Dr Paolo Bazzurro describes how catastrophe modelling can be used to quantify the potential impact of earthquakes on corporate facilities.

Quantifying the potential impact of earthquakes on corporate facilities located in regions of seismic hazard is of primary interest to property owners, insurance and reinsurance companies, capital lending institutions, local government agencies and structural engineers. Each is likely to have a different viewpoint and different requirements.

Sophisticated owners and corporate risk managers cope with seismic risk using a variety of strategies. These range from establishing self-insurance programmes (expense funds for post-earthquake emergency, for example, or proactive seismic retrofit) to buying earthquake insurance coverage and, more recently, to taking advantage of alternative risk transfer arrangements, including captives and insurance-linked securities or catastrophe bonds. Self-insurance programmes can be effective for relatively minor events, while insurance coverage and alternative risk transfer tools are the rational choice to manage large scale losses that cannot be absorbed internally.

Seismic risk assessment
Seismic risk assessment studies are crucial to providing risk managers and owners of corporate facilities with the necessary knowledge to design an optimal risk management strategy. Information on the likelihood of losses of varying amounts is vital for making informed decisions on the adequacy of out-of-pocket expense funding, for example, how to structure a cost-effective insurance programme or whether to access the capital markets directly.

Seismic risk analyses can be tailored to fit the specific needs of the corporation or the owners that commission them. Typically, they focus on one or a combination of the following assessments:

  • loss estimation - the likelihood of losses of different amounts in a given exposure period;

  • damage estimation - the likelihood of building damage levels, ranging from the onset of damage to complete collapse, within a given exposure period;

  • `system performance' estimation - the likelihood that a given level of system performance will not be met in a specific time frame; and

  • l earthquake occurrence probability estimation - the likelihood of occurrence of events of a specified magnitude within a predefined geographical area.

    Loss estimation
    A loss estimation analysis can be performed either for a specific earthquake scenario (a magnitude 7.5 on the San Andreas fault, for example) or for large suites of possible future events. In many cases, such an analysis will provide all the information necessary to manage most risk-related decisions. Catastrophe modellers use probabilistic procedures that couple both the random aspects of earthquake phenomena and the response of the structure to different levels of ground shaking.

    A building vulnerability assessment methodology developed by AIR Worldwide estimates building response by subjecting a computer model of the structure to increasing levels of ground shaking. The computed response can be characterised in terms of various engineering parameters, such as maximum roof displacement, floor acceleration or interstorey drift, that past earthquakes have shown to be well correlated with observed damage.

    Application of a cost model identifies specific repair/replacement strategies appropriate to the level of damage, and the costs of implementing those strategies are used to obtain a final estimate of losses. The most advanced of such studies take into account the uncertainty intrinsic in parameters of the engineering models (uncertainty in the rate of activity of a particular fault, for example, or in the strength of materials in an existing building). The explicit treatment of such uncertainty allows the quantification of the confidence in the computed loss estimates, which is extremely valuable information to the decision maker.

    Damage estimation
    In addition to probability distributions of monetary loss, corporate risk managers may be interested in estimates of actual physical damage likely. Such studies might be performed when occupation of a particular building is of paramount importance for continued operation. The likelihood that a building will be either `yellow' or `red' tagged after an earthquake, which limits access to the facility, is the controlling decision variable.

    The colour of the placard assigned to a damaged building depends, of course, on the level of physical damage that the building has sustained. However, sophisticated studies also recognise that the degree of occupancy restriction also depends on the hazard from ground shaking due to the high aftershock activity that almost invariably follows large earthquakes. It is the combination of decreased capacity and increased ground motion hazard that controls the occupancy status. A slightly damaged building may be `yellow tagged', not because of a great loss in structural capacity, but because the chance of further ground shaking greatly increases in the immediate aftermath of an earthquake. The aftershock hazard, however, diminishes within days, and in some cases the occupancy restriction may be lifted even if no significant repair has yet been made.

    `System performance'
    System performance studies are most often associated with infrastructure or so-called `lifelines', such as highways, water and power supply, and distribution systems that provide critical services to the corporation. Given their spatial extent, lifelines tend to be particularly vulnerable to earthquakes. A system performance study can evaluate the likelihood that the system will not meet a predetermined standard of performance.

    A lifeline-specific failure may be defined in a number of ways, such as the disruption of service from point A to point B for more than some specified amount of time. This could be disruption in the delivery of necessary supplies from vendors, the delivery of goods to customers, or availability of electricity for plant operations. A system performance analysis can be used to determine whether sufficient redundancy currently exists to minimise the interruption of operations. In addition, seismic hazard may not be the same for all system components, and an analysis can reveal which components are more vulnerable than others. Seismic risk analyses of lifelines are among the most challenging to engineers.

    Earthquake probabilities
    Estimating earthquake hazard, or the probability of earthquake occurrence, is one of the building blocks of all the seismic risk analyses discussed above. But it takes on an importance of its own, independent of such analyses, when it is used as a tool to define the likelihood of occurrence of events that trigger coverage under earthquake-related catastrophe bonds. In earthquake hazard studies, catastrophe modellers use sophisticated statistical techniques based on historical seismicity, paleoseismic data (evidence of ancient events, usually obtained by trenching) and geodetic data measurements taken by global positioning system (GPS) instruments of the earth's crustal movement to estimate rates of earthquake occurrence.

    Loss estimation studies, whether for a single facility or for a portfolio of corporate properties, often consider both direct losses to buildings, and contents and indirect losses, such as those due to business interruption. The concept of direct costs is straightforward to grasp, but they are often not the main contributor to the enterprise-wide losses induced by a natural disaster. Indirect losses stemming from local and regional effects can overshadow direct losses, and may include not only the downtime of the facility itself, but downtime experienced by suppliers (and even suppliers of suppliers).

    Business interruption may be extended if access to the corporate facilities is restricted because of infrastructure failure or because demand for a product or service has temporarily vanished in the aftermath of the earthquake. This may very well happen, for example, to a tourism-based San Francisco enterprise should a large earthquake strike the Bay Area. The San Francisco facility may be fully operational after the event, but the disruption of one or more bridges across the bay could prevent employees from reaching the facility and tourist numbers may temporarily decline. In other cases, business interruption losses may be reduced if the product can be manufactured at a different facility with almost no extra cost. A realistic quantification of indirect costs is case-specific and may require considerations that go beyond the local facilities to the entire corporation.

    We have developed customisable seismic risk assessment methodologies that are designed to provide corporate owners and risk managers with the requisite information to answer a wide variety of questions, such as: How severe can earthquakes relevant to my location be? How often are they likely to occur? What is the range of losses that can occur? What is the probability of any given loss amount? How much insurance should I buy? How large should my deductible be and where should it apply?

    The answers to these and other questions can be used for corporate risk management planning, insurance-buying decisions, developing a seismic retrofit strategy or optimising performance-based design of new facilities.

    By Paolo Bazzurro

    Dr Paolo Bazzurro is Manager of Engineering Analysis in the San Francisco office of AIR Worldwide, now part of Insurance Services Office (ISO).

    http://www. www.air-worldwide.com .