 1996). As with businesses, even insured individuals carry deductibles on their policies. Other individuals may receive some federal aid, which may or may not fully compensate their property losses and temporary expenses. And still other individuals and their families may suffer losses but receive no compensation from either private or public sources. For example, a 1997 winter storm caused $25 million of uninsured losses to property owners in Lincoln, Nebraska, and a 1996 snowstorm in Cleveland caused homeowners $5 million in uninsured losses (see Appendix A). Again, in principle, insurance companies—property-casualty and medical—should be able to compute or estimate the total amounts of their

OCR for page 15
--> deductibles for given events or types of disasters. Similarly, it should be possible for FEMA and/or other government agencies that now provide disaster aid to individuals to ascertain their total property damages. In practice, none of these data on self-insured losses absorbed by individuals are systematically compiled by any organization. The same is true with respect to losses suffered by individuals who do not qualify for or who do not seek either private insurance payments or government assistance. Even more important than the monetary damages suffered by individuals in disasters are the injuries and fatalities that often occur. Through the National Weather Service (NWS), the federal government currently collects comprehensive data on injuries and fatalities in all weather-related events, however small. These data are provided to FEMA. As valuable as it is, the NWS data set contains two shortcomings for purposes of this report: it excludes fatalities and injuries from earthquakes and other geohazards, and it is not clear if the data on disasters below some dollar loss threshold can be easily separated from the larger disasters that, in our view, should be the primary focus of a comprehensive federal disaster data collection effort. Standardizing Loss Estimates In addition to the lack of a comprehensive data base, there exists no standardized estimation technique or framework for compiling loss estimates from individual disasters. Most estimates are ad hoc, consisting of those losses that were significant in a particular event. As a result, the range of loss estimates of a natural disaster tends to vary widely, sometimes as much as 10-fold. Table 2-1 is an example of one framework used in compiling a loss estimate for Hurricane Andrew. This estimate, like all others, is not standardized, and different groups and individuals compiled their own, unique loss estimates from Hurricane Andrew. There is a range of loss estimates following a disaster, but no official estimate (or official scorekeeper). The lack of a consistent framework for loss estimation makes it difficult to accurately compare the losses of natural disaster events to one another. For example, did Hurricane Georges actually cause less damage than Hurricane Hugo or was the loss estimation framework simply different? And of the varying estimates of losses, which one is to be consistently used? Clearly, the lack of a standard framework makes it extremely difficult to accurately identify trends in natural disaster losses (see Howe and Cochrane, 1993, for guidelines on the uniform measurement of economic damages from disasters). Moreover, this inability makes it more difficult for the federal government to identify which disaster mitigation policies represent the more cost-effective options.

OCR for page 15
--> TABLE 2–1. Current Dollar Estimates of $30 Billion in Damages Directly Related to Hurricane Andrew in South Florida. Type of Loss Amount (billions of dollars) Sources and Notes Common insured private property 16.5 Sheets, 1994; includes homes, mobile homes, commercial and industrial properties and their contents, boats, autos, farm equipment and structures, etc. Uninsured homes 0.35 Miami Herald, 16 February 1993, reported in Rappaport, 1993. Federal disaster package 6.5 Anderson et al., 1992; represents 90% of $7.2 billion package (the rest went to Louisiana). Public infrastructure     State County City Schools 0.050 0.287 0.060 1.0 Filkins, 1994; tax revenue shortfall. Rappaport, 1993. Tanfani, 1992; Miami only. Rappaport, 1993. Agriculture     Damages Lost sales 1.04 0.48 McNair, 1992a,b. Fatsis, 1992. Environment 2.124 Rappaport, 1993; includes state request for cleanup and repair of parks, marinas, beaches, and reefs. Aircraft Food claims 0.02 0.096 Rappaport, 1993. FEMA Flood Insurance Administration, reported in Rappaport, 1993. Red Cross Defense Department 0.070 1.412 Swenson, 1993. GAO, 1993; for DOE and USACE expenses during recovery.   SOURCE: Pielke, 1995. In an effort to help standardize the data used in estimating the direct losses in natural disasters, we suggest the framework shown in Table 2-2. This table could be refined by distributing it to parties affected by disasters and asking them for input regarding additional items to be included. If used consistently, this framework should allow the federal government to begin to compile more consistent loss estimates, better understand trends in losses, and ultimately provide a basis for better decisions in hazard mitigation policy.

OCR for page 15
--> The committee recommends that this framework be applied to all of the various types of hazards identified earlier in this chapter. In compiling data for loss estimates, it is generally recommended that losses be calculated as the cost required to restore buildings and structures to their pre-disaster condition. Different hazards naturally affect structures, infrastructure, and people differently. Use of the framework proposed in Table 2-2 should promote a more systematic compilation of the types of losses associated with different disasters. TABLE 2-2 Sample Data on Direct Impacts per Each "Major" Event (dollar amounts should be entered in each cell in the table, except for human losses) Who Initially Bears The Loss Type of Loss Insurers Governmenta Business Individuals NGO Property: Government Structures Contents           Business Structures Contents           Residential Structures Contents Landscapes           Autos, boats and planes           Infrastructure: Utilities Transportation           Agricultural products: Crops Livestock           Human losses: Deaths Injuries           Cleanup and response costsb           Adjustment costs, temporary living aidc           NOTE: If possible, direct primary and secondary losses should be tabulated separately. a Ideally, losses of federal, state, and local and tribal governments should be separately collected and recorded. b Includes costs of added police protection immediately after the event. c Includes expenditures of charities such as the Red Cross.

OCR for page 15
--> Recommendations We recommend the following steps be taken: RECOMMENDATION 2-1: One agency of the federal government should be made responsible for compiling a comprehensive data base containing the losses of natural disasters, adhering to the structure outlined in Table 2-2 wherever it is feasible. The committee believes that the Bureau of Economic Analysis (BEA) within the U.S. Department of Commerce, in consultation with FEMA and other federal agencies involved in natural disaster preparedness, response, and mitigation activities, is best suited for this purpose. The U.S. Department of Commerce is a logical agency to carry out this assignment because two of its major components already have related responsibilities: the National Weather Service compiles loss estimates for all weather-related disasters and the Bureau of Economic Analysis regularly compiles and reports data on the nation's economic performance, an activity closely related to collecting and reporting data on the economic impacts of natural disasters. Along with the Census Bureau and STAT-USA, BEA is part of the Commerce Department's Economics and Statistics Administration. BEA's mission is to produce and disseminate accurate, timely, and relevant statistics that provide government, businesses, households, and individuals with a comprehensive, up-to-date picture of economic activity. BEA's national, regional, and international economic accounts present basic information on key issues such as U.S. economic growth, regional economic development, and the nation's position in the world economy. The BEA develops its figures of economic performance in a setting relatively free of political bias and vested interests. For all these reasons, it appears to be the agency most capable of compiling consistent disaster loss estimates to the nation. In compiling its loss estimates on earthquakes in particular, the U.S. Department of Commerce should draw on the data supplied by state, local, and regional governments that use the HAZUS (Hazards, U.S.) earthquake loss estimation tools developed by the National Institute of Building Sciences for FEMA (see Box 2-1). The data bases, tools (e.g., GIS), and engineering and technical knowledge used in the HAZUS model for earthquake loss estimation appear to be applicable to estimating losses from major hurricanes. As such, Commerce should explore with FEMA the prospects for extending the HAZUS model to cover loss estimation from major hurricanes. The U.S. Department of Commerce might also find it useful to solicit comments on Table 2-2 and specifically should request interested parties to identify types of loss data that could be collected at reasonable cost following a major disaster. Similarly, the U.S. Department of Commerce could make use of simulation models such as the HAZUS model used by FEMA (see Box 2-1).

OCR for page 15
--> It is worth noting that a program for the collection of accurate natural disaster loss data must proceed on various time scales. That is, some post disaster information tends to dissipate rapidly, requiring a rapid response with professional expertise. Other loss information can take a long time to stabilize, requiring a long-term commitment to data collection. Finally, any agency charged with the additional data collection responsibilities recommended here should be given an appropriate level of resources to do the job effectively. BOX 2-1 Description of the HAZUS Loss Estimation Methodology FEMA and the National Institute of Building Safety (NIBS) have developed an earthquake loss estimation methodology entitled HAZUS (Hazards, U.S.). This software program was developed to help provide a standardized methodology for estimating the losses associated with earthquakes. FEMA began work on HAZUS in October 1992 and released the program in the spring of 1997. HAZUS uses mathematical formulas and information about building stock, local geology and the location and size of potential earthquakes, economic data, and other information to estimate losses from potential earthquakes. HAZUS a geographic information system (GIS) to map and display ground shaking the patterns of building damage, and local demographic information. Given a hypothetical earthquake event, HAZUS estimates (among other things) the violence of ground shaking, the number of buildings damaged, the number of casualties, and the estimated cost of repairing projected damage and other effects. Not only can HAZUS be used to estimate local impacts, it can be used to compare seismic risks across regions throughout the continental United States. HAZUS aims to provide local, state, and regional officials information to plan for earthquakes, mitigate against future losses, and prepare for emergency response and recovery. In addition, HAZUS may be used to prepare a quick loss estimate following an earthquake or to provide the basis to assess the nationwide risk of losses from earthquakes. RECOMMENDATION 2-2: The agency charged with the overall data collection should obtain insured paid claims data from available sources, such as PCS and IBHS. In addition, the agency should work with relevant trade associations, especially the National Association of Insurance Commissioners (NAIC), to obtain any additional data that may be useful. The NAIC could be instrumental in achieving consensus among state regulators to induce regulated insurance companies to provide the appropriate data. The agency charged with data

OCR for page 15
--> collection should do its best to avoid double counting losses reported by different sources. RECOMMENDATION 2-3: The agency charged with data collection responsibilities must also strive to collect data on losses incurred by uninsured individuals, businesses, and governments that are not otherwise reimbursed by disaster aid from some other level of government (typically the federal government). Several avenues for estimating the disaster losses absorbed by the uninsured should be explored, including: (1) post-event sampling (for very large disasters); (2) extrapolations from other data bases, such as the data compiled by the SBA from loan applicants who presumably are not insured or not well-insured; (3) data compiled by insurers and commercial sources of insurance data (such as PCS) that may indicate the amounts of deductibles absorbed by individuals and businesses who do receive insurance payments; and (4) extrapolations from insured claims data. RECOMMENDATION 2-4: The federal agency charged with the overall data collection effort should encourage and work with states and localities to collect disaster-related data. Such data need not be reported to the federal government after every disaster but could be reported annually. Congress could amend the Stafford Act to require such reporting as a condition for states and localities to receive federal disaster aid in the future, although such a requirement could be viewed as an unfunded mandate and thus subject to further analysis before being implemented. In addition, it is not clear if the federal government has the legal authority to audit any cost data submitted by the states and localities.14 This hurdle might be overcome, however, if the National Emergency Management Association (NEMA) developed ways to collect the data from the individual states. FEMA should work with NEMA to bring this result about. 14   States and localities are already required by the Stafford Act to submit estimates of damage immediately following an event for which they seek a presidential disaster declaration. These estimates, however, are necessarily based on the very limited data that are available at the time of such an event. The recommendation outlined above calling for annual reporting would require these jurisdictions to assemble more complete and accurate costs, at least for the expenditures they make on account of all disasters, including events that may not qualify for a presidential declaration but that nevertheless fit the definition set by the federal government for purposes of its comprehensive data collection program.

OCR for page 15
--> RECOMMENDATION 2-5: The Office of Management and Budget, in consultation with FEMA, should develop annual, comprehensive estimates of the payouts for all disaster costs incurred by federal agencies. As outlined above, these costs at the very least should be broken down into four categories: compensation payments to individuals and businesses (including the estimated subsidy cost of any loans designed to help cover disaster-related expenses); response costs; losses to government-owned infrastructure (including both state and local costs that are reimbursed by the federal government and damage to federally owned facilities and property); and payouts from federal insurance programs (with annual premium receipts being shown separately). It would also be useful if the data in each category were disaggregated by type of event and loss (such as losses to buildings, other infrastructure, and compensation for lost income, in the case of disaster-related unemployment insurance payments). Furthermore, to provide some perspective on current and future loss estimates, it would be extremely helpful if these data were assembled for some historical period. The results could be published in OMB's Analytical Perspectives that accompanies the annual budget. To carry out such an exercise it would be necessary for OMB (working with FEMA and the federal agency charged with the more comprehensive data collection effort) to develop a standardized definition of what events to include in the data base. One obvious defining characteristic could be dollar loss above a certain threshold, as determined by a preliminary assessment of direct losses (for example, by using loss estimation models such as HAZUS). In this connection, the relevant agencies should explore with organizations that currently maintain insurance claims data the feasibility of using different dollar cost thresholds than the ones they may currently be using (such as the $25 million per-event threshold now employed by PCS) for purposes of determining which events should be included in the data base. It may also be appropriate to add human losses to the defining criteria.15 The committee recognizes that because it may not be possible, or practical in light of the costs, to require all agencies immediately to work with a standardized definition. Accordingly, the use of any agreed upon definition should be phased in over some reasonable period. Once the standard becomes effective, it should not be costly for agencies to organize their data around it. If modest resources were appropriated, it would also be useful for the agencies to reorganize their historical cost data to be consistent, to the extent possible, with the new definitions so policymakers would have sufficient data from the past to make long-term projections. 15   The Stafford Act does not provide clear and objective criteria for establishing an appropriate threshold, however.

OCR for page 15
--> RECOMMENDATION 2-6: An effort should be made to collect loss data for direct primary and secondary losses separately. Secondary losses can be significantly affected by the availability and effectiveness of emergency response measures. Separate data on secondary losses can help policymakers to assess existing response measures, in design and in practice, and to develop improvements in the future.