The Impact of Uncertainty in Managing Seismic Risk: The Case of Earthquake Frequency and Structural Vulnerability
Summary: The research presented in this paper is an update to the preliminary analyses presented at the Public Private Partnership 2000 forum that was held in Washington DC in December of 1997. Using more accurate cost and benefit information than was available at that time, this paper presents a more precise look at the effects of mitigation on reducing losses. Furthermore, the issue of uncertainty in estimating earthquake frequency and structural vulnerability is systematically analyzed.
The authors begin by explaining in detail why the need to understand the causes and degree of uncertainty in the loss estimation process is so important. Each of the four main steps in the methodology of earthquake loss estimation (hazard definition, inventory exposure characteristics, inventory damage, and loss calculation) is shown to have a characteristic "mix" of aleatory (ie random), epistemic(lack of knowledge), modeling (simplification of natural processes to allow model building to take place), and parametric (intrinsic to estimating values) uncertainty.
Two specific studies were performed to examine uncertainty in the frequency of earthquake events in defining hazard, and the vulnerability of residential structures to estimate inventory damage. Risk Management Solutions Incorporated and EQE International provided data on the residential makeup of Oakland and Long Beach CA respectively. Pre-1940 wood frame houses, good candidates for quake damage mitigation via the bracing a bolting of cripple walls to the foundation, were selected for inclusion in the hypothetical "book of business" to be examined in each city. In addition, surveys were sent to contractors and structural engineers to solicit better estimates of damagablity of these structures and the actual costs of mitigation. The authors set the time horizon for homeowners' "to-mitigate or not-to-mitigate" decision at thirty years. The "worst case loss scenario" for the hypothetical insurance companies was defined as having an exceedance probability figure of .01. Earthquake frequency and structure vulnerablity curves were provided by RMS and EQE and were evaluated to yield a 90% confidence interval.
The concluding section of the paper presents tables of numerical estimates for "expected losses", "insolvency probability", "book of business effects", and "profitability" for the hypothetical insurance firms. Tables are also presented which give of estimates of "insured homeowner total expected cost"(mitigation and insurance), "uninsured homeowner's total expected cost" "homeowners total worst case loss versus, homeowners total expected cost", and "impact of uncertainty on homeowners' decision to mitigate"
The authors conclude on the basis of these numerical estimates that for both the Oakland and Long Beach cases, that there is significant impact of uncertainty on insurers insolvency probability, book of business, and profitability. Further, they show that mitigation can be of significant benefit to insurers by increasing profitability due to lower capacity restrictions. For homeowners the picture is less dramatic. While the outcomes of homeowners' decision processes to insure and/or mitigate can be strongly effected by uncertainty in earthquake frequency and structure vulnerability, other factors such as assumed discount rate, time horizons, insurance pricing and coverage limits are expected to be more important.
Updated February 29, 2000
Copyright ©1995-2002 The Wharton Financial Institutions Center and The Wharton Risk Management and Decision Processes Center. All rights reserved.