2.0 SEISMIC RISK ANALYSIS METHODS IN A FEDERAL EARTHQUAKE INSURANCE PROGRAM WITH A LOSS-RIEDU ON COMPONENT1 Seismic risk analysis methods and their possible applications in a federal earthquake insurance program involving a loss-reduction component are central to this project Risk analyses are needed to answer such questions as: o Will the implementation of a loss-reduction measure be cost-effective and a sound policy decision? How is this affected by differences in seismic zone? o How will the implementation of proposed LRMs affect various stakeholders? o How can equity among stakeholders be assured in a rate-setting system for earthquake insurance? o To what extent can a rate-setting system for earthquake insurance consider both catastrophic losses that may occur in the near term and expected losses over the long term? This section uses findings developed in Eguchi et al., (1989) and in the project workshop in order to help answer these questions. by discussing: o the types of risk methods available and their applications and limitations., o risk methods appropriate for assessing LRMs and for use in setting insurance rates, and o risk mapping considerations. Section 2.1 outlines the processes and methods involved in seismic risk analysis. We maintain that two decades of science and engineering have reduced uncertainties in seismic risk analyses, but that systematic efforts, such as post-disaster damage and loss studies, are desirable to reduce uncertainties further (NRC, 1989; Eguchi et al.,1989. In FIA, 1971, un- certainties in seismic risk analysis methods were considered to eliminate any consideration of a federal earthquake insurance program with a loss-reduction component.) We illustrate the linkages between risk analysis methods, loss-contro, and insurance and then summarize project recommendationsr for uses of risk analysis. Section 2.2 describes the types of methods suitable for earthquake insurance rate- setting. Because such rating requires comparison of reserves with all possible earthquake losses affecting them, these are probabilistic methods, not probable maximum loss (PML) or 1. The reader who desires fuller details of the risk analysis methods discussed here may refer to the project report by Eguchi et al. (1989). 2- 1 other deterministic methods. Fuller accounts of earthquake insurance rating methods are provided in Eguchi et al. (1989) and Taylor, Hayne, and Tiliman (1990). Section 2.3 discusses pertinent mapping considerations. For the application of loss- reduction measures, we recommend initially small-scale consensus maps as shown in Figure 2-5. Progress on such maps may soon include information regarding local shaking hazards and seismicity. (See Whitman, 1989.) Hence, small-scale maps should be constantly reviewed and updated. In higher seismic zones (3 and 4), larger-scale maps of clear local potential ground deformation hazards may be suitable. Research on estimating probabilities of various degrees of permanent ground displacement and on associated losses is desirable to make microzonation more effective in a federal earthquake insurance program. At first glance, and even at a small scale, seismic maps for LRMs are different in kind from seismic maps for insurance rate-setting; however, making radically different maps for rates and LRMs would have severe administrative weaknesses in a federal earthquake insurance program. 2.1 Risk Analysis Methods: Applications and Limitations This subsection provides an overview of o steps required in risk analysis, o types of risk methods available, o major uncertainties of interest in a federal insurance context, and o recommended risk analysis methods and uses. In Section 3 we use risk analyses to identify cost-effective LRMs for incorporation into a federal earthquake insurance program. In Section 5, we show how risk analyses may be used to develop partially risk-based insurance rates as one major tool for the incorporation of LRMs into this federal program. A fuller discussion of detailed risk analysis methods available is found in the working report developed for this project. Basic Processes in General Risk Analysis Figure 2-1 provides an overview of the essential steps in risk analysis. 2-2 Identify and Define >a Assess Hazards Vulnerabilities Assess Risks Figure 2-1. General Risk Analysis Approach Exposure definition is an inventory process central to the collection of information for use in risk analysis. This involves use of a geographic system (e.g., longitude-latitude, Universal Transverse Mercator,, State Plane Coordinates, township-and-ranges, zip codes) in order to identify facilities (e.g., buildings) and record, other pertinent information such as facility descriptions, replacement cost estimates, building contents, number of occupants, and equity position. Hazard identification and assessment involves examination of the natural sources and effects that give rise to risks (e.g., such potential natural hazard sources as floods, earth- quakes, and hurricanes and such local hazards as liquefaction susceptible sites). Vulnerability analysis determines the degree to which exposures (e.g., buildings, their contents, business operators, and people) are susceptible to damage as the result of specific hazard occurrances at various degrees of severity. A quantitative account of vulnerabilities, as of hazards, is needed in order to produce a quantitative assessment of risks. Risk assessment takes into account identified exposures and their vulnerabilities to identified hazards. These processes are often used individually as low-cost means to make specific deci- sions. This is very appropriate when other elements of the decision are known within reasonable bounds. For instance, the strong motion hazards to a building and its contents in Los Angeles may be known well enough to warrant recommendation of anchorage to critical equipment; however, a specific large-scale program of equipment anchorage needs to be guided by risk-reduction analysis methods. 2-3 Under most circumstances, however, used in isolation, the individual steps in the risk assessment process provide ineffective approaches to guide and monitor risk reduction programs. The identification of exposures by itself indicates little or nothing about the presence of hazards or the vulnerabilities involved. Hazard assessments by themselves indi- cate little about the vulnerabilities of various exposures or the types of exposures. Vul- nerability assessment indicates little or nothing about the severities and likelihoods of the hazards or the values involved. Risk analysis methods are better suited to formulate, guide and monitor risk-reduction programs. As suggested in Figure 2-1, when used in risk analysis, these basic processes of exposure definition, hazard identification and assessment, and vulnerability definition are not con- ducted independently or even sequentially. The exposure definition process must be closely tied to the other steps in the risk analysis process so that the hazards assessed and the vul- nerabilities defined can be correlated with the facilities inventoried and the types of risk out- puts desired (e.g., risks to schools versus risks to residential construction). Thus, the inven- tory process is central to the management of information that can be used in risk analysis, and contains elements that are neither hazards nor vulnerabilities. Hazard identification and assessment becomes part of risk analysis and risk reduction efforts to the extent that (a) exposures are identified relative to hazards and (b) these ex- posures are determined to have vulnerabilities relative to the hazards. Quantitative risk es- timates are possible only if hazards are defined so that the likelihood of the hazard occur- rence can be assessed. For instance, identifying a region highly susceptible to landsliding is not enough to quantify risks. One needs to quantify the likelihood of landsliding either un- conditionally or conditionally as a result of triggering events and then apply these probabil- ities to existing structures in the landslide area. Risk analysis, then, employs the data accumulated through exposure definition, hazard identification and assessment, and vulnerability analysis to develop statistics that can be used to evaluate the extent to which specific exposures with their vulnerabilities are at risk to specific hazards. In the sense used here, risk assessment may terminate at the production of risk statistics or may involve a more complex procedure of assessing the significance of those statistics for risk evaluation and risk-control purposes. 2- 4 Types of Seismic Risk Methods Figure 2-2 summarizes types of risk methods available for analyzing LRMs and for earthquake rating purposes. I DETERMINSTIC PROBABIUSTIC 1 Deterministic Probabilistic SITE- SPECIFIC Site-Specific Site-Specific Methods Methods Deterministic Probabilistic MUULTI- Multisite Multisite SiTE Methods Methods Figure 2-2. Types of Seismic Risk Methods Deterministic methods are those in which "one or more earthquakes are postulated without explicit consideration of the probability that those events will occur" (NRC. 1989, p. 20). Often these postulated earthquakes are large, but deterministic methods may be used which postulate smaller magnitude events. (See Spangle, 1987; Algermissen et al., 1988.) Probabilistic methods include probabilities for the entire suite of earthquakes poten- tially damaging an exposure or group of exposures. Site-specific methods model each site as an independent exposure unit -- whether or not detailed geologic hazards and/or building vulnerability stu dies are performed. Multisite methods model exposures (buildings) at two or more sites as potentially suf- fering losses from the same earthquake. Examples of deterministic site-specific methods. include analyses of the potential damageability of a specific building relative to a selected "response spectrum". Examples of deterministic multisite methods include many regional studies which have examined in detail the consequences of postulating selected, often large-magnitude earthquakes. (See Alger- missen et al., 1973; Davis et al., 1982; Hopper et al., 1975; Algermissen et al., 1976; Alger- missen et al., 1988; and Spangle, 1987.) Analyses of multiple sites which predict probable maximum losses (PMLs) are instances of deterministic multisite methods. Examples of probabilistic site-specific methods, typically estimate expected annual earthquake-related losses. These estimates of annual earthquake losses are used to calculate 2- 5 "pure premiums" -- premiums which do not include loadings for administrative, claims ad- justment, and other costs and fees - for earthquake insurance rating of annualized benefits of proposed LRMs. (See Liu and Neghabat, 1972; Wiggins et al., 1974; Whitman and Cor- nell, 1976; Whitman, 1975; Taylor and Ward, 1979; Petak and Atkisson, 1982; and Boisson- nade and Shah, 1985.) Examples of probabilistic multisite methods simulate a large number of potential earthquakes affecting facilities at diverse locations. (See Algermissen, Steinbrugge, and Lagorio, 1978; Taylor et al., 1985; Ostrom and Gould, 1986; and Taylor et al., 1990.) As we shall see in the following discussion, these models are most valuable for estimating extreme losses in the short term as well as overall annual expected losses. As a consequence, while more mathematically and computationally sophisticated than other methods, these can yield important information related to rates and to various types of benefits from proposed LRMs. Similar methods were used by Friedman and Roy to develop rates for the National Flood In- surance Program. (See Kaplan, 1971-72.) Uncertainties in Risk Analysis in a Federal Insurance Context Twenty years ago uncertainties in risk- and loss-estimation methods made the actuarial basis for an earthquake insurance rate-setting system very dubious (FLA, 1971). Methods for estimating seismic risk have advanced over the last twenty years, but significant uncertainties remain. We regard the discussion of suitability of methods (Section 2.2 as an example) as key to removing some major ambiguities or uncertainties. As a consequence of considerable studies over the past two decades, the uncertainties in seismic risk analysis are no longer by themselves sufficient grounds for maintaining that earthquake perils are uninsurable. (For a much fuller discussion of the insurability issue see Butler et al., 1988.) Nevertheless, further advancements are needed if risk analysis is to meet the needs of a federal earthquake insurance program. In particular systematic efforts are needed to: o collect and publish post-earthquake loss dat o collect exposure data including building structure values and locations at risk, o assess probabilities of ground failures induced by liquefaction and landslides and to correlate to these phenomena both buildings and land values, o examine macroseismicity (seismicity on a regional basis and "rock" attenuation func- tions -- how seismic waves attenuate in amplitude as they pass through "rock"). o examine the feasibility of microzonation for relative ground motion site response fac- tors. 2-6 The relative importance of each of these types of information in risk analysis is a matter of some debate. Even though it has been maintained that regional seismicity factors are dominant in risk analyses, several different applications of seismic risk analyses have sug- gested that under some circumstances applications, of loss models for diverse seismic con- struction classes may be dominant. (See Algermissen et al., 1988; Taylor et al., 1988.) Under other circumstances (e.g. the 1989 lDoma Prieta earthquake and the 1985 Mexico City earthquake), relative site response factors become extremely critical. Hence, continued progress in macroseismi city studies is needed to remove uncertainties in seismr icity and at- tenuation parameters, but other parameters cannot be ignored and should be investigated. The need for systematic recovery and accumulation of earthquake loss data has been repeatedly emphasized throughout earthquake-related literature. (See NRC, 1989, and Wig- gins and Taylor, 1986.) For application to a federal program, current institutional con- straints on systematization and publication of existing loss data need to be overcome. Since inventorying can be mundane, costly, and time-consuming, many investigators of- ten fail to meet the critical need for systematic exposure data collection (NRC, 1989). Project workshop participants suggested that inventory data be collected as properties are included in a federal insurance program. Systematic development of basic exposure data on locations, values at risk, and general building structure types is critical for both primary and secondary federal earthquake insurance programs, since prospective liabilities in those programs cannot be assessed and effectively reduced without these data. As suggested in this report, current weaknesses in scientific models estimating losses caused by permanent ground deformation make it difficult to assess the probable success of LRMs associated with potential ground failures. These include virtually all landuse LRMs. In order to improve the state-of-the-art in landuse planning for earthquake loss-reduction, it is necessary for planners also to begin to consider how to reduce the effects of high relative site response factors (the likelihood that a site will respond severely to an earthquake), which are potentially very high contributors to risk. This consideration may enter indirectly in seismic building code provisions as knowledge regarding site factors evolves along with microzonation identifying relative site response factors. Linkages between Risk Analysis Methods, Loss-Reduction, and Insurance When risk analysis (as defined above) is used as a tool in decision-making, the com- bined process is known as risk and decision analysis. As illustrated in Figure 2-3, risk and decision analysis is central to addressing the objectives of,this project. 2-7 L Define Decision Alternatives F Assess Technically Develop Risk and Feasible Insurance Decision Loss-Reduction , Rates Measures Analysis Define Implementation Steps Figure 2-3. The Centrality of Risk and Decision Analysis in Linking Insurance and Loss-Reduction Measures Decision alternatives considered in this project include incorporation of proposed LRMs in the following contexts: o the status quo (no major program changes), o federal government as primary insurer, o federal government as secondary insurer, and o federal government as both primary and secondary insurer. In the assessment and selection of loss-reduction measures, risk and decision analysis methods assist in o estimating the aggregate benefits and costs of various proposed LRMs, and o estimating the benefits and costs to various key stakeholders. Thus, these applications are very important in assuring that monies are directed toward those activities that have the greatest returns. Moreover, analysis of diverse stakeholder benefits and costs is useful for determining what types of existing and potential incentive systems (e.g., policy instruments such as subsidies and/or tax breaks) may be used to remove possible inequities and to incorporate cost-effective LRMs under the four decision alternatives. These applications are elaborated further in Sections 3 and 4 of this report. 2-8 Risk analysis methods are also required to develop insurance rates. Although rate determinations depend on a number of reserve development and cost considerations which are beyond the scope of this project,, seismic risk analysis does play an important role in facilitating the relationship between insurance rates and LRMs. Some rate determinations can be counterproductive to instituting LRMs. From a broader perspective, risk analysis methods can be used as a tool to determine the trade-offs between costs of finer rate deter- mination.s and the benefits of loss-reduction. For instance, benefits of detailed mapping of poor site conditions and of establishing rate differentials based on site conditions can be compared to costs of developing and applying these rate differentials. Successful implementation of insurance rating schemes and of proposed LRMs depends on how both factors together affect governments, individuals, and other types of stakeholders. For instance, federally subsidized rates may be desirable if the loss-reductions otherwise involved offset potential federal liabilities. For another instance, finer rate deter- minations may be desirable if the resulting rates induce implementation of loss-reduction measures and if the costs of these LRMs are not unduly borne by any of the individual stakeholiders, including insurance companies, insurance regulators, and the general taxpayer. Implementation also requires monitoring of programs. Risk and decision analysis methods are needed both to predict and to evaluate the success of implementation of proposed rating schemes and loss-control measures. In sunmmary, risk and decision analysis methods serve as important tools in examining insurance rate-setting and LRMs within the context of a federal program. Nevertheless, the risk analysis procedures that we have been discussing are primarily directed toward specific actuarial, professional, economic, and technical considerations, with relevance to other fac- tors. For the purposes of this study, a broader approach which incorporates an even wider range of considerations was needed. The STAPLE (social, technical, administrative, politi- cal, legal, economic-- but also, as H. Kunreu-ther has indicated -- actuarial, procedural, ethi- cal, environmental, professional, and so on) approach discussed by Petak (1981) was used to ensure consideration of all relevant issues. The further application of risk and decision analysis methods to the selection of cost-effective LRMs is a major concern of Section 3 of this report. The variety of additional considerations such as political costs and legal con- straints that are involved in assessing diverse decision alternatives is considered in Section 4 of this report. Their application to insurance rate-setting is discussed in Section 5. 2-9 Recommended Uses of Risk Analysis Based on the foregoing considerations and on the project workshop, the following uses of risk analysis are highly recommended:, o providing the basis for benefit-cost analysis in order to determine the suitability of building practice LRMs for inclusion in a federal program; o providing nationwide estimates of losses and casualties and of reduced losses through unplementation of specific measures; o providing estimates for monitoring and modifying loss-reduction programs undertaken; o providing a basis for developing premiums, deductibles, and limits of liability for an earthquake insurance program; o providing a basis for proposed zoning schemes (macrozonation and/or microzona- tion) to be used in a federal program; o clarifying the degree to which one should place confidence in seismic loss estimates and perhaps improving the reliability of these estimates; o clarifying those parameters which contribute most to the uncertainty in seismic loss es- timates and suggesting various approaches to reducing this uncertainty. Although rated less highly in the project workshop, the following uses also deserve consideration: o providing estimates that serve as a basis for examining potential derivative losses to dif- ferent sectors of the economy (the stakeholder analysis in Section 4 is an example of this method); o examining in greater depth the earthquake damageability of critical or essential facilities and determining the most cost-effective means of reducing this earthquake damage (see especially Section 3.3); o providing a basis for estimating the cost-effectiveness of various landuse programs such as specific programs designed to reduce seismically-induced landslide losses (see Sec- tion 3); and o providing a basis for estimating time-varying reserves and subsidy levels within the con- text of an earthquake insurance program (see Section 2.2). 2.2 The Use of Risk Analysis in Insurance Rating Methods In this section we examine the application of risk analysis methods to evaluating in- surance rates under different federal involvements in earthquake insurance. 2 -10 Risk Analysis Methods Suitable for Use by Private Insurers Eguchi et al. (1989) and Taylor, Hayne, and TillInan (1990), have maintained that for private insurers (and reinsurers) probabilistic multisite methods (variants of ruin analyses used in actuarial science to ensure that insurance companies do not face high risks of insolvency) are most suitable for defining insurance rates (premiums deductibles, and limits of liability). Succinctly put, they argue mathematically that: o Probabilistic site-specific methods, which develop rates based on expected annual losses (as in Boissonnade and Shah, 1985), fail to take into account extreme short-tern fluctuations in earthquake losses. These extreme short-term fluctuations can lead to insurer and/or reinsurer insolvencies which in turn harm consumers and more prudent insurers/reinsurers. o Deterministic multisite methods, such as "probable maximum loss" (PML) methods in Anderson, 1981, add some information regarding possible extreme events. But these PML methods do not by themselves necessarily cover expected losses, and the ratio of PML to expected annual loss varies too extremely -- by a factor of twenty across dif- ferent regions of the United States -- for PML to be an equitable basis for rate-setting. Indeed, adverse selection may be encouraged by a PML rating method inasmuch as higher risks have lower ratios of PML divided by expected annual loss. The informa- tion provided by these PML methods, even with respect to extreme events posing short- term solvency concerns, can be made much more coherent through probabilistic methods. o Products of probabilistic multisite methods (see Figure 24 for examples) can be used to determine rates relative both to short-term fluctuations (the catastrophic element needed to assure that contractual obligations with consumers are fulfilled) and the ex- pected loss element (needled to cover costs). (Taylor et al., 1985, justify the use of 2- 4(b), a model better known for application to individual sites, for application to mul- tiple sites.) Thus, it can be concluded that private insurers (status quo) currently benefit froom probabil- istic multisite mrethods. which incorporate expected annual loss and catastrophic loss poten- tial into a coherent framework. 2- 11 a1) R -J -tc .2 0 a- 2 l SI 2 L-Q /2 Loss Level Distance of Rupture Center Along Fault Trace (b) Illustrative Loss Levels for a Portfolio Relative to a (a) Illustrative Probabillity of Loss as a Function of Loss Level Output Specific Fault Trace and Magnitude of Interest (L is Total Fault Length; .Q is Rupture Length) U) 0 2-1) a 0 U) a) 0.g 0 V cr (L V Time Time (c) Illustrative Simulation of (d) Illustrative Simulation of Effect of Portfolio Losses Over Time Portfolio Losses on Reserve Figure 2-4. Illustrative Probabilistic Multisite Analysis Outputs Risk Analysis Methods Suitable for Use with Federal Insurance With direct federal involvement in earthquake insurance, the above conclusions may need to be revised in view of public policy considerations required to guide actuarial rate assessment work. Public policy analysis is needed to determine which of the following will guide actuarial analysis: o The net reserve is positive trending (premium income exceeds expected program costs including expected earthquake losses). o The net reserve is zero. o The net reserve is negative-trending (a subsidy situation). Moreover, public policy analysis is needed to assure that short-term potentially extreme fluc- tuations in losses can be handled within the political system (e.g., that these extreme fluctua- tions will not create extreme program instabilities). In addition, detailed public policy and economic analysis would be needed to assess possible regressive or progressive effects of various rate structures such as effects of specific proposed rate structures on low-income citizens. (No systematically documented work on this subject is known to the project direc- tor.) Obviously, public policy evaluation of these topics will depend on the type of federal in- surance involvement considered (as a primary insurer, as a secondary insurer, as a banker" of last resort) and the distribution of and types of properties covered. A mandated program, with monopolistic implications, will differ from a voluntary program. A residential program will differ from a commercial program. With respect to a mandatory priAmary federal earthquake insurance involvement (as for selected classes of buildings) expected annual loss (probabilistic site-specific] methods may suffice in order to develop crude approirmations for earthquake insurance rates. As ex- plained in Eguchi et al. (1989), although these expected loss methods contain important un- certainties and are often performed incorrectly, they include advantages derived from many developments originally designed for the building code and nuclear power plant industries. (See Algermissen, 1983; Bender and Perkins, 1987; McGuire et al., 1988.) Probabilistic m.ul- tisite methods are helpful in studying extreme short-term loss fluctuations which may be sig- nificant in a mandated primarv progam. but such results will be less important for required programs than for voluntary programs. 2- 13 In voluntary primary programs, problems of adverse selection can easily arise if in- surance rates do not adequately reflect risk levels. This is particularly the case when the in- sured knows more about his risk than the insurer. Cases of adverse selection may involve in- dividual exposures, such as when the insured is in a highly risky location and the insured building is more vulnerable than the "typical" building used for rate-setting. Cases of "adverse selection" in a broader sense may also involve aggregate (more than one) ex- posures, such as when many property owners in one small seismic region purchase insurance and so create a catastrophic loss potential that has not been anticipated fully in rates. Problems of adverse selection are reduced to the extent that rates reflect risk. Nevertheless, the "adverse selection" or unanticipated catastrophic element involving aggregate exposures still poses problems for a voluntary program. To anticipate this latter problem, probabilistic multisite methods are most useful in rate development for voluntary primary federal in- surance programs. Secondary federal earthquake insurance involvements will typically require examina- tion of higher earthquake loss-potentials. Hence, in the typical case, the suitable methods to be used will be those that consider the extreme tails of the distribution suggested in Figure 2- 4(a). Expected annual federal liabilities can be calculated from such a distribution, and so can be used along with administrative and other cost factors to estimate expected aggregate prices for federal involvement. Thus, probabilistic multisite methods are desirable for as- sessing potential federal liabilities (if any) in a secondary federal program. In order to apply these methods, one critical factor will be inclusion of inventory data on o values at risk, o locations, and o general types of structures. Although inventory problems can be severe, and aggregation methods will be needed (e.g., to estimate exposures by zip code or by county), workshop participants suggested that these in- ventories can be developed at the time of purchase. This proposal appears to apply best to a primary earthquake insurance program. Development of exposure data to support risk analysis in support of a secondary program presents a formidable task, given the obstacles that currently exist to systematic recording and transfer of exposure data within the insurance and reinsurance industries. (See Section 5 for information on a secondary federal earth- quake insurance program.) Also essential to the implementation of these methods are 2 - 14 o use of earthquake source zones (and further examination of sensitivities of rates developed to alternative selections of source zones), o use of "rocklr attenuation functions for different regions of the country, and o use of soft soil factors (whether to microzone or merely to assess more accurately overall potential federal liabilities). The first two factors have been well developed in procedures for model building code organizations and the nuclear power industry. The third factor is included by the project director on the grounds that soft soil site dynamic amplification factors have contributed sig- nificantly to the degree of damage resulting from varying shaldng intensities in many past earthquakes (see Hays, 1980, 1981). In regions of very low seismicity,, sensitivity analyses may be used to determine whether or not rates would fluctuate significantly should these soil factors be included. In regions of higher seismicity (certainly zones 3 and 4) microzonation of soft soils is needed to determine aggregate earthquake loss potential within these zones. In summary, deterministic methods, including probable maximum loss (PML) ap- proaches, have extremely limited utility for both socioeconomic evaluations. of LRMs and earthquake insurance rate-setting. For most earthquake insurance rating purposes, prob- abilistic multisite methods are appropriate. Under a mandatory primary national program, probabilistic site-specific methods may be adequate for basic rate-making even though mul- tisite methods would be desirable to assess short-term expense fluctuations in losses. Further applications of these approaches are discussed in Section 5. Mapping Considerations 2.3 In this section, we discuss briefly mapping considerations in a federal earthquake in- surance program involving LRMs. We first discuss the sorts of small-scale maps used for building codes. Next, we discuss generally larger scale maps for lLRMs. Third, we discuss how rating maps could diverge significantly from maps for LRMs. We maintain that the ad- ministrative difficulties of employing two widely divergent sets of maps, one for LRMs and another for rating, provides a strong reason for developing to the extent possible a single set of maps for both LRMs and rating. Seismic Zone Maps Figure 2-5 is an adaptation of one small-scale consensus seismic zone map. Tlhe original map uses probabilistically derived contours of strong ground motion values (in this 2- 15 mtar II= * EXPLANATION For Adapting Figure C-1, We Have Used the Following Relationships Between Seismic Zones and Peak Horizonta Ground Accelerations: Seismic Zone Peak Ground Acceleration Shaking (Percentage of Gravity) Hazard Level 0 < 0.05 Lowest 1 2 0.05 and < 0.1 2 2 0.1 and<0.2 3 2 0.2 and c0.4 V.&U 4 2 0.4 Highest Figure 2-5. Illustrative Seismic Zone Map for the United States (Adapted from "NEHRP Recommended Provisions for the Development of Seismic Regulations for New Buildings," FEMA Publication 18 by the Building Seismic Safety Advisory Council, 1988) 2 - 16 case, peak ground acceleration values derived by conversion from peak ground velocity values). Scientifically derived contours are subjected to a consensus process that assures consideration f a wide variety of scientific, engineering, and bulilding construction practices within the contours produced. These consensus maps are designed primarily for uses in design and in construction and in the development of seismic building codes. The adaptation in this report has resulted in seismic zone designations O.1, 2, 3, and 4. The resulting map in Figure 2-5 differs from other seismic zone maps. For instance, seismic zone designations in the 1988 edition of the Uniform Building Code will be higher in Oregon, portions of Alaska, California, and other states west of the Rocky Mountains. Since we do not urge modifica- tions of the seismic provisions of the Uniform Building Code, Figure 2-5 can serve as a min- imum initial basis for the LRMs proposed in this report. However, in a federal earthquake insurance program, development of consensus seismic maps will be a first priority both for insurance rating and LRMs. From the standpoint of science and engineering, the basis for such maps has evolved significantly over the past twenty years, especially as a result of the national hazard mapping program and also as a result of special studies of major sites and facilities. Continued re- search (see Whitman, 1989) is ongoing in an attempt o to combine geological and historic earthquake data, o to estimate "rock& attenuation patterns, and o to estimate effects of soft soils and special geologic configurations in amplifying the effects of strong ground motions. In regions with limited seismicity, or where historic seismicity may not be an adequate clue to expected seismicity, estimates of seismicity may vary, and these may lead to large variations in loss estimates. Thus, in the administration of a federal earthquake insurance program, updates of seismic hazard information and loss estimates will be needed periodically as scientific and engineering knowledge grows. Also, it should be noted that the effects of these variations on proposed rates may be as significant in moderate-to-high seismic zones (zones 3 and 4) as they are in lower seismic zones. In lower seismic zones, costs of underwriting can become a significant consideration in rate development; that is, given the low expected an- nual losses estimated in these regions, minimum rates may be required merely to cover costs of writing the insurance. In using such maps, administrators must be aware of the constant possible progress in refining such maps. Specifically, incorporation of soft soil site dynamic amplification factors 2-17 into such maps is imminent. (See Whitman, 1989.) In the near term, new seismic maps are expected to be developed that reflect o additional studies of seismicity, including studies of the eastern U.S., and o relative site-response factors critical in assessing potential losses. As a result, federal officials should be aware that seismic maps are constantly being aug- mented and that refinements in the next few years likely will be more consistent with the LRMs proposed in this report. While all such new maps developed undergo consensus processes which in many cases may take decades to implement into building and landuse practices, they should be readily available for use in implementing LRMs and evaluating in- surance risks. Mapping Local Hazards Although relative site response factors are becoming integral to seismic zone maps for building codes, and although this information is critical to estimating losses (as discussed in Section 2.1 and Eguchi et al., 1989), the use of ground failure potential maps for purposes of local loss-reduction is far more ambiguous. While techniques are currently available for delineating surface fault rupture, landslide, liquefaction, and subsidence local hazards, am- biguity results from o the lack of data with which to estimate losses associated with various types of ground failures (except in clear cases such as when total constructive loss would be expected), and o the lack of convincing methods to calculate on a regional basis the probability of ground failure or its severity. As an example, with respect to liquefaction zones, it is currently not known how much poten- tial ground deformation can be expected given earthquake magnitudes of less than 6.5. (See Taylor et al., 1988.) Yet, since occurrences of magnitudes below 6.5 are far more likely than those above 6.5, the contribution of these lower magnitude levels to expected losses is very great. (See Taylor, Atkisson, and Petak, 1981.) Likewise, seismic risks from landslides have not been examined thoroughly. Furthermore, mapping on a local level appears to be expen- sive. We recommend investigations to develop economical mapping procedures before any large-scale mapping program is initiated. Validation of the assumption that old landslides serve as good indicators of potential earthquake-induced landslides could, for instance, produce economic means for mapping potential earthquake-induced landslides. 2- 18 In spite of these caveats, it is clear that permanent ground failures do cause losses and that there are clear-cut cases of very high potential for such ground failures. Adverse selec- don can result if these local hazards are ignored in insurance rate-making. Hence, in Section 4, we maintain a pragmatic attitude toward these hazards, and recommend initial compila- tion of existing sources of information and an intermediate scale (not larger than 1:24,000) mapping program only in high risk seismic zones (3 and 4). These supporting activities are modest in cost given projects already completed through the United States Geological Sur- vey and through state and local programs. Review of findings on probabilities and potential extent of these hazards, and also on losses associated with these hazards, will assist in later program developments. Mapping and Insurance Rating An earthquake insurance program could be designed in ways in which maps for LRMs diverge somewhat significantly from maps that provide insurance "rating zones." Unlike the National Flood Insurance Program, which begins with large-scale maps and integrates them into the insurance rating process, an earthquake insurance program may begin with small- scale maps. Public policy considerations must be addressed before maps for earthquake in- surance rating can be developed. For instance, chiefly for political, jurisdictional, or administrative reasons, it may be decided that rates should be developed relative to political boundaries that do not cross- subsidize each other. If probabilistic multisite approaches are used, then the catastrophic loss potential in a given territory will have significant impact on how rates are developed. Moreover, key decisions affecting rate-setting in a federal insurance program will per- tain to o how detailed the niicrozonation should be for rate development, o how detailed the seismic building vulnerabilities should be (what proxies may be used to estim ate roughly these vulnerabilities, which may be very great for the large majority of buildings in the United States, especially older buildings or buildings located east of the Rockies), and o whether a subsidy situation, a net zero financial gain, or a surplus reserve is desirable. Before such issues are addressed, it is unclear what insurance rating maps should look like. For instance, if seismic building vulnerabilities vary from region to regiorn then contours of expected or average annual losses (so-called pure premiums, or premiums without catastrophic loadings -- the terminology varies among actuaries) will not necessarily be iden- tical with contours ofprobabilistic peak strong ground motion values. If political boundaries 2-19 are used with different catastrophic loadings for different-territories, then contours of basic premiums (pure premiums loaded to account for catastrophic occurrences) may differ from contours for expected annual losses. Moreover smoothing techniques may be used that produce jurisdictional rating maps rather than maps based primarily on loss potential. Hence, mapping for earthquake insurance rating purposes will not necessarily mirror map- ping for LRMs. Various administrative pitfalls occur if maps for LRMs diverge too greatly from maps for insurance rating. The first is internal - the program would involve separate functions, each producing maps of a very different sort and hence not working on a common basis. Use of a risk basis for maps would avoid this internal administrative problem. The second is ex- ternal -- it would become readily apparent to the public-at-large and to their representatives that maps for rating did not correspond to maps for LRMs. Hence, for administrative simplicity, maps for LRMs should correspond as closely as possible to maps for rates. We later conclude that to encourage cost-effective LRMs, partially risk-based rates are desirable. A risk basis for both LRM maps and rating maps would offer a further very large administrative advantage. 2.4 Conclusions In this section we conclude that: o Although uncertainties still characterize seismic risk analysis methods, these uncertainties have been significantly reduced through many studies over the past two decades so that seismic risk methods are now acceptable for use in a national earthquake insurance program. o Deterministic methods including probable maximum loss (PML) approaches have ex- tremely limited utility in either socioeconomic evaluations of LRMs or earthquake in- surance rating. For most earthquake insurance rating purposes, probabilistic multisite methods are appropriate. Under a mandatory national program, site-specific probabil- istic methods may be adequate for basic rate-making even though multisite methods would be desirable to assess short-term fluctuations in losses. o Small scale seismic zone maps at a national or regional scale are critical to the implemen- tation of LRMs. Various theoretical and practical limitations exist to large-scale 2-20 micozonation mapping, but these may soon be largely overcome through research tar- geted to resolving issues related to assessing probabilities of permanent ground displace- ments. Severe administrative disadvantages occur if maps for LRMs, especially at a small scale, diverge too significantly from maps for rating. 2-21 THE IDENTIFICATION, DEVELOPMENT AND ECONOMIC EVALUATION 3.9 OF PROMISING LOSS-REDUCTION MEASURES The primary objective of this project is to identify feasible alternative earthquake loss- reduction provisions and develop a strategy to FEMA for incorporation of recommended loss-reduction provisions into a national earthquake insurance oT reinsurance program. In order to achieve this two-part goal it was necessary first to identify earthquake loss- reduction provisions for recommendation. To be recommended, a loss-reduction measure muist be technically sound, cost-effective, and otherwise acceptable to a wide variety of in- dividuals who will be affected by the measure. 'This section reviews the procedures used to accomplish this task and presents the results of efforts to identify and develop cost-effective loss-reduction measures, (LRMs) which could serve as loss-reduction provisions within the context of a national insurance or reinsurance program. Specifically, 3.1 summarizes the process used to identify feasible earthquake loss-reduction provi- sions. reviews the results of efforts to identify hazard reduction activities for candidacy as 3.2 feasible loss-reduction measures (LRMs). reports the findings of socioeconomic analyses conducted to determine whether or 3.3 not these candidates could be considered cost-effective. Section 4 summarizes the process used to evaluate the '"acceptability" of these measures, presents the resulting set of feasible loss-reduction measures (LRMs) and neces- sary supporting activities, and reviews issues of acceptability as discussed during the course of this study. Throughout this process the STAPLE (Social, Technical, Administrative, Political, Legal, Economic) approach was utilized to ensure that (a) a wide range of available earth- quake hazard reduction activities was considered and (b) the LRMs. developed were analyzed and evaluated from all necessary perspectives. Section 5 will sort out the various ways in which cost-effective LRMs or earthquake or- dinances may fit into various types of federal involvements. The Process of LRM Identification and Evaluation 3.1 In order to achieve the primary objective of this study, a significant effort was directed toward identifying, defining, and assessing possible loss-reduction provisions appropriate for inclusion in a national program for earthquake insurance or reinsurance. The steps used in this process are illustrated in Figure 3-1. 3 -1 1. Identify earthquake hazard reduction activities through a thorough information search. 2. Select and restate a smaller number to be subjected to more thorough analysis a : 3. Analyze these activities on socioeconomic grounds n~~ 4. Present initial results at a Project Workshop for final development of LRMs and further analysis by knowledgable and interested parties. .~~ S. Present results for future analysis at the final Advisory Panel meeting and to other reviewers of draft final reports. I Figure 3-1. Steps Used to Develop and Assess Recommended Loss-Reduction Measures 1. Conduct a thorough information search to identify hazard reduction activities which could be used as components of a loss-reduction program. The very thorough infor- mation search was designed to assure that a wide variety of sources was considered from the growing literature on earthquake loss-reduction. Additionally, the very thoroughness of the information search was designed to provide a resource for future efforts to re-examine pos- sible efforts. This search produced a collection of ninety-six earthquake hazard reduction ac- tivities for consideration. 3 -2 2. Reduce the number of possible earthquake hazard reduction measures and restate them as needed so as to facilitate analysis and further development. In the second stage of this process, the project team used expert judgment to select, synthesize, and restate hazard reduction activities as a smaller number of more technically feasible loss-reduction provi- sions.. 3. Perform socioeconomic risk-and-decision analyses of the technically feasible loss- reduction activities. Questions of practicality were of utmost importance in the third step of the process. In this step the technically feasible loss-reduction activities developed in Step 2 were subjected to socioeconomic analyses in order to determine their level of cost- effectiveness (economic efficiency and their economic impact on various stakeholders (economic allocation). 4. Present the preliminary results of technically and economically feasible loss- reduction activities at a Project Workshop for further revision, refinement, and definition as loss-reduction provisions within the context of a national earthquake insurance program. Once the list of possible loss-reduction activities had been limited to those identified as cost- effective, they were synthesized and restated. A report of this work was generated and sent to Project- Workshop participants (knowledgeable individuals representing a variety of stakeholder interests). At the Workshop, participants were asked to review and critique the socioeconomic analyses -and then re-examine, refine, and augment the candidate loss- reduction measures in the light of their various interests. The collection of promising loss- reduction activities was condensed and synthesized into fifteen Loss Reduction Measures (LRMs) with corresponding Supporting Elements. Discussions evaluating the acceptability of these measures from the perspectives of various stakeholders addressed risk analysis, in- surance, and public policy issues. S. Present initially recommended loss-reduction measures at a final advisory panel meeting and in two fmal draft reports for further review and revision. Workshop findings resulting in fifteen initially recommended LRMs were developed in a preliminary final draft report and presented at the third advisory panel meeting for review. A final draft report was also submitted for review by advisory panel members. This process resulted in restatement and elaboration of many of the LRMs to make them more acceptable in the light of objec- tions raised. The remainder of this section discusses. in further detail each of the processes used to identify technically feasible loss-reduction measures and to evaluate these technically feasible LRMs on the basis of cost-effectiveness. 3- 3 3.2 The Identification and Selection of Technically Feasible LRMs Information Search As a starting point for accomplishing the tasks required by this project, a comprehen- sive search of earthquake-related information was conducted to identify earthquake hazard reduction activities which could satisfy the definition of loss-reduction measure as an activity involving a physical intervention or restraint thereon that reduces expected building losses associated with earthquakes. In order to assure thoroughness we developed a data collection and evaluation (DACE) instrument through which each activity was characterized by five descriptors. These could later be used to sort and review results to assure that all aspects of the descriptor had been considered in the course of the search. Thus, each hazard reduction activity was characterized by: o a general description of the earthquake hazard reduction activity in terms of (a) a physical description of the fix", which qualifies the activity to be considered as a possible LRM as defined in section 1.3, (b) specification of the activities necessary to support the LRM (e.g., maps, building inventories), and (c) identification of the methods needed to verify implementation (e.g., engineer sign-off, random sampling techniques, risk analyses, public policy analyses), o the source document(s) used to identify and develop the earthquake hazard reduction activities, includ- ing - library searches [in-house Dames & Moore; Natural Hazards Research and Applications Informa- tion Center (NHRAIC) in Boulder, Colorado; the National Center for Earthquake Engineering Re- search (NCEER) Information Service in Buffalo, N.Y.; and the Earthquake Engineering Research Center (EERC) Library at Richmond, California], - legislative, bibliographic, and other materials developed from a number of other federal, state and local sources suggested by the advisory panel and those listed in the acknowledgments, and - reports from projects previously developed by project team members and advisory panel members. o general type of application (e.g., landuse and building practices; residential, commercial, and critical public building usages) to assure that all applications required by the project (see section 1.3) were addressed, o type of structure affected to assure that all main building types were covered, and o region of application as defined by seismic zone, to assure that all regions of the country were covered, Ninety-six earthquake hazard reduction activities were identified and stated in great detail. A full report of these findings would be useful for those interested in re-examining a broader range of possible loss-reduction activities. Their importance to this project was primarily to ensure a thorough preliminary basis from which promising loss-reduction measures and their supporting elements could be identified, systhesized, and developed. As a result, they are not included in this report. The remainder of this section reports the results of this process. 3-4 Application of Engineering, Scientific, and Risk Criteria to Identify Possible Loss-Reduction Activities In the next step, the project team used expert judgment in order to produce a shorter list of possible loss-reduction measures that could be subjected to a risk-and-decision analysis. In many cases, the information search identified overlapping activities that could be synthesized into a fewer number of possible loss-reduction measures. In other cases, ac- tivities were eliminated for one of many possible reasons, including the judgment of geotechnical and o lack of technical feasibility (as detersined4 for instance, through structural engineers on the project), o obvious lack of political feasibility, o narrowness of application relative to a national risk-reduction or insurance program (eg., the ac- tivity applied to a limited number of buildings or in such rare circumstances that state and local ad- ministrators would tend to ignore them), o imprecision of pertinent technical description (e.g., administrators would not know how to imple- ment the earthquake hazard reduction activity), o absence of technical evidence that losses would significantly be reduced through the implementation of the putative hazard reduction activity (eg., for exsting steel-frame buildings, technically possible seismic retrofits may not improve seismic performance significantly), and o obvious lack of cost-effectiveness (e.g., practitioners currently do not use this measure because more cost-effective techniques are available). Once a shorter list of activities was produced, these activities were restated in order to facilitate socioeconomic analysis. Candidate activities naturally fell into two groupings -- Landuse Measures and Building Practices. Under landuse measures we included various practices, such as site preparation, to minimize or avoid geological hazards. Under building practices we included various measures, to modify the building structure through design or redesign. This distinction between landuse measures and building practices is well- recognized in different bureaucratic and professional functions, with site preparation ac- tivities forming one link between the two types of activities. In section 5 we use the expres- sion "earthquake ordinances" in order to facilitate greater administrative linkage between landuse measures and building practices. The following two subsections present, respectively, the landuse measures and building practices, that were subjected to socioecononiic analysis. 3-5 Landuse Measures Table 3-1 lists the candidate loss-reduction measures related to landuse practices that were submitted to socioeconomic analyses. Seismic zones specified in these tables are il- lustrated in Figure 2-5. Landuse measures generally involve policies, ordinances, and legisla- tion regulating three classifications of landuse planning: o regulation of new development of sites, o removal/conversion of existing structures and sites, and o use of appropriate engineering and construction measures to mitigate damage for both new and existing developments. Policies, ordinances, and legislation to be applied to new development include o development of public information and notification programs to place developers, realtors, and pur- chasers on notice of existing hazards; o discouragement of development in potentially hazardous (e.g., poor soil and/or ground-failure prone) areas by means of disclosure and/or such financial mechanisms as disallowance of tax deductibles after losses or special local taxes; o regulation of new development through subdivision regulations, grading ordinances, hillside-develop- ment regulations, and other zoning specifications; o state legislation related to liquefaction and landslide hazards similar to the 1972 Alquist-Priolo Special Studies Zones Act1 , which covered surface faulting. Policies, ordinances, and legislation regulating removal/conversion of existing development should ensure o acquisition, exchange, and relocation of properties in high hazard locations (e.g., high liquefaction sus- ceptible sites), o discontinuance of non-conforming land uses, o removal of unsafe structures through public nuisance abatement powers, o use of a non-conforming building ordinance requiring eventual removal of structures in the greatest danger, and ,,..A. rlceta. raul~vnalnnm..nt * kYV,3L-%bLQ SI UGVGs" %J 1. California Senate Bill 520 (1972), the Alquist-Priolo Geologic Hazard Zones Act, requires the State Min- ing and Geology Board to prepare policies and criteria for the development of areas encompassing major ac- tive fault traces, which are to be mapped by the state geologist. It was amended in 1975 to require disclosure to prospective buyers (see Palm, 1981). 3 -6 Table 3-1 Landuse Measures Analyzed Number Brief Description 1000 Purchase (if needed) existing construction or properties in very active fault zones of deformation (hence in seismic zone 4) and convert to low-density purposes or open space only. (The expression "very active is defined operationally in Appendix B.) 1100 Purchase (if needed) existing construction or properties in moderately active fault zones of deformation and convert to low-density usage or open space only. (The expression "moderately activew is defined operationally in Appendix B.) 120 Restrict new development in very active fault zones of deformation (in seismic zone 4) to low-rise residential construction. (Assume that residences and other construction would be designed to seismic code.) 1300 Restrict new development in moderately active fault zones of deformation in seismic zone 4 to low-rise residential construction. (Assume that residences and other construction would be designed to code.) 1400 Restrict new development in moderately active fault zones of deformation in seismic.zone 3 to low-rise residential construction. (Assume that residences and other construction would be designed to code.) 1500 In seismic zone 4 as deemed appropriate by geotechnical engineers prior to development, drive piles, use vibro-compaction, or use dynamic deep compaction in order to minimize potential ground failures owing to liquefaction. (Assume that seismic codes are adopted and enforced.) 1600 In seismic zone 3 as deemed appropriate by geotechnical engineers prior to development, drive piles, use vibro-compaction, or use dynamic deep compaction in order to minimize ground failures owing to liquefaction. (Assume that seismic codes are adopted and en- forced.) 1700 In seismic zone 4 restrict new development in very susceptible liquefaction zones to low- rise residential structures. 1800 In seismic zone 3 restrict new development in very susceptible liquefaction zones to low- rise residential structures. 1900 In seismic zone 4 allow major modifications of existing structures in very susceptible liquefaction zones only for which suitable geotechnical techniques can be used to mini- mize hazards resulting from ground failures. 2000 In seismic zone 3 allow major modifications of existing structures in very susceptible liquefaction zones only for which suitable geotechnical techniques, are used to minimize hazards resulting from ground failures. 2100 In seismic zone 4, in very susceptible landslide locales, restrict new development to open- space uses. 2200 In seismic zone 3, in landslide locales, restrict new development to open-space uses. 2300 In seismic zone 4 purchase (if necessary) land and/or severely damaged construction and convert existing development in very susceptible landslide locales to open-space uses. 2400 In seismic zone 3 purchase (if necessary) land and/or severely damaged construction and convert existing development in very susceptible landslide locales to open-space uses 3-7 Policies, ordinances, and legislation regulating engineering and construction measures used to mitigate damage should include o measures to protect existing development from damage in landslide areas, including measures for slide and slump control, mud- and debris-flow control, and rockfall control; and o measures to protect existing development in high liquefaction susceptible areas including nonstructural solutions, improving the soil in-place, and changing the project structure. Few very active fault zones of deformation exist in the United States. For operational purposes, we have defined these as having surface expression at any site along the fault every five hundred years or less. Chief examples of these very active fault systems include the San Andreas, Hayward, and San Jacinto fault systems, but not the Wasatch or Newport- Inglewood fault systems. "Blind" thrust fault systems, such as generated the 1987 Whittier Narrows earthquake, and offshore fault systems are excluded from consideration inasmuch as they do not generate surface faulting hazards. For very susceptible liquefaction, landslide, or settlement zones, we refer to such zones as Turnagain Heights in Anchorage, the Marina District in San Francisco, sites adjacent to the Duwamish River in the Puget Sound, sites ad- jacent to the Jordan River or Great Salt Lake in the Wasatch Front region, and similarly severe landslide zones. The emphasis of this report is toward site preparation and/or avoidance of "severe" ground failure susceptibilities since current evidence does not suggest large risks in zones with moderately susceptible ground failure hazards. At the same time, risks of building losses from strong ground motion are emphasized in this report. This em- phasis is not merely presupposed, but also is tested in terms of the cost-effectiveness analysis in this section (with greater detail in Appendix B). Building Practices Resulting in Loss Reduction Table 3-2 lists the candidate activities related to building practices that were submitted to socioeconomic analyses. Again, seismic zones specified in these tables are illustrated in Figure 2-5. In describing building practices that augment current efforts toward loss-reduction, we have assumed that for the most part adequate seismic building codes are in effect in seismic zones 3 and 4 west of the Rocky Mountains. In selected municipalities and possibly following a disaster, this assumption may prove to be untrue. (See Selkregg et al., 1984, p. 185, for seismic code disputes even in seismic zone 4.) Nevertheless, we further propose for analysis purposes that the two eastern Model Codes -- the BOCA National Building Code (formerly the Basic Building Code) published by BOCA and the Standard Building Code (SBC) pub- lished by SBCCI -- incorporate by transcription the National Earthquake Hazard Reduction 3 -8 Table 3-2 Building Practices Analyzed Number Brief Description 100 In seismic zone 2, design new comm ercial buildings in accordance with adequate seismic code provi- sions 110 In seismic zone 3, design new commercial buildings in accordance with adequate seismic code provi- Sions 12 In regions of seismic zone 2 where high catastrophic loss potential exists (zone 2*), require seismic zone 3 detailing requirements for new commercial and governmental buildings 13 In seismic zone 2, design new residential buildings in accordance with adequate seismic code provi- sions 140 In seismic zone 3, design new residential buildings in accordance with adequate seismic code provi- sions 150 In seismic zone 2, design new governmental buildings in accordance with adequate seismic provi- sions 160 In seismic zone 3, design new governmental buildings in accordance with adequate seismic provi- sions 170 In seismic zone 2,seismically retrofit all potentially hazardous- buildings 180 In seismic zone 3,. seismically retrofit all "potentially hazardous' buildings 190 In seismic zone 4, seismically retrofit all "potentially hazrdous" buildings 200 In seismic zone 2, seismically retrofit all poorly anchored or poorly reinforced wood frame dwellings 210 In seismic zone 3, seismically retrofit all poorly anchored or poorly reinforced wood frame dwellings 220 In seismic zone 4, seismically retrofit all poorly anchored or poorly reinforced wood frame dwellings 230 In seismic zone 4, reinforce/anchor/or brace parapets and ornaments to withstand 50 percent of seismic design forces specified in a current model code 240 In seismic zone 4 require that gas water heaters be strapped to structural framing in all multi-family dwellings and apartments (two-family dwellings and above) 250 In seismic zone 3, seismically anchor or restrain all life-safety related equipment (see Table 3-4) in essential buildings, public schools (including colleges), buildings whose primary function is assembly for more than 300 people, other public buildings, and for buildings with more than 500 occupants 260 In seismic zone 4, seismically anchor or restrain all life-safety related equipment in buildings referenced in number 250. 3-9 Program (NEHRP) seismic design provisions. Program administrators may in the future consider whether or not 1992 ASCE/ANSI provisions satisfy national seismic design code requirements, and so can be regarded as the equivalent to incorporation of NEHRP seismic provisions. Uniformity of seismic design provisions is very important for (1) removing competition among jurisdictions (often neighboring) with respect to seismic codes in place and (2) providing greater uniformity of standards for reducing underwriting costs for writing earthquake insurance. Since many municipalities regularly adopt one of these two model codes, incorporation of adequate seismic provisions in these model codes would be an im- portant step toward their implementation in local municipalities throughout the country. Since losses are reduced only if these provisions are adopted and enforced, the project team suggested for evaluation a loss-reduction measure that uniform seismic design provisions be adopted and enforced, especially in the eastern United States. At the same time, this recommendation recognizes the important role of the three major model codes in construc- tion practices throughout the United States. For purposes of analysis, we also propose that the concept of "seismic zone 2`"be in- corporated into model codes. This concept recommends that seismic zone 3 detailing prac- tice be required in those seismic zone 2 regions exhibiting high catastrophic loss potential (hence 2*)2. The expression "high catastrophic loss potential" is important from an earthquake insurance standpoint since, as explained in section 3.0, what makes earthquake insurance different from many other forms of insurance (such as life and casualty) is that eg gregate earthquake losses are characterized by extreme fluctuations on a year-to-year basis. Regions with high catastrophic losses are more likely to trigger state, regional, and national concerns as associated effects ripple throughout the economy. 2. One early step in the development of a federal earthquake insurance program would be to work with code- making organizations to determine how to define seismic zone 2*. Possible operational definitions of seismic zone 2* include o those regions where PGA2500/PGA 500 exceeds a specified number such as 1.5, o those regions where PHV2500/PHVSW exceeds a specific number such as 1.5, or o those regions where direct losses due to earthquake could exceed a given amount (such as $5 billion). In these definitions, PGA represents peak horizontal bedrock acceleration, PHV represents peak horizontal bedrock velocity, and subscripts refer to the mean return interval at a given site for these estimates. Use of such site-specific probabilistic shaking hazard estimates could serve to define zones with thigh catastrophic loss potential,' but it would need to be understood that these estimates render the interpretation of "high catastrophic loss potential" as being building-specific rather than as area-wide or regional. 3-10 For programs covering existing buildings, we define potentially hazardous buildings as those buildings which have high damageability and whose direct damage poses significant life-safety risks. In order to simplify the administrative application of the expression "potentially hazardous building," we define for public policy purposes this expression in terms of a list of nine construction classes in Table 3-3. These generally exclude single-family dwellings and duplexes, although the potential hazards of these buildings (except for Table 3-3 Potentially Hazardous Building Construction Classes Identified for Public Policy Purposes (1) Buildings with unreinforced masonry bearing walls, which do not have complete or adequate load paths for seismic forces. (2) Concrete tilt-up or reinforced masonry structures with flexible roofs. Flexible roofs include those of wood or steel deck without concrete fill. Structures having one or more of the fol- lowing inadequate features: !(a) wood ledgers used in cross-grained bending or tension, (b) no bolts or anchor straps for anchorage of walls to roof diaphragm, ;(c) excessive spacing or inadequate capacity of roof to wall anchors, *(d) chord elements that are discontinuous (not supplied with continuity plates, etc.), and/or (e) inadequate connection of tilt-up wall panels to foundation. (3) Non-ductile concrete frames -- concrete moment-resisting frames not conforming to the detailing provisions of the 1976 or later editions of the Uniform Building Code (UBC), in- cluding "pre-cast' frames. (4) Buildings with 'soft" or 'weak" first stories -- particularly those having story strengths less than 65 percent of the strength of the story above, as per 1988 UBC. (5) Buildings having unreinforced or inadequately braced parapet walls or inadequately attached exterior ornamentation. (6) Buildings with inadequately attached or rigidly attached (inadequate allowance for story drift) exterior glazing or pre-cast concrete, masonry, or stone curtain wall panels. (7) Unreinforced masonry "infill" exterior walls. (8) Unreinforced masonry interior partitions or "infill" walls in stairwells and elevator shafts. (9) Buildings where no lateral force resisting system is present or can be identified either for the whole building or for one story of the building. Buildings in which the seismic lateral force resisting system is incomplete, or has significant gaps that could allow portions of the struc- ture to collapse. 3 -11 predominantly wood-framed residential construction) remains to be analyzed. Certain classes of one- and two-story wood frame residential construction are not so "potentially hazardous" as they are highly damageable (i.e., having high likelihood of direct property damage without there being a correspondingly high life-safety hazard). (See, for instance, building practices 210 and 220 in Table 3-2.) Unfortunately, however, in the October 17, 1989, Loma Prieta earthquake, five deaths were reported in three- and four-story wood- frame residences (with weak or soft stories) in the Marina District of San Francisco. Likewise, significant numbers of deaths were associated with wood-frame dwelling damage in the 1906 San Francisco earthquake (Hansen and Condon, 1989). The existing stock of buildings that are either "potentially hazardous" or have high property damage potential is very large. (See Petak and Atkisson, 1982; Algermissen et al., 1988; Hopper et al., 1975; CSSC, 1985a; May et al., 1989; Taylor et al., 1988.) Table 3-4 lists examples of life-safety related equipment (i.e., equipment whose failure can lead to or exacerbate life-safety hazards) associated with LRMs 250 and 260 in Table 3-2. Later in this section and in sections 5 and 6 we discuss possible roles of life-safety protection programs in loss-control programs involving federal insurance involvement. As with landuse elements, we characterize at the end of this section additional supporting measures for build- ing practice LRMs. Table 3-4 Examples of Life-Safety Related Equipment o Emergency generators including batteries and fuel tanks o Fire/water storage tanks and pumps o Boilers and other equipment using natural gas o Vessels and their support structures which contain sufficient quantities of sub- stances which, if released, could endanger the general public. These should primarily include liquified gases which can form heavier-than-air vapor clouds, and which are either explosive or toxic o Elevators and elevator equipment o Emergency communication equipment 3-12 General Remarks on Loss-Reduction Activities Selected for Analysis When analyzing the candidate activities listed in Tables 3-1 and 3-2, we have included the following structural subcases as appropriate: o masonry (unreinforced versus partially reinforced) o cast-in-place concrete frames (non-ductile versus semi-ductile or with concrete shear walls) o tilt-up shear walls (unimproved or seismically improved) o pre-cast concrete frames (non-ductile versus semi-ductile or with concrete shear walls) o parapets/ornaments (unbraced versus braced or removed) o wood-frame dwellings (not anchored to foundation or with unreinforced cripple walls versus anchored or with wall reinforced at foundation) o chimneys (unreinforced versus reinforced or removed) o dwellings with story over garage (unreinforced versus reinforced) o mobile homes (unbraced versus braced) This list does not include steel frame structures for which seismiuc design requirements are generally less expensive and retrofit possibilities are more dubious. TBe types of activities and sub cases selected for analysis, with a few exceptions, fit into what we shall call community-based programs. Currently, life-safety, economic, political, or legal reasons may lead to individual instances of exemplary seismic safety practices. Espe- cially within higher seismic regions, individuals or firms may on their own decide to build to high seismic standards, to restrain, anchor, or isolate equipment and contents, and to retrofit buildings to higher seismic standards. Owners or property managers mayvoluntarily decide, or decide based on existing market incentives, to include seismic safety practices that exceed existing regulations and code requirements.. Selected insurance companies, as part of their business strategy, may systematically encourage various loss-reduction activities and may provide premium reductions (rate credits) as means to provide incentives for these measures. 3 These types of risk-redu ction activities are based on umique motivations or cir- 3. As later indicated, virtually all prudent insurers will provide some degree of partially risk-based rates to minimize adverse selection, or poor risks at average or flat rates. The degree to which underwriting efforts are made to rate accurately, to avoid poor risks, and to assist clients in reducing losses will depend in the private market on business considerations. 3 -13 cumstances. Because such motivations or circumstances occur only randomly or rarely, loss- reduction also occurs only randomly. That is, the programs are not community-based. In contrast, community-based programs generally require some degree of regulation including enforcement at local, state, or federal levels and generally require uniform applica- tion. Additionally, most activities listed in Tables 3-1 and 3-2 will require some degree of of consolidation of state and local government resources. In some instances it may be advisable to provide better information to owners, insurers and others in an effort to involve competi- tive forces in encouraging implementation of loss-control measures. The application of regulation to loss-control programs that emphasize damage-control also implicitly recognizes concerns for life-safety, and vice-versa. Since direct damage to structures and contents is the primary cause of deaths and injuries in earthquakes, reduction in property damage is the primary means to effect reductions in casualties. As a con- sequence, a comprehensive program designed to reduce propertM losses will in turn have the benefit of reducing expected casualties. (This of course does not imply that each increment of property-loss reduction has a corresponding increment in life-safety protection. For in- stance, adequate seismic design of new concrete and masonry frame structures may have high casualty reduction benefits per dollar spent. In contrast, reduction in property damage to low-occupancy and one- and two-story buildings and wood-frame structures, as well as to existing buildings, may have lesser life-safety benefits per dollar spent.) Additionally, the use of civil and police powers to enforce regulation may be significantly questioned if damage- control alone is regulated, but the use of such authority is generally justified by the clear presence of safety concerns. Hence, with a few exceptions, the community orientation of many of the candidate activities, recognizing associated life-safety goals, is consistent with a comprehensive program to reduce expected property losses associated with earthquakes. 3.3 Socioeconomic Analysis of Technically Feasible Loss-Reduction Activities This section reviews the process of socioeconomic analysis, details the methods used to estimate costs and loss reductions, and summarizes the results of the socioeconomic analyses of technically feasible loss-reduction activities listed in Tables 3-1 and 3-2 as they relate to cost-effectiveness. This analysis of cost-effectiveness is an essential ingredient in a full- fledged consensus process for identifying recommended LRMs, since without this analysis LRMs might be proposed that have appeal to a wide variety of interested parties but that have low benefits relative to their costs. 3-14 Process of Socioeconomic Analysis Figure 3-2 identifies the steps taken to accomplish the socioeconomic analysis. By and large, this analysis evaluated candidate activities on two grounds o economic efficiency or cost-effectiveness and o economic allocation - identification of those stakeholders who benefit from and those stakeholders who pay for implementation of loss-reduction activities. For this project, an interactive model was developed as a means to provide this evalua- tion. For each analyzed activity, this model yields aggregate benefit/cost ratios and also produces dollar estimates of costs and benefits to diverse stakeholders. Sensitivity analyses were performed relative to some of the many parameters that could -affect outcomes. Relative to the classification of risk analysis methods discussed in Section 2, the method used in this project for socioeconomic analysis has been probabilistic site-specific, which is effective for defining long-range costs and benefits. Costs for loss-reduction measures include chiefly construction costs with small percentages for local administrative costs. Benefits include the entire stream of expected future benefits in constant dollar values discounted to reflect the value of current investment capital. At the most restrictive level, only reduction in direct property losses were considered amon g benefits. At less restrictive levels, reductions in casualties, in the need for temporary housing, and in business interrupt- ion were also considered among benefits. Tlis expected annual loss methodology ignores large-scale fluctuations in loss levels that may occur over multiple-year periods, of time. Large-scale fluctuations often discourage capital outlays, since returns may not be immediate or may not occur within a stakeholder's tine-frame of interest. During the Project Workshop participants re-affirned the use of the interactive model, but encouraged the use of probabilistic multisite methods and further ex- amination of secondary and higher order economic losses. Should a fuller examination be made of these issues, probabilistic multisite methods would be desirable. These have the ad- vantage of defining various threshold level losses (such as losses resulting from mortgage defaults as they affect lenders, and ultimately the federal government and/or catastrophe insurers) that can create secondary financial losses through bankruptcies, insolvencies, defaults, the need for federal assistance, and the like. The interactive model was used in order to o assure that activities identified for inclusion as LRMs in a federal insurance program are economically sound (i.e., have benefits which exceed costs), and 3- 15 Define LRM alternatives: the status quo (no mitigation) and candidate activities Define marginal costs of the candidate activity Define expected direct property losses reduced through implementation of candidate activity Define expected indirect losses reduced through the candidate activity_ Define aggregate costs and benefits of mitigation A='~~~~~~~~~~~~p Define costs and benefits to various stakeholders Figure 3-2. Steps in Socioeconomic Analysis 3- 16 provi de a preliminary determination of who pays for and who benefits from specific 0 L~tms. Thus, the cost-effectiveness analysis served to assure that selected activities would not in- voIve poor economic decisions and to provide data with which to evaluate candidates rela- five to cost/benefit criteria. The stakeholder analysis, to the degree possible in this project, provided farther important information regarding the acceptability of candidate activities to various stakeholders. In application, this information can be used to determine, to the extent possible within the limitations of this project, the degree to which various stakeholders who benefit from mitigations should bear the cost of their implementation. For instance, if lend- ing institutions benefit from cost-effective loss-reduction measures, then it is, at least con- ceivable that they should bear the burden of some of the costs of these LRMs. Such con- siderations are the subject of Sections 4 and 5. It should be noted that o The decision procedures used here permit a continuous LRM evaluation scale largely dependent on the discount rate selected. 4 (Although project workshop par- ticipants favored a three percent real discount rate, OMB Circular A-94 requires a ten percent discount rate, and no discount rate is easily warranted with respect to lives saved.) o Numerous individual decision situations are reflected in the real world (and in potential distributions used for such parameters as income level, site-specific seis- inicity factors, deductible levels, limits of liability, and premium levels under dif- ferent kinds of federal involvement, for both earthquake and non-earthquake insurance). Neither this range of cases nor helpful human behavioral models can be reflected in the limited mechanical cases produced for this project. o The model as developed applies to individual decision-makers, and hence accounts neither for widespread decisions nor for community impacts (such as those resulting from large-magmitude earthquakes) of those decisions. 4. Discount rates purport to quantify the time-value of money. Money today is more valuable than money tomorrow - even discounting for inflation -- to the extent that money as capital has an investment earning power. Thus, money may earn at xo above inflation. For public policy analyses, this earning power of money has many critical implications for the benefit-cost ratios developed. Using the variable xo allows for a gradated or continuous evaluation of LRMs in terms of the discount rate at which they would have a favorable benefit/cost ratio. (See Taylor, Atkisson, and Petal, The real discount rate is the constant dollar/earning value of current money, and is used as a continuous in- teractive parameter. Hence, an LRM may be evaluated in terms of the lowest discount rate at which its benefit-cost ratio exceeds unity. Discount rates to be used are subject to many controversies. 3-17 Estimations of Costs and Expected Annual Direct Loss Reductions Resulting from Implementation of Candidate Loss-Reduction Activities In order to perform socioeconomic evaluations of mitigations, one must have available data that at the very least can be used to indicate the costs and benefits of mitigations. Ap- pendix B presents the results of analyses performed to estimate (1) risk levels (degree of loss) associated with various levels of seismic (earthquake) frequency and intensity, (2) dollar values for both implementation costs and loss-reduction benefits of can- didate building practice LRMs, and (3) dollar values for both implementation costs and loss-reduction benefits of can- didate landuse planning LRMs. Again, it should be noted that estimates relate to expected annual direct loss reductions and were designed for average conditions. Annual direct loss reductions incorporate all possible earthquakes affecting buildings analyzed, not merely rare serious ones. As indicated in Eguchi et al. (1989), ignoring effects of smaller but damaging earthquakes is a serious error not only in earthquake insurance rating but also in analysis of the benefits of LRMs. In order to account for variations in frequency of earthquakes of varying seismic inten- sities, short-cut methods were used that considered both (a) variations in seismic intensity levels (owing to distance from sites to various earthquake sources) and (b) effects of strong ground motions on various local soils. (See Appendix B.) Assumptions made are conserva- tive since one goal of this analysis was to eliminate candidate LRMs which are, as a general rule, poor investments. Even more conservative assumptions typically were made with respect to landuse measures mitigating surface fault rupture, liquefaction, and landslide hazards. Sensitivity of outcomes to variations in estimates of frequencies of shaking inten- sities were analyzed within the interactive model. For both shaking intensity and analysis of potential permanent ground displacements, it is possible at greater cost to provide much finer estimates. Current methods for probabilistically analyzing liquefaction and landslide hazards require additional systematic applied research in order to become credible for microzonation analyses (Eguchi et al., 1989). Seismic vulnerability estimates were derived largely from ATC-13 (1985), Wiggins and Taylor (1986), and engineering judgment. (See Appendix B.) As indicated in Wiggins and Taylor (1986), for selected classes of construction, procedures previously used may sig- nificantly underestimate the losses reduced through seismic improvements. Consequently, sensitivity analyses were performed in order to determine the effects of using diverse seismic vulnerability estimates to evaluate specific loss-reduction activities. 3 - 18 Estimates of the direct initial costs of loss-reduction activities were derived from a number of sources along with engineering experience. These represent average initial cost estimates for instances when the activity is technically feasible. This study as well as previous studies strongly indicates that initial cost estimates are an extremely critical parameter in the assessment of IRMs (see Taylor, Atkisson, and Petak-, 1981). By the same token, techniques developed in order to reduce expected capital outlays needed to achieve seismic design and redesign objectives even by small margins when applied broadly can have potentially - - large-scale aggregate effects in reducing the costs of losses associated with earthquakes. In other words, research programs that yield small percentage reductions in seismic design and redesign costs (e.g., 0.1% of replacement value) can have large-scale aggregate effects (e.g., millions of dollars in reduced costs). In addition to initial outlay costs and estimates of direct property benefits for specific LRMs, an adequate risk-and-decision analysis requires consideration of secondary costs and losses associated directly wiith the property and its stakeholders. Appendix C discusses addi- tional estimates of economi c and stakeholder costs and losses. In addition to estimates of direct property losses, analysts considered losses associated with o temporary housing, o business interruption, o deaths and injuries, o the cost of money, and o insurance premiums. Not considered were losses associated with o contents, o infrastructure facilities and fire following, o release of toxic or hazardous chemicals, o transaction, foreclosure and business closure (except for crude assumptions on load- ings for premiums), o post-disaster clean-up, o price deflation, o unemployment resulting from general economic decline. 3 - 19 Results of the Economic Efficiency Analysis In order to rate candidate LRMs in terms of cost-effectiveness, a three-tier criteria ap- proach was used. This permits a continuum of economic evaluations with those LRMs that pass at tier 1 obviously passing at tiers 2 and 3, those passing at tier 2 obviously passing at tier 3, and those failing at tier 3 obviously failing at all tiers. This continuum of economic evalua- tions recognizes that diverse economic criteria, such as are implicit in discount rates selected, may be used to determine economic acceptibility. At tier 1, the most rigorous, only direct property losses were considered, and an eight (8) percent discount rate was used. Only the most cost-effective LRMs satisfy tier 1 criteria. At tier 2, clear monetary considerations were included as secondary losses associated with temporary housing, business interruption, and deaths and injuries were considered, and a three (3) percent discount rate was used. Deaths were calculated to have an average value of $300,000 (as a monetary measure only), which represents a conservative appraisal of a lifetime discounted income. Tier 2 represents the level for sound economic policy supported by project workshop participants. At tier 3, losses above primary and secondary levels were considered, deaths were valued at $1,000,000 (a value exceeding average present discounted value of potential future earnings), and a zero (0) percent discount rate applied. Tier 3 may be appropriate for community health, safety, and welfare programs. LRMs not satisfying tiers 1, 2, or 3 criteria are as a rule not economically sound policies. The main results of the efficiency (cost-effectiveness) analysis are outlined below. Loss-reduction activity numbers refer to Tables 3-1 and 3-2. At tier 1, most of the activities covering new seismic design (#s 100, 110, 120, 130, 140, 150, 160) passed, with all cases passing at tier 2. Generally speaking, implementation of seismic building codes has a favorable benefit/cost ratio--even from the standpoint of property loss reductions alone. Seismic retrofit of unbolted and/or poorly anchored wood-frame residences in seismic zone 4 (# 220) passed at tier 1. The low per-unit cost of this measure combined with sig- nificant property loss-reduction make this an economically attractive candidate mitigation measure. Use of geotechnical means to minimize landsliding, severe liquefaction, and/or sub- sidence hazards in seismic zone 4 (# 1500) passed at tier 1 for commercial and public developments and for large-scale residential tracts. Economies of scale are significant with respect to residential structures; this measure appears to be economically unattractive for unit-by-unit single-family and duplex residential developments. 3-20 At tier 2, seismic retrofit oftiit-up construction in seismic zone 4 (one subcase of # 190) appears to be economically attractive owing to the comparatively low costs of this retrofit. Other economic measures are seismic retrofit of unreinforced masonry structures in seismic zone 4 (another subcase of # 190), and seismic retrofit of tilt-up construction in seismic zones 2 and 3 (subcases of #s 170 and 180). Also at tier 2, a number of landuse/geotechnical en gineerig measures passed for residential construction other than single family dwellings or duplexes. These include, for seismic zone 4, restrictions on new development in very active fault zones of deformation (# 1200), in highly susceptible liquefaction zones (# 1700), and in highly susceptible landslide locales (# 2100). For landslide locales, possible purchase of land with conversion to open- space uses can in some cases be economically warranted (# 2300). Considerations ofpublic health and welfare can enter into these landuse decisions. Seismically poor land can also correspond to regions of intensive infrastructure earthquake damage with associated health and welfare concerns (see Selkregg et aL., 1984). Minimization of severe liquefaction and/or subsidence hazards in seismic zone 3 (# 1600) also passed at tier 2 for new commercial and public development. Also, for modifica- tions of commercial structures in severe liquefaction zones in seismic zone 4, use of geotech- nical techniques to minimize these hazards passed at tier 2 (# 1900). Mitigations passing tier 1 and tier 2 represent, as a rule, economically sound policies. These include life-safety factors only in defensible economic terms. Activities that pass at tier 3 represent policies that have lower ratings on economic grounds, but that may be pursued as sound public policies if life-safety hazards, potential community disruption, and other factors are emphasized. At tier 3, seismic retrofit of unreinforced masonry buildings appears to be warranted in seismic zones 2 and 3 (subcases of #s 170 and 180 in Table 3-2). Seismic retrofit of "potentially hazardous" buildings in seismic zone 4 passes at tier 3 for all pertinent building types (retrofit of tilt-up and unreinforced masonry buildings passed at tier 2) (# 190). Seis- mic reinforcement/anchorage/bracing of parapets and ornaments (# 230) passed at tier 2. With respect to landuse measures in seismic zone 4, restriction of new development in moderately active fault zones of deformation passed at tier 3 (# 1300). In, seismic zone 3, restrictions of new developments in highly susceptible liquefaction or landslide locales (#s 1800 and 2200) appear to pass at tier 3, as does use of geotechnical techniques to nmninimize 3 -21 ground failure potential damage whenever major modifications of structures are made in highly susceptible liquefaction zones (# 2000). Seismic retrofit of wood-frame dwellings in seismic zones 2 and 3 (#s 200 and 210) is of marginal cost-effectiveness and needs further ex- amination. Not passing on any of the three tiers (and as a rule with exceptions that can be assessed through individual benefit-cost analyses) are o seismic retrofit in seismic zones 2 and 3 of non-ductile and pre-cast concrete frames (subcases of #s 170 and 180) o purchase of existing (undamaged) construction or properties in fault zones of deformation, or zones with severe liquefaction, landslide or rockfall potential (#s 1000, 1100, as illustrations) o restriction of new development in moderately active fault zones of deformation in seismic zones 2 or 3 (# 1400) o purchase and conversion of land in highly susceptible landslide susceptible locales in seismic zone 3 (# 2400) Activities numbered 200 and 210 only marginally passed for subcases considered at tier 3. No formal analysis was provided for activities numbered 240,250, and 260. 3.4 Summary In this section we outlined the procedures used to identify and define feasible Loss Reduction Measures (LRMs) for possible incorporation in a federal earthquake insurance program. The procedures used involved 1. a thorough information search, which identified 96 earthquake hazard reduction ac- tivities for consideration, 2. reduction of this list to 32 loss-reduction activities for further analysis, 3. socioeconomic risk-and-decision analyses to evaluate these 32 loss-reduction ac- tivities for cost-effectiveness and stakeholder impact, and 4. a project workshop during which cost-effective loss-reduction activities were further analyzed, synthesized and developed into initially recommended Loss-Reduction Measures. 5. a final advisory panel meeting and review by advisory panel members and other knowledgeable parties of these initially recommended LRMs. 3-22 The socioeconomic analysis permitted the evaluation of candidate loss-reduction ac- tivties in terms of their cost-efficiency and stakeholder impact. Those that are especially cost-effective include o adoption of, compliance with, and enforcement of adequate seismic design provisions in new construction in all seismic zones (with minimal or no costs in seismic zones 0 and 1), o seismic retrofit of unbolted and/or poorly anchored wood-frame residences in seismic zone 4, and o minimization through geotechnical techniques of severe landslides liquefaction and/or subsidence hazards in seismic zone 4. As discussed in the next section, these results were considered further in the course of an ac- ceptability analysis. 3 -23 4.0 THE IDENTIFICATION, DEVELOPMENT AN]) EVALUATION OF ACCEPTABLE LOSS-REDUCTION MEASURES As explained in Section 1, the characterization of earthquake loss-reduction provisions as "feasible"r requires that they must prove to be both cost-effective and otherwise acceptable to a wide variety of individuals who will be affected by the measure. As a result, three processes were utilized to help ensure that recommended loss- reduction provisions would be acceptable. (1) The socioeconomic analysis conducted as. part of this study included an economic al- locative or stakeholder analysis of technically feasible loss-reduction activities. (2) Cost-effective loss-reduction activities were presented to a Project Workshop for further consideration and modification by diverse interest groups. (3) Initially recommmended loss-reduction measures were submitted to advisory panel members and other knowledgable parties at a final advisory panel meeting and through draft final reports. This section reviews the procedures used to conduct these processes and reports the results of these efforts. 4.1 Economic Allocative (Stakeholder) Analysis A stakeholder analysis has been developed in this project in order to estimate gains and losses to stakeholders with respect to the status quo and to alternatives so that various loss- reduction policy measures may be recommended. The following stakeholders were con- sidered in the allocative analysis: o the general taxpayer (the federal government) o the state taxpayer (for specific state governments) o the local taxpayer (for specific municipalities) o lending institutions o realtors o developers o building contractors and subcontractors o owners o small businesses o workers/employees o tenants o nonprofit organizations o low-income residents 4-1 o companies writing earthquake insurance o companies writing non-earthquake insurance These and a wide variety of other stakeholders may be directly involved in a proposed loss- reduction program. In this list, taxpayers are included in three categories in order to evaluate how losses are distributed among individual states and municipalities. Hence, a taxpayer in a highly seismically prone municipality or state may expect to pay higher state or local taxes to offset earthquake damage than a taxpayer in a less seismically prone municipality or state. Accordingly, individuals may have interests that are associated with potentially competing stakeholder-claims: one may desire lower earthquake insurance rates for one's residence or business; one may desire lower local, state, and federal tax burdens; one may desire to occupy more seismically resistant buildings; and one may have additional interests in the above stakeholder positions. Stakeholders do not always accurately evaluate their economic interests. People may support or oppose programs based on only partial or incorrect information. We have con- ducted a limited stakeholder analysis in an effort to clarify where economic interests in proposed measures lie -- who gains and who pays. In this regard, the stakeholder analysis may estimate gains and losses to stakeholders with respect to (a) the status quo, (b) diverse proposed loss-reduction measures, and (c) alternative types of federal involvements in earthquake insurance. Thus, stakeholder analyses clarify potential gains and losses to existing or prospective con- stituencies. These analyses are also useful in determining which public policy vehicles may be useful in view of the stakeholder benefits and costs of a proposed loss-reduction measure. 4.2 Results of Economic Allocative (Stakeholder) Analysis In discussing the results of the stakeholder analysis, we distinguish among diverse stakeholder populations. In some instances, as with taxpayers, individuals may fall into a number of stakeholder positions (e.g., federal taxpayer, taxpayer for a specific state, taxpayer for a specific municipality). Key results obtained from the stakeholder analysis include the following: o Federal taxpayers currently bear the largest burden of earthquake property losses associated with public and selected private nonprofit buildings. Since current dis- aster relief policy places the financial burden of recovery on the federal government, the benefit/cost ratio of implementing loss-reduction activities to state and local governments is significantly reduced. In effect, current disaster relief policy serves as a major disincentive to mitigation for state and local and selected private non- profit buildings. 4-2 o All stakeholders modeled benefit fromtcost-effective IRMs except for state and lo- cal taxpayers for activities pertaining to state and local buildings. This general con- clusion may not necessarily hold for short-term owners, tenants, realtors, and con- tractors, who were not included in the model (in these cases, short-term time horizons, possibly reflected in high discount rates, may need to be considered in the stakeholder analysis). o Given current contingent federal liabilities with respect to ublicly owned buildings, the costs of federal subsidies, for cost-effective LRMs for these buildings can be off- set by reduced federal liabilities. o Mortgage-lending institutions benefit from cost-effective IRMs. A finer analysis of the risks borne by these institutions is therefore needed in order to estimate more precisely the degree to which these lenders benefit from implementation of LRMs. o Workers generally benefit from implementation of cost-effective LRMs owing to reduced expected casualties and unemployment. o Since losses in other lines of insurance (auto, theft, etc.) are expected to be reduced by the implementation of LRMs, insurers who write policies in these other lines and who may expect losses after earthquakes benefit from these LRMs. o Taxpayers in general gain from cost-effective loss-reduction activities, since their implementation - reduces post-disaster loans (subsidies) and grants, and - reduces the need for te mporary housing (relative to residential LRMs). o Tax credits for homeowners who undertake cost-effective loss-reduction activities are not effective vehicles to offset contingent federal liabilities, since benefits of tax credits would largely accrue to higher income segments of the population who are least expected to require post-disaster assistance. A fuller study of the benefits of tax credits would be needed before they should be recommen ded as means to induce LRMs. o Subsidies including tax deductions and low interest loans should be considered for cost-effective commercial seismic retrofit programs to the extent that these programs reduce post-disaster costs of - unemployment insurance, - workers' compensation claims, and - loss of tax revenues resulting from business interruptions and closures. In review, virtually all stakeholders benefit fromr implementation of cost-effective loss- reduction measures. The notable exception is the state and local taxpayer and private non- profit institutions eligible for federal disaster assistance, because federal disaster relief assis- tance policies for public and private nonprofit building provide a disincentive for additional investment in loss-reduction activities. (See Appendix D.) 4-3 4.3 The Project Workshop and Final Advisory Panel Meeting The efforts outlined thus far were ultimately aimed at identifying earthquake loss- reduction provisions which could potentially be worked into a national insurance or rein- surance program. After the socioeconomic analyses were completed, the remaining loss- reduction activities were presented at a Project Workshop for further revision, refinement, and definition as Loss Reduction Measures (LRMs) for potential incorporation into a na- tional insurance or reinsurance program. Afterwards, a final advisory panel meeting was held, and draft final reports were reviewed by advisory panel members and other knowledge- able parties. The Project Workshop The Project Workshop was composed of recognized experts and interested and af- fected parties representing a very broad spectrum of interests, professions, and geographical regions. The workshop was to provide for further consideration of the views of diverse inter- est groups and to assure that the proposed measures are professionally supportable, and ac- ceptable to communities, the insurance industry and policyholders. The workshop was to en- courage jhe introduction of new information and innovative ideas. Workshop participants were divided into five working groups: o building issues, o landuse planning/geotechnical issues, o risk analysis issues, o socioeconomic/insurance issues, and o public policy/legal issues. The workshop was designed to encourage participants in each of these five topic areas to freely discuss, augment, and revise preliminary findings and finally to develop and evaluate specific conclusions. In both the landuse and building issues sessions, participants were first asked to review a list of cost-effective loss-reduction measures (LRMs) and to add to this list. Next, positive and negative impacts of these LRMs were to be listed. Then, the resulting LRMs were to be evaluated on the basis of their probability of successful inclusion in a national insurance program. A similar set of steps was used to develop lists of supporting elements -- activities required to initiate, support, or sustain promising LRMs. 4-4 The risk analysis session was organized around - risk analysis methods, - other model elements of risk analysis, and - possible uses of risk methods in a federal earthquake insurance program. First, prepared lists of these methods, elements, and uses, respectively, were added to. Next pros and cons of these lists were discussed. Finally, the resulting lists were evaluated. The design of the working session on socioeconomic issues was, first to augment the list of stakeholders previously modeled, next to evaluate the merits of including various stakeholders, and then to rate the inclusion of these stakeholders in the model socioeconomic analysis. Then, preliminary socioeconomic analysis results, were listed, to be added to, discussed, and evaluated. The working session on public policy/legal issues was designed to evaluate the impor- tance of various political issues relating to LRMs that had been developed prior to the workshop. Advantages and disadvantages of including these and other issues in the public policy analysis were to be listed and evaluated. The design of the workshop did not include a concerted effort to make every matter discussed and every recommendation developed consistent in each of the five working ses- sions. For instance, LRMs that may have been added or altered in the building issues and landuse issues sessions were not necessarily specifically evaluated in the session on public policy analysis. For another instance, recommendations developed in one session may have been qualified or discussed in another session. However, the intent of the workshop was to convene recognized experts and interested parties in each of a number of disciplines and in- terest areas and to have those parties critique, modify, and augment conclusions developed so far in this, project. Workshop participants were not asked to provide an overall integration of results. Also discussed by Workshop participants were status quo obstacles to implementation of IRMs and strategies for overcoming these obstacles. Additional public policy/legal analysis are included in Section 5, with respect to the inclusion of LRMs in various types of federal insurance involvements. The Final Advisory Meeting and Reviews of Draft Final Reports The overall integration of results was developed in an initial draft final report which was discussed at a final advisory panel meeting. Owing to the complexity of the report and to serious objections to initial incorporation proposals, especially with respect to federal rein- 4-5 surance involvement, an additional draft final report was also submitted. These draft final reports were reviewed by advisory panel members, by the project officer, and, through the project liaison, by federal officials on the interagency task force on earthquake insurance. Although much of the attention of these reviews of draft final reports concentrated on incorporation issues, reviewers also added valuable insights that led to further refinements and revisions of the initially recommended LRMs. One central issue raised in the final Ad- visory Panel meeting and afterwards was the status of the recommended LRMs: Were they to be explicitly stated in any bills before Congress, were they to be included in administrative guidelines, or were they to be sufficiently detailed to be readily implemented as expressed? Given the consensus process used in this project, it is clear that more specific LRMs than are recommended here would require further administrative and technical discussion. In con- trast, inclusion of the LRMs as stated may lead to an undesirable degree of administrative in- flexibility. Hence, the LRMs recommended in this project are designed to be used in ad- ministrative rules and regulations within existing and/or major modifications of federal programs. Moreover, this report is advisory only to FEMA and so does not reflect their final considerations. 4.4 Cost-Effective LRMs Recommended LRMs and Supporting Activities/Elements Developed The landuse and building groups of the project workshop were asked to refine, revise, and augment those loss-reduction activities defined as being cost-effective. The resulting LRMs were then evaluated, and supporting or framework elements were identified and defined. Results of the evaluation process were intended to refer to a likelihood of success- ful inclusion in a federal earthquake insurance program having a loss-prevention element. This subsection presents the results of those two working sessions as revised by the project team in response to comments by the project advisory panel and other considerations. Landuse LRMs Table 4-1 lists promising landuse LRMs, all of which were evaluated by the working landuse group at the Project Workshop as having a high probability of successful inclusion in a federal earthquake insurance program. All four LRMs apply chiefly to seismic zones 3 and 4, although LRM 3, pertaining to post-damage situations, may be extended to seismic zone 2. LRM 1 does not apply as cost-effectively to scattered construction of single-family dwellings, and administrative and legal versions of this LRM may consider scattered or unit-by-unit 4-6 construction of single-family dwellings as a possible exclusion. The other three LRMs apply to all structures. All-four LRMs require governmental regulation as a prm instrument for implementation, and all four emphasize potential permanent ground failures. Local extreme strong motion amplification effects owing to high relative site response factors were added to LRM 2, because an increasing body of evidence demonstrates that these factors are sig- nificant contributors to earthquake loss potential. Some concern was expressed over the use of 50 percent of replacement cost as a criterion in LRM 4. In administrative versions of this LRM, consideration may be given to use of "actual cash value" or "market value" in lieu of "'replacement cost." Table 4-1 Recommended Landuse LRMs (Applicable only in seismic zones 3 and 4) New Developments Require in high liquefaction susceptible zones that geotechnical techniques be used to LRM 1 minimize potential ground failures for: o new commercial, public, and residential subdivision development, and o major modifications of commercial, public and residential subdivision development. (Exceptions of scattered construction of single-family dwellings may be considered in legal and administrative versions of this loss-reduction measure.) Use zoning ordinances, subdivision ordinances and other techniques to control new LRM 2 development in active fault zones, high" site amplification, landslide and liquefaction sus- ceptible zones. Existing Developments Permit reconstruction or replacement of existing development in areas identified as active LRM 3 fault zones, high landslide, or liquefaction susceptible zones experiencing damage of more than 50% of its replacement value only if the identified risk is reduced to an acceptable level. Consider purchase of existing damaged properties in high landslide susceptible zones unless suitable measures are used to protect existing development from damage. Permit no additions to buildings in areas identified as active fault zones, high landslide or LRM 4 liquefaction susceptible zones unless the risks are reduced to an acceptable level, except additions to single-family dwellings up to 50% of the replacement cost, which can be made without such risk reduction. 4-7 In evaluating these four LRMs, the working group emphasized two general advantages: o reduction in expected property losses (reduction in exposure/risk, limitation of number of potentially hazardous buildings, limitation of potential losses to the housing stock, and reduction of post-disaster clean-up costs), and o improvement in public safety and health in regions where damage potential is high to such infrastructure facilities as sewage lift-stations, natural gas mains, and water mains. The working groups highlighted the following general disadvantages: o apparent loss of market property value in high hazard zones (Further legal and socioeconomic analysis is needed to determine whether or not externalization of earthquake risks can be assumed in assessment of market values.), and o possible litigation and political dissension resulting from this loss of property value. Supporting Elements for Landuse LRMs Table 4-2 provides a list of candidate activities that support or sustain LRMs described in Table 4-1 as determined by the workshop attendees. In the main, these are restricted to seismic zones 3 and 4 for which available data have already been developed. Hence, these supporting activities do not require huge sums of money as some reviewers of earlier final drafts have contended. Li consists of small-scale maps needed principally to define seismic zones. These maps are being updated continually. The remainder of the supporting elements apply primarily in seismic zones 3 and 4. Intermediate scale maps (12) requiring examination of local geologic effects on strong ground motion are helpful in defining "poor sites" relative to strong ground motion. Their identification is useful in further delineating liquefaction/subsidence/landslide hazards. Fur- thermore, as landuse planning evolves, inclusion of LRMs pertaining to poor sites will gain more support. The project director strongly endorses continuing improvements of building codes in order to determine the feasibility of including microzones for poor soils within the seismic provisions of those codes. Moreover, consideration of poor sites relative to strong ground motion are incorporated into building practice LRMs. 1S requires maps of active fault zones of deformation in seismic zones 3 and 4. The State of California already has such maps, and these are also available from Utah Geological and Mineral Survey for portions of the Wasatch Front, Utah. As the socioeconomic analysis indicated, in many cases, these maps have more symbolic than economic value, although they are useful in general earthquake source studies and in some cases indicate fault creep 4-8 (gradual movement of identified faults) and additional hazards (new faults). Cases in which development in moderately active fault zones of deformation can be avoided -- without sig- nificant economic loss - should be encouraged. These maps can also be extremely useful in controlling the development of critical and very high occupancy structures. Table 4e2 Activities Supporting Recommended Landuse LRMs Supporting Element Description For the entire United States, development of small scale maps (1:5,000,000) of ground Li motion, evaluated by an expert panel. For urban areas with a minimum population (e.g., 50,000) development of inter- L2 mediate scale maps (1:100,000) of ground motion that include examination of local geological effects on strong ground motion (e.g., maps of relative site velocities for dif- ferent spectra). Compilation and as necessary development of large scale maps (1:24,000) of Quater- I3 nary surface faulting within a 50-mile band outside the perimeter of urban areas having a minimum population. Compilation and development of intermediate scale maps (1:100,000) elsewhere in seismic zones 3 and 4. Compilation and development of large scale liquefaction and landslide high seismic IA susceptibility maps (1:24,000) for urban areas having a certain minimum population. Greater attention should be placed on quantitative interpretation of such expressions as "high susceptibility." Areas mapped should be large enough to accommodate short- term growth in undeveloped areas around the city. Construction of information databases and transfer mechanisms so that the foregoing 15 maps may be readily available and understandable to local officials, realtors, developers, insurance companies, and the general public. Requirement that general plans include a seismic safety element that sets development L6 policy for local geological hazards including high relative site response factors, fault zones, and regions of high liquefaction and/or landslide susceptibility. Development of requirements that in areas identified as active fault zones, and high 17 landslide or liquefaction susceptible zones that a geologic/geotechnical report be prepared for critical facilities, high-occupancy buildings, new subdivisions, and major modifications of high-occupancy (and/or critical) buildings, and that these be reviewed by a suitable licensed professional Development of guidelines for preparation and review of geologic/geotechnical 1.8 reports. Provision of resources for state and local programs, procedures, and staffing to effect L9 LRMs. 4-9 LA requires landslide and liquefaction maps. Data for these, for instance, are available for portions of California, Alaska, Utah, Nevada, and Washington. Of special interest is more detailed delineation of very high susceptibility zones, and, for many purposes, the desirability of improved techniques to quantify these hazards (see Eguchi et al., 1989). The broad zones as defined for the Wasatch Front, for instance, could be more narrowly defined based on future detailed site investigations and research that better determines probabilities of liquefaction-induced ground failure and that provides more consistent quantitative defini- tions of "high" and "very high." (See Anderson et al., 1982; Idriss et al., 1986; Taylor et al., 1988.) Economic means to develop seismically-induced landslide maps should be en- couraged with emphasis possibly on areas where previous landslides have occurred. Once again, quantitative interpretation of landslide susceptibilities needs to be improved. L5 encourages the development of systematic means for the transfer of these maps to interested parties. L6 requires a seismic safety element within general plans of municipalities with a cer- tain minimum population. This activity is regarded as a supporting element rather than as an LRM because there has been considerable controversy over the effectiveness of seismic safety planning elements. L7 requires geologic/geotechnical reports for critical facilities, high-occupancy facilities, new subdivisions, and major modifications of high-occupancy (and/or critical) buildings. L8 requires guidelines for preparation and review of these geologic/geotechnical reports. L9 was added by the project director based on Selkregg et al., 1984, pp. 160ff. Support for capable state and local government LRM programs, procedures, and professional staff- ing is an essential means to carry out loss-reduction policy. Continuity of organizations and professional staffing is needed to take advantage of loss-reduction opportunities. Ap- propriate capabilities include technical knowledge. The federal govermnent may justifiably provide partial financial support for these programs, procedures, and staffing to the extent that landuse measures are expected to offset existing contingent federal liabilities. This sup- port (made later also with respect to building LRMs) is consistent with the recommendations by S. Scott (1988a and b) that implementation needs both (a) local agents of experimentation and innovation and also (b) continuing education and training programs. For the bulk of these supporting elements, many previous investigations and activities can be used to provide a basis for LRMs proposed, especially in Alaska, California, Utah, and Washington. 4 - 10 LRMs Related to Building Practices Ten LRMhs affecting building practices were carefully formulated by workshop par- ticipants and were evaluated as, having a high probability of successful inclusion in a federal earthquake insurance program. An eleventh (LRM 6) was added by the project director on consideration of a primary program that includes residential structures. These LRMs are listed in Table 4-3. Of the LRMs affecting building practices, those evaluated most highly in terms of the likelihood of successful inclusion in a federal earthquake insurance program having a loss- reduction element were: LRM 7 - in seismic zones, 2, 3, and 4, design of new essential buildings and public schools (including colleges and universities) in conformance with current model code seismic provisions. LRM 10 - seismic retrofit of unreinforced masonry buildings in seismic zone 4. LRM 11 - seismic securing/strengthening of building parapets and external ornamenta- tion in seismic zone 4. LRM 12 - seismic retrofit of potentially hazardous essential buildings and public schools (including colleges and universities) in seismic zone 4. LRM 13 - in seismic zones 3 and 4, inspection and disclosure of anchorage and presence of unbraced cripple walls at time of property transfer for buildings including one- and two-family dwellings. and of anchorage of mobile homes. LRM 14 - In seismic zone 4, state law should require that gas water heaters in new and existing multi-family dwellings be braced or strapped to structural framing. Evaluated almost as highly are LRM 5 - incorporation of NEHRP seismic provisions in eastern model codes (seismic zones 0,.1, 2,3) and LRM 15 - require seismic retrofit of pre-1976 concrete tilt-up construction in seismic zone 4 within 10 years. Evaluated as passing (as less convincing to workshop participants) are LRM 8 - addin superior detailing requirements in seismic zones currently designated 2 with high catastrophic loss potential. LRM 9 - in seismic zones 3 and 4, disclose a "hazard rating" to potential buyers well before the time of escrow. 4 - 11 Table 4-3 Recommended Building Practice LRMs New Construction Eastern model codes shall be encouraged to incorporate (adopt by transcription) the latest version LRM 5 of the NEHRP seismic provisions. All model codes should incorporate a geotechnical component that considers local site amplification effects on strong ground motion and minimization of potential ground failure effects. Building regulatory authorities should adopt and enforce model codes that have adequate seismic LRM 6 provisions for one- and two-family dwellings and anchorage of mobile homes. The building code should apply also to repairs of earthquake-damaged buildings to assure that losses are not repeated in subsequent earthquakes. In seismic zones 2, 3, and 4, new essential buildings and public schools, including colleges and LRM 7 universities, should be designed in conformance with current model code seismic provisions. In seismic zones currently designated 2 but with high seismic catastrophic loss potential (designated LRM 8 2*) model codes should require the detailing requirements applied to zones of high seismicity. For new construction in seismic zones 3 and 4, a building "hazard rating" must be disclosed to LRM 9 potential buyers well before the close of escrow. Existing Construction LRM 10 In seismic zone 4, local jurisdictions should institute ordinances with requirements for seismic retrofit of unreinforced masonry (URM) bearing wall buildings. These buildings should be required to be upgraded to a minimum-level or else demolished within a 20-year period. LRM 11 In seismic zone 4, local jurisdictions should institute ordinances for the securing/ strengthening of building parapets and external ornamentation within a 20-year period. LRM 12 In seismic zone 4, potentially hazardous (other than URM) essential buildings and public schools in- cluding colleges and universities must be retrofitted or phased out within a 20-year period. LRM 13 In seismic zones 3 and 4, inspections of buildings including one- and two-family dwellings and an- chorage of mobile homes should be performed prior to significant financial commitment or property transfer and hence well before the close of escrow. A report to the potential buyer should indicate whether or not: a. the dwelling is anchored to the foundation, b. unbraced cripple walls are present, and c. gas water heaters (if present) are adequately braced or strapped to the framing. LRM 14 In seismic zone 4, state law should require that gas water heaters in multi-family dwellings (new and existing) be braced or strapped to structural framing. LRM 15 In seismic zone 4, concrete tilt-up construction which does not have adequate roof-to-wall anchors and continuity ties shall be required to be retrofitted within 10 years. 4 - 12 LRM 9 has been the subject of several criticisms by project participants. First, studies have shown that disclosure requirements in California have sometimes had little if any dis- cernible effect on purchases in active fault zones in California. (See Palm, 198 1.) However, LRM 9 emphasizes disclosure well before the close of escrow. Second, if LRM 5 is accepted, the LRM 9 might initially appear to be redundant. Yet, even if L1RM 5 is adopted, LRM 9 provides a basis for distinguishing between lbuildings conforming to current model seismic code requirements and older buildings that may not so conform. More elaborately, a hazard rating system can include the following categories: o Conforming to current model seismic code requirements o Potentially seismically hazardous (as defined in Table 4-3) o Nonconforming to current model seismic code requirements and not potentially haz- ardious o Seismically retrofitted to 65% of current model seismic code design force requirements Although LRM 9 is not so essential as most of the other LRMs recommended, efficient use *of such a hazard rating system can help in o insurance rating, o mortgage lending, pension fund, mortgage securty, and other financia decisions which are concerned with potential catastrophic losses from mortgage defaults, and o building purchase decisions. In short, LRM 9 has the potential to discourage through a wide variety of financial incentives the prolonged use of potentially hazardous buildings. LRM 5 recognizes the autonomy and importance of the three major model codes in the United States, and so does not propose a national building code. Moreover, LRM 5 acknow- ledges the leadership of the Structural Engineers Association of California in developing seismic provisions for the Unifrm BuildingCode. In two years, LRM 5 might be revised after adequate review to consider incorporation of ASCE/ANSI (American Society of Civil Engineers/American National Standards Institute) seismicprovisions. One LRM noticeably absent from workshop discussions, but recommended in LRM 6, is the development and incorporation in model codes of adequate seismic provisions for residential construction. Progress has already been made in developing agreement among model code organizations on codes for residential construction. (See CABO, 2989.) These codes represent suitable minimum standards for an LRM in a federal program. Basedl on 4-13 codes represent suitable minimum standards for an LRM in a federal program. Based on other Workshop recommendations on seismic building codes, the project director believes that LRM 6 would also be rated highly. The LRMs thus presented for inclusion in a federal earthquake insurance program having a loss-reduction element cover a wide range of structures, from public to residential to commercial/industrial, and all seismic zones. The LRMs proposed also differ significantly with respect to socioeconomic criteria. In particular, life-safety factors and ease of ad- ministration were emphasized in o LRMs proposed for essential facilities, and o LRMs proposed for seismic retrofit of "potentially hazardous" buildings (only some of which are clearly cost-effective). An optional LRM, not discussed in the workshop but clearly technically feasible is the an- chorage or restraint of life-safety related equipment (see Table 4-4) in essential buildings in seismic zones 3 and 4. This and many other candidate LRMs may on further examination, or with additional research, be found to be cost-effective and acceptable. Review of research and applications advances, including those in using base-isolation strategies for design and retrofit, should be used to reassess periodically cost-effective and acceptable LRMs. Supporting Elements for Building Practice LRMs Table 4-4 summarizes supporting elements needed for successful implementation of the building LRMs listed in Table 4-3. All supporting elements except B6, B7, and B8 were evaluated highly by participants in the building working sessions. B6, B7, B8, B9, and B10 were added by the project director in response to later considerations provided by project reviewers. Except for B2, which applies to new construction in seismic zone 2, and B8 and B9, which apply generally, the supporting elements are designed chiefly for existing construction in seismic zones 3 and especially 4. The marginal costs of these supporting elements are low. B1 defines "potentially hazardous" buildings as a means to support LRM 6, LRM 15, and partially risk-based rating systems (See Table 4-3). B2, which provides for the definition of seismic zone 2*, facilitates application of LRM 8. Our previous discussion in subsection 3.2 describes alternative definitions of seismic zone 2 that could be considered in determining the application of LRM 8. 4 - 14 Table 4:4 Supporting Elements for Recommended Building Practice LRMs Supporting Element Description BI Definition of 'potentially hazardous' buildings as in Table 4-3. B2 Definition of seismic zone 2* -as those seismic zone 2 areas with high seismic potential at extended recurrence intervals and/or with high seismic loss poten- tial. B3 Definition of criteria and a program for seismic evaluation and retrofit of exist- ing buildings. B4 Provision for limitations on liability of local jurisdictions and their building official(s) when they provide and permit criteria (as in B3) for evaluation and retrofit design which is less stringent than building code requirements for new construction. B5 Permission for voluntary seismic upgrades without mandated upgrades for non- safety related functions. B6 Support for development of programs and procedures and of professional state and local building staffing to effect LRMs. B7 Support for dislocated tenants during seismic retrofit programs. BS Continued research directed at reducing costs for seismic construction, both new and existing. B9 Continued work to incorporate a geotechnical component into model seismic codes. B10 Continued research into development of codes that emphasize property damage control and maintenance of function over and above critical life-safety protec- tion. Supporting elements B3 through BS support LRMs applicable to existing construction. B3 involves definition of seismic retrofit criteria and state legislation in order to make retrofitting cost-effective -andfeasible at standards more realistically below those required by current model seismic provisions and supports LRMs 10, 11, 12, 13, 14, and 15). B4 provides (through state legislation or local ordinance) for limitations on liability of local jurisdictions and their building officials overseeing a seismic retrofit program. This action supports LRMs 10, 11, 12, 13, 14, and 15. B5 provides for the development of legislation, etc., to facilitate seismic upgrading without triggering costly mandatory upgrades for non-safety-related func- tions (in support of LRMs 10, 11, 12, 13, 14, and 15). 4-15 B6 supports all building practice LRMs by providing federal funding for state and local governments in support of programs, procedures, and professional staffing (e.g., structural engineers, architects, inspectors) to effect LRMs. B6, like L10, may require federal assis- tance. This partial federal assistance may be warranted by reductions in contingent federal liabilities as a result of state and local activities and so may not be considered an additional burden on a federal earthquake insurance program. Consistent with this recommendation is that by Selkregg et al. (1984) for support for state-level commissions to oversee and en- courage loss-reduction activities. Suitable representatives of private industry, including in- surers, engineers, architects, and others engaged in loss-prevention activities, should be in- cluded on these commissions. B7 addresses concerns identified in the stakeholder analysis by providing for inclusion of a social program to assist dislocated tenants during major seismic retrofit construction. Its inclusion as a supporting element derives from discussions by B. Zeidman at the project workshop and by Comerio (1989). In some instances seismic retrofit programs in seismic zone 4 may involve serious social dislocations. Examination of and minimization of these dislocations should be an integral element in these retrofit programs. B8 recommends continued research efforts in order to reduce future costs of seismic design and retrofit. As our socioeconomic analysis confirmed, these costs are very significant factors in assessing the cost-efficiency of LRMs; even small reductions in costs for the LRMs proposed can have large-scale aggregate benefits. B9 derives from cautions voiced at the third Advisory Panel meeting that one must not assume that the geotechnical component of LRMs for new construction is adequately covered in existing model building codes or in NEHRP provisions. Strengthening code con- cerns for geotechnical remediation of landslide, liquefaction, and subsidence sites and sub- divisions, and continued progress in incorporating local high shaking relative site response factors into codes is important in assuring that sound landuse practices are buttressed by and often enforced in terms of sound building practices. Status Ouo Obstacles to Implementation of LRMs Unfortunately, because of the many proposals that face them, key decision makers of- ten make decisions simply in order to reduce the number of proposals remaining in their purview. (See Cohen et al., 1972, and March and Olsen, 1976.) Instead of evaluating each problem logically and searching for optimal solutions, decision makers are overwhelmed so that, fundamentally, 4-16 o problems, solutions, and participants are seen as separate streams which flow through the system, and o system outcomes depend heavily on coupling the three streams in a timely fashion to take advantage of an opportunity for a decision. In applying this approach to describe policy formulation in complex governmental organiza- tions, Kingdon (1984) defined the three major process streams as o problem recognition, o formation and refining of poicyproposals, and o politics. Figure 4-1 presents a general illustration of the major elements contained in a model for as- sessing policy formulation, adoption, and implementation. With respect to problem recognition, risk analysis has been the chief means discussed in this project to explain the extent to which earthquakes pose a problem. Risk evaluation has been used to answer critical questions related to social, technical, political, legal, and economic components of the problem streams. Inside advocates (i.e., policy entrepreneurs) who understand risk results and windows of opportunity will be able more fully to bring about an effective coupling of the problem, policy, and political streams. In other words, the process of risk and decision analysis, as discussed in Section 2.1, is essential to policy for- mulation, adoption, and implemenation. In the project workshop, participants indicated that LRMs facing the greatest obstacles to implementation are: o those pertaining to existing commercial/industrial/institutional structures and o those in smaller local jurisdictions (e.g., those with less financial, technical, and ad- ministrative capacity to effectuate LRMs), Additional barriers already discussed include o disincentives to mitigate hazardous publicly owned buildings when 75 percent of all disaster costs are borne by the federal government in a Presidentially declared dis- aster and o disincentives to mitigate when insurance rates are insensitive to cost-effective mitigations. 4-17 Policy Policy Formulation and Adoption Risk Analysis mplementatio I Risk Problem Risk Estimation Expected Loss and Consequences __liiiiiiii Pollcy/Solution Implementation Stream and L Pol Policy Evaluation Agenda Windw Social, Technical Administrative, Objectives Acceptable .Opportunit Political, Legal, and --4 Risk for Choice I-a. Programs Economic 80 __~ Components Risk Evaluation Political Stream I Participants/ Adminlstratoi rs Inside Advocates / Policy Entrepreneurs Stakeholders Engineers Researchers Budget/Economy Contractors Analysts Environment P __ l .1 111411WWRJN II I I I WOMEN OWN 11.11 11INS WINNOWER -1-115 laid I Nim-111,11111111 1111,11INNIIIIII 111,111's I IN III-pill" Figure 4-1. A Risk Management Model for Assessing the Process of Policy Formulation, Adoption, and Implementation Means to overcome barriers faced in the process of implementation LRMs for natural hazards include: o identifying, mapping, and classifying natural hazard zones and estimating earthquake losses so as to clarify the extent of the risk and facilitate ILRM im- plementation, o creating procedures and data bases to facilitate benefit-cost analyses, o reducing significant differences among model building codes in order to provide consistency among municipalities, o increasing technical capacity among local planning and buildingregulation depart- ments, o developing political constituencies to support LRM i mplementation, and o understanding the short-term focus in decision-maaking so as to maximize efforts- within the political system. Workshop participants also proposed strategies to increase implementation success. These include: o making LRMs acceptable to the largest number of people and interest groups in the community, o creating advocacy groups within professional code organizations to support code changes, o having federal staff work with ad hoc committees of code organizations to facilitate the adoption of LRMs, o promoting simple code provisions for one-to-four family residences, and o promoting education/training programs to improve seismic design and construction practice. Workshop participants emphasized linkage of earthquake insurance rate-structuring programs to those insurance programs for other perils. They also emphasized the integra- tion of programs for public buildings, disaster relief, and insurance. 4.5 Summary The stakeholder analysis highlighted the fact that current disaster relief and other federal policies constitute disincentives to state and local governments in implementing loss- reduction measures. Lenders benefit both from implementation of cost-effective mitigations and increased volumes of earthquake insurance purchase. Under current circumstances, federal taxpayers. may benefit from implementation of mitigations for public construction 4 -19 even to the extent that federal cost-sharing programs may be considered cost-effective. Providing these cost-sharing programs will make benefit-cost ratios for state and local mitigations of public buildings more attractive. (See Appendix D.) Using the results of the socioeconomic analysis, the Project Workshop participants, in- cluding the Advisory Panel and other project reviewers, and the project team were able to synthesize and further develop the fifteen LRMs that were recommended for inclusion in a federal insurance program. These fifteen LRMs affect all regions of the country. They em- phasize new construction in seismic zones 2, 3, and 4 and existing construction in seismic zones 3 and 4. They emphasize landuse practices in seismic zones 3 and 4. All LRMs may be considered community-based in that key loci of enforcement lie in state and local govern- ments. In effect, the fifteen LRMs can be incorporated into potential earthquake or- dinances. For a particular community, details of earthquake ordinances would be defined by the seismic zone (i.e., 0, 1, 2, 3, and 4) of the community in question in conjunction with those LRMs in Tables 4-1 and 4-3 applicable to the zone. In a federal program there is a need for model earthquake ordinances. The fifteen LRMs developed in this report provide adequate guidance for ordinances which would be cost-effective and acceptable. Project Advisory Panel members, workshop participants, reviewers, and the project team also developed supporting elements for these promising LRMs, and strategies for overcoming status quo obstacles to implementation of these LRMs. In the majority of cases, these supporting elements would have low costs relative to a national program. Higher costs might be involved in federal assistance and cost-sharing programs for training and staffing and for phasing or retrofitting potentially hazardous public buildings. In Section 5, we main- tain that some of the higher costs of supporting promising LRMs may be justified with reference to programs specifically targeted to reduce existing contingent federal liabilities owing to disaster relief policy and other federal statutes. 4-20