• Keine Ergebnisse gefunden

The Role of Risk Assessment in Facility Siting: An Example from California

N/A
N/A
Protected

Academic year: 2022

Aktie "The Role of Risk Assessment in Facility Siting: An Example from California"

Copied!
20
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Working Paper

T H E R O L E O F R I S K A S S E S S M E N T IN FACILITY SITING:

AN E X A M P L E FROM CALIFORNIA

John

W.

Lathrop

WP-80-150 October 1980

International Institute for Applied Systems Analysis

A-2361 Laxenburg, Austria

(2)

NOT FOR QUOTATION WITHOUT PERMISSION OF THE AUTHOR

THE ROLE OF RISK ASSESSMENT IN FACILITY SITING:

AN EXAMPLE FROM CALIFORNIA

John W. Lathrop

WP-80- 150 October 1980

Working Papers a r e i n t e r i m r e p o r t s on work of t h e Interrlational Institute for Applied S y s t e m s Analysis a n d have received only limited review. Views o r opinions expressed h e r e i n do n o t necessarily r e p r e s e n t t h o s e of t h e Institute o r of its National Member Organizations.

INTERXATIONAL INSTITUTE FOK APTLIED SYSTEMS ANALYSIS A-2361 Laxenburg, Austria

(3)

CONTENTS

11. DIFFERENCES IN DEFINITION

AND

EVALUATION OF RISK Focusing the Discussion

Technical Perspective Societal Perspective

ImpIicaions of the Two Perspectives

111. DIFFERENCES IN APPROACHES TO RISK MLVAGEMENT The Technical Approach

The Political Approach

IV. THE ROLE OF ANALYSIS IN THE POLITICAL PROCESS Disaggregated Nature of the Process

Role of Analysis in forming Siting Legislation, Regulation

Role of Analysis in Hearings Section Conclusions

V. CONC1,USIONS: FUTURE RESEARCH DIRECTIONS REFERENCES

(4)

THE ROLE OF RISK ASSESSMENT IN FACILITY SITING:

AN EXAMPLE FROM CALIFORNIA

John W. Lathrop

As the number and scale of energy facilities increase, govern- m e n t agencies a r e faced with more and more difficult decisions involving the management of societal risk. Since t h e risks can be complex and not well understood. risk management decisions c a n be extremely difficult t o make. I t would seem, then, t h a t t h e r e should be an important role for formal analyses in assessing risks 'to aid those decisions. This paper ex- plores t h a t role in the context of the a t t e m p t e d siting of a Liquefied Na- tural Gas (LNG) facility o n the California coast. While the first site appli- cation was filed in 1973, as of this writing (October 1980) no site has been approved. The story of those eight years, a fascinating case study in so- cietal risk management, has been told and analyzed with insight by two of t h e participants in the process: William Ahern (1980a,b) and Randolph Deutsch (1980). This paper concentrates on one aspect of one episode in t h a t siting process, taking t h e point of view of t h e frustrated decision analyst, trying to understand the role his tools play--and could play--in t h e political process of siting. After investigating t h a t role, the paper concludes with some brief and very general recommendations as t o what r e s e a r c h is called for t o make those tools more useful.

he

research reported in this paper is supported by the Bundesmin- isteriurn fiir Forschung und Technologie, F.R.G., contract no.

321/7591/RGB 8001. While support for t.his work is gratefully ack- nowledged, the views expressed a r e the author's own, and a r e not neces- sarily shared by t h e sponsor.

(5)

In early 1977, s t a t e and federal regulatory agencies were favor- ing a n LNG terminal site a t Oxnard, California. As p a r t of t h e site appro- val procedure, assessments of t h e risks to surrounding a r e a s due to t h e LNG terminal h a d been done by the federal regulatory staff (Federal Power Commission, FPC). and by consultants hired by the gas company 2 (Science Applications. Inc., SAI). Both risk assessments estimated t h e probabilities of e a c h of many possible accidents and derived from t h e m various measures of t h e risk in probabilistic terms. such a s expected numbers of fatalities, probability of fatality p e r exposed person p e r year, etc. The implicit goal of both of these assessments was t o m e a s u r e t h e risk in s u m m a r y t e r m s t h a t could assist in the determination of whether o r not t h e risk was low enough for t h e site t o be acceptable. Both assess- m e n t s indicated t h a t t h e LNG risks t o Oxnard were extremely low or negligible (SAI 1975; FPC 1976). The appropriate federal agency deemed t h e Oxnard terminal acceptably safe, and approved it in December 1977.

Considering all of these events, things s e e m e d t o be moving smoothly to- ward approval for the Oxnard terminal. However, t h e city of Oxnard had commissioned another consulting firm, Socio-Economic Systems. Inc.

(SES) t o d o a n environmental impact report. As p a r t of t h a t effort, SES did its own risk assessment, which combined assumptions and model results from SAI, FPC, and U.S. Coast Guard studies t o calculate about 5.7 expected annual fatalities a s a summary m e a s u r e of the risk of t h e LNG terminal (SES 1976). This was about 380 times higher t h a n t h e corresponding estimate by SAI. However, according t o Ahern (1980a) t h e politically m o r e important p a r t of the SES report was a s e t of descrip- tions of several deterministic worst case scenarios, with flammable vapor clouds covering up t o 70,000 people, presented without estimated proba- bilities of occurrence. The publication of those scenarios "electrified op- position t o t h e terminal," t o quote Ahern.

In the face of opposition to t h e terrninal based on concern for safety, t h e California s t a t e legislature passed the LNG Terminal Act of 1977 (S.B.1081, September), which excluded t h e Oxnard site. That bill may have passed with or without t h e SES worst c a s e scenarios, b u t t h e fact remains t h a t t h e SAI and FPC probabilistic risk assessments did not help t h e g a s company o r t h e regulatory agencies gauge the actual politi- cal acceptability of the Oxnard site. The assessments could have been in- tended for any of several purposes, among t h e m two in particular: t o warn t h e company away from a site where i t would b e imposing politically unacceptable risk, or t o convince all appropriate government agencies t h a t t h e s i t e was acceptably safe. Measured against either of these two purposes, the SAI and FPC risk assessments failed. This paper p r e s e n t s and discusses four very general reasons for this failu.re:

2 ~ h e FPC staff m o r e o r less b e c a m e t h e Federal Energy Regulatory Com- mission (FERC) staff in t h e course of t h e creation of t h e Department of Energy in 1.977. Their work will be referred t o a s t h e FPC assessment throughout this paper.

(6)

differences in definition and evaluation of risk differences in approaches t o risk management

t h e n a t u r e of t h e role of analysis in t h e political process, t h e n a t u r e of t h e political process itself

The last section of t h e p a p e r summarizes the lessons t h a t c a n b e learned f r o m t h e discussions of t h e s e points, bringing t h e m together in t h e form of very brief descriptions of t h e most promising a r e a s for r e s e a r c h directed a t improving t h e usefulness of risk a s s e s s m e n t i n t h e political siting process.

11. DIFFERENCES IN DEFINITION AN11 EVALUATION OF RISK

Given t h e e x t e n t of controversy surrounding questions of socie- tal risk, it s e e m s odd t h a t t h e d e b a t e goes on without a clear definition of what risk is. In fact, this section suggests t h a t differences in definitions of risk a r e basic p a r t s of t h e problem of managing societal risk.

A. FOCUSING THE DISCUSSION

An LNG siting problem, a s discussed here, consists of two inter- dependent decisions: whether o r not t,o have the LNG project, and if so, where t o site t h e plant(s). A disinterested observer might view t h e s e de- cisions a s involving t h e consideration of very uncertain benefits and losses, all viewed in relation t o t h e alternative action (pipeline, oil, coal, nuclear, conservation. o t h e r site). The LNG plant could c a u s e a n uncer- tain reduction in t h e num-ber and severity of supply interruptions, an uncertain change in t h e finances of t h e utility and its c u s t o m e r s , a n unc- ertain loss of flora and fauna, an u n c e r t a i n degradation in people's enjoy- m e n t of the coast, a n uncertain r e d u c t i o r ~ in air pollution h e a l t h effects, and a n uncertain increase in accident-related loss of life and limb. While t h e s e s t a t e m e n t s suggest directions of LNG plant effects ( r e d u c t i o n , in- c r e a s e , etc.), those directions depend o n what alternative is a s s u m e d t o t a k e t h e place of t h e LNG project if i t is not p e r m i t t e d , and on assump- tions made in t h e r e l a t e d analyses. While i t would be interesting t o con- s i d e r all of t h e factors listed, this p a p e r focuses on health effects, includ- ing air pollution effects and accident.-related loss of life and limb, factors c e n t r a l t o the example given in t h e introduction. The word "risk" is used t o denote some m e a s u r e of those uncertain losses, reflecting t h e typical use of t h e word in s u c h t e r m s a s risk assessment and societal risk. Even with this focus, uncertain health effects a r e many-dimensional things, so t h a t any definition of risk rnust involve assumptions concerning how those losses a r e t o b e evaluated. That is the source of a key a s p e c t of t h e socie- tal risk rnanagement problem: There a r e two different types of definitions of risk, which arise f r o m two basically different perspectives o n risk, technical and societal. These two perspectives can lead t o substantial.ly different evaluations of t h e acceptability of the risk resulting from an LNG alternat.ive.

(7)

B. TECHNICAL PERSPECTIVE

From the technical perspective, risk is some probability distri- bution over s e t s of health effects. For example, in one paper Keeney (1980a) defines the risk of a technology as a probability distribution over s e t s of probabilities

1

p i , i = l ,

. . .

, N

1,

each s e t denoted in short a s pi

1,

where p, is t h e probability t h a t the i t h individual of a group of N individu- als will die due t o the technology before t h e end of the next time period.

The probability sets

I

p i

1

a r e constructed in such a way t h a t within e a c h set, e a c h probability is independent of any other. The probability distri- bution over those sets is required t o represent probabilistic dependencies between t h e fatalities. For example, in the LNG case t h e probability dis- tribution over s e t s

I

p, could be essentially the probability distribution over accidents. To illustrate, suppose only one accident was possible for an LNG facility. that accident could occur with an annual probability of

and it would expose e a c h of t h e nearest 100 people t o a very high fa- tality risk of 80%. While t h e possibility of such an accident results in a n annual mortality probability of . 8 * 1 0 - ~ for each of those people, t h a t number alone does not reflect t h e f a c t t h a t t h e fatalities would occur in one large accident of about 70 t o 90 fatalities, if any occur a t all. The probability distribution over sets

I

pi

1

can be used t o calculate both t h e probability of fatality p e r exposed person per year and t h e potential for high-fatality accidents. P a s t risk assessments have used these and other summary measures of t h a t probability distribution. The SAI a n d FPC stu- dies r e f e r r e d to earlier used among other measures the probability of fa- tality p e r exposed person per year, a n index convenient for comparing in- dividual risks from LNG t o risks from other sources. In the Reactor Safe- ty Study ("Rasmussen report"), t h e risk of nuclear power was represented by a reverse cumulative probability distribution over numbers of fatali- ties per year p e r reference reactor (USNRC 1975). One can read directly from a g r a p h of t h a t distribution the annual probability t h a t t h e number of fatalities will exceed 10, will exceed 100, will exceed 1000, etc. It fol- lows t h a t such a distribution, often referred t o as a Rasmussen curve, very directly indicates the potential for catastrophe. The same type of distribution has been used with LNG risk assessments (Hazelwood and Philipson 1977). Some risk assessments go one s t e p further and reduce t h e distribution t o t h e expected number of fatalities, a rneasure reported in t h e FPC and SES studies (see also Pate' 1978).

C. SOCIETAL PERSPECTIVE

In contrast t o t h e fairly straightforward technical-perspective risk measures listed above, extensive psychological research in t h e field of risk perception has suggested many more aspects t o be included in a measure of risk t h a t would be more sensitive to t h e concerns of political and societal processes ( ~ i s c h h o f f , e t al. 1978; Linnerooth 1978; Otway and Pahner 1976). Some of t h e s e aspects a r e listeci and briefly discussed h e r e as a way of describing risk from a societal perspective. This section is not intended t o cover these aspects in any detail, as they have all been h e a t -

(8)

ed at length in t h e cited papers. The aspects a r e listed roughly in order from t h e most easily adapted to a simple technical index t o t h e most dif- ficult.

1 . P o s s i b i l i t y of c a t a s t r o p h e

Some technologies cause fatalities t h a t occur one a t a time, s c a t t e r e d widely over a geographic area. An example would be the fatali- ties due t o air pollution caused by burning substitutes for natural gas.

While those effects a r e "spread out" over space and time, t h e fatalities from a major LNG accident would be "bunched" into a catastrophe a t one place and time. That bunching can be very important for how society evaluates t h e risk. While the Keeney and Rasmussen measures reflect bunching, o r possibility of catastrophe, t h e other technical-perspective measures listed above, probability of fatality per exposed person p e r year and expected n u m b e r of fatalities, do not.

2 . I n e q u i t y of i m p a c t s

An LNG plant t h a t exposes people in its immediate neighborhood to a risk for t h e benefit of all Californians is a case where inequity is im- portant. The air pollution due to a lack of natural gas would be a risk spread more widely over t h e benefiting population, and so would be more spatially equitable. A risk t h a t may s e e m acceptable by a n aggregate measure, such a s expected number of fatalities, may not be politically ac- ceptable if it is inequitably distributed.

3. D e g r e e of c o n t r o l

This a s p e c t is a much more general version of t h e voluntary/involur~tary distinction rnade by S t a r r (1969). The central con- c e p t h e r e is t h e level of participation of each potential impactee in each of two decisions: t o expose himself to the risk, and t o deploy t h e technolo- gy in the first place. S t a r r presented evidence which he interpreted as indicating t h a t society has a much higher threshold of acceptability for risks involving voluntary exposure t h a n for risks incurred involuntarily.

Arguments against that hypothesis have appeared in the literature. most recently in t h e p a p e r by Slovic, Fischhoff and Lichtenstein (1980), which concludes t h a t a p p a r e n t aversion to involuntary risk can be b e t t e r ex- plained by t h e higher potential for catastrophe and inequity t h a t often acconlpany t h a t type of risk. However, in another paper t h e s a m e first two authors s t r e s s the importance of public participation in t h e second of the two decisions listed above, concerning t h e deployment of the technol- ogy (Slovic and Fi sc hhoff 1979).

(9)

4. Attribution

This is an important aspect of societal reaction t o a risk t h a t is often overlooked because of the cause-specific way risk assessments a r e performed. There is no doubt as to the cause when a person has been in- jured o r killed by an LNG accident. On the other hand, i t m a y not even occur t o someone when an elderly relative dies t h a t he may have lived longer if t h e r e had been less air pollution. Beyond t h a t , i t would be irn- possible t o attribute any death in particular t o the increment in air pollu- tion caused by a low share of natural gas in t h e energy mix. I t may not seem reasonable in a narrowly prescriptive sense t o evaluate easily- attributed health effects any differently than more subtly-linked health effects. However any more broad, strategic effort t o assess t h e political acceptability of a project should t a k e attributability into account.

5. Non- decision comparisons

The search for criteria for acceptable risk often falls back on comparisons not involved in actual decisions. For example, comparisons a r e often made between LNG o r nuclear risk and risk of disease, or risk of the r e s t of t h e fuel cycle, or risk of smoking an e x t r a cigarette. Yet very few people choose between living near an LNG plant a n d exposure t o risk of disease. The l a t t e r risk is always there. as is t h e risk of t h e r e s t of t h e fuel cycle, so t h e LNG risk adds t o it. Perhaps even fewer people choose between living near an LNG plant and smoking an e x t r a cigarette, though t h a t decision involves some interesting interactions. Yet the use of such benchmarks in evaluating societal risk may be helpful in placing assess- m e n t s of risks in more meaningful terrrls t h a n probabilistic measures (Lichtenstein e t al. 1978).

6. Non- probabilistic evaluation

Perhaps the most serious mismatch between technical and so- cietal perspectives lies in t h e societal sensitivity t o descriptions of severe outcomes with no regard for t h e estimated probability for those out- comes. This aspect may have been a key one in t h e rejection of the Ox- nard site mentioned earlier. Worst case scenarios. with uncalculated but extremely low probabilities, s e e m e d t o be effective in furthering opposi- tion t o the site. While it may be tempting t o consider this an example of faulty information processing, of misuse of technical information, in fact there a r e reasonable justifications for this aspect of risk evaluation:

doubt in the ability of probabilistic assessments t o avoid underestimating probabilities of accidents, sensitivity t o t h e feeling of dread t h a t could ac- company living in t h e shadow of a vividly described catastrophe, or desire for a resilient, soc:ial system. I-Iowever, this aspect of risk evaluation is directly incompatible with the probabilistic orientation of any of th.e technical-perspective risk measures listed above. One difficu1.t~ with any prescriptive evaluation incorporating this aspect is t h a t t h e most severe outcome described is dependent upon the imagination and motivation of t h e analyst. Yet once again, any a t t e m p t to gauge the political accepta-

(10)

bility of a risk should t a k e this aspect into account.

7. Indirect effects

Some aspects of a technology having little to do with possible health effects m a y have a g r e a t deal t o do with a societal evaluation of risk. Those aspects include t h e degree of centralization, how closely a new technology is linked with a high-consumption lifestyle, e t c . While these aspects s e e m far removed from evaluating the' risk t o life and limb of LNG, t h e y should be included in any effort to understand possible oppo- sition to a n LNG site.

D. IMPLICATIONS O F THE TWO PERSPECTIVES

The difference between the two perspectives on risk described above may have very d i r e c t relevance for t h e role of a risk assessment in a political risk management process. This is illustrated by t.he LNG siting example given in the introduction. The SAI and FPC risk assessments could be viewed a s adopting t h e technical perspective. While p a r t s of t h e SES assessment adopted t h a t same technical perspective, o t h e r p a r t s in- cluded worst case scenarios t h a t caught t h e attention of people with a so- cietal perspective. While the gas company and regulatory agencies were planning on t h e basis of the technical perspective of risk, t h e political process they had to deal with was quite naturally sensitive t o the societal perspective. The result: probabilistic risk assessments were not effective in guiding the g a s company a.nd regulators, nor were those assessments persuasive in the political process. From the other point of view, t h e pol- itical process was not aided by t h e analytic effort t h a t went into t h e SAI and FPC assessments. Clearly, no p a r t y was served by the mismatch between t h e technical perspectives of t h e probabilistic risk assessments a n d the societal perspective of the political process. Just a s clearly, if risk assessrnents a r e t o serve t h e applicant seeking to site a hazardous facility and the political process seeking t o manage the resulting risk, t h e n either the assessments should adopt more of a societal perspective of risk, or t h e political process should a d a p t itself t o assessing risks from a more technical perspective.

As the previous paragraphs have made clear, t h e r e a r e many more aspects t o the evaluation of societal risk than summary m e a s u r e s of probability distributior~s on health effects. The limited m e a s u r e s of past probabilistic risk assessments make i t easy t o suggest t h a t adding m o r e evaluation dimensions t o an assessment will t h e n provide more use- ful guidance t o t h e parties in the siting process. In fact, decision analysts have proposed significant and interesting ways t o expand probabilistic risk assessments to explicitly incorporate dimensions t h a t represent; so- cietal concerns (Bodily 1980; Keeney 1.980 a,b). These approaches a r e based on mu1tiattri.but.e utility analysis, which c a n generate evalua.tion functions t h a t not only r e p r e s e n t value tradeoffs between several dimen- sions, but also r e p r e s e n t attitudes toward uncertainty (Keeney a n d Raiffa 1976). Such functions seem t o hold g r e a t promise for bringing more of a societal perspective t o risk assessments, making t h o s e assessrnents nlore

(11)

useful and used. However, a s the next sections demonstrate, t h e r e a r e m o r e fundamental problems in t h e interface between risk assessment and t h e political process t h a n a lack of dimensions.

111. DIFFERENCES IN APPROACHES TO RISK MANAGEMENT

Beyond differences in how risks a r e evaluated, political organi- zations cope with the uncertainties inherent in risk management in ways very different from t h a t assumed in typical risk assessments. There a r e two types of uncertainty involved. First, t h e r e is uncertainty in t h e out- come of any action since t h e future is full of unknowns, accidents may or may n o t happen, physical processes a r e not completely understood, etc.

Second, t h e r e is uncertainty in t h e results of a n y analysis. F r o m a partic- ularly narrow analytic point of view, looking only a t a decision t o b e made immediately, t h e r e is no operational difference between these two uncer- tainties. But from a m o r e strategic point of view, t h e second uncertainty reflects how likely t h e results of an analysis a r e t o be subject t o argu- ment, o r worse yet, how likely those results a r e t o change. The possibility of analytic results changing within a few y e a r s gives very reasonable pause t o a government agency about t o commit resources t o an action with effects spanning decades (e.g., setting a regulation or allowing a plant to be built). The next two sections explore how t h e analytic a n d pol- itical communities cope with these two uncertainties.

A. THE 7'ECHNICAL APPROACH

Expectecl numbers of health effects is a risk rrleasure t h a t does not reflect outcome uncertainty, though i t is sometimes r e p o r t e d with a n e r r o r band indicating analysis uncertainty. The other technical- perspective risk m e a s u r e s r e p o r t e d above explicitly account for a t least some a s p e c t s of outcome uncertainty, though often t h e analytic uncer- tainty is either n o t r e p o r t e d or is not clearly r e p o r t e d along with t h e risk measures. a s was the case with t h e SAI and FPC assessments. Sometimes Rasmussen curves a r e r e p o r t e d with e r r o r bands reflecting analytic un- certainty, a s in Hazelwood and Philipson (1977). In many c a s e s analytic uncertainty i s coped with by adopting conservative assumptions, then maintaining or implying t h a t t h e r e p o r t e d risk m e a s u r e is a t the conser- vative end of t h e analytic uncertainty e r r o r bar. The fact t h a t t h e SAI and FPC assessments differ in many of their measures, while both claim conservative assumptions, illustrates t h a t t h e r e c a n be differences of opinion regarding what is a n appropriate conservative s e t of assumpt,ions.

While t h e Lechnical-perspective risk m e a s u r e s discussed so far r e p o r t uncertainty t o varying degrees, evaluation is limited t o observing how small t h e probabilities a r e , perhaps in non-decision comparisons a s described above. In contrast, t h e decision analytic expansions t o the technical perspective discussed before (Keeney 198Ob; Bodily 1980) e m - ploy e x p e c t e d utility theory (von Neurnann and Morgenstern 194-7), which incorporates a t t i t u d e toward outcome uncertainty in t.he risk measure it- self. In this type of approach, t h e risk m e a s u r e is t h e expectation of a n index (utility) t h a t is a nonlinear function of t h e outcorne measures. On

(12)

any one dimension, then, if t h e utility function is concave downward, t h e expected utility risk m e a s u r e will discount t h e value of a n option with high outcome uncertainty relative t o another option with t h e s a m e ex- p e c t e d outcome but less uncertainty. In such a n approach, analysis un- c e r t a i n t y can be gauged by a sensitivity analysis which a d d s e r r o r b a r s to t h e risk measure.

In t h e technical approach, then, outcome uncertainty is reflect- e d in all but one of the risk measures, and a t t i t u d e toward outcome un- c e r t a i n t y is incorporated within t h e expected utility risk measures. Ana- lytic uncertainty is r e p r e s e n t e d , when it is a t all, by e r r o r b a r s or bands around any of t h e m e a s u r e s , or is dismissed by maintaining o r implying t h a t t h e r e p o r t e d m e a s u r e is a t t h e conservative e n d of t h e e r r o r bar. I t follows t h a t e a c h technical approach to coping with uncertainty is based on o n e of two assumptions. The approaches not based on e x p e c t e d utility a s s u m e t h a t once outcome and analytic uncertainties a r e m e a s u r e d and p r e s e n t e d , t h e political process will know what t o d o with t h e m . The ex- p e c t e d utility approach incorporates a t t i t u d e toward outcome uncertain- ty directly in t h e risk m e a s u r e , assuming t h a t t h e political process will want t o t r a d e off outcome uncertainty with e x p e c t e d outcomes in t h e way prescribed by von Neumann and Morgenstern (1947).

El. THE POLITICAL AF'PROACH

The political process, t o t h e e x t e n t t h a t one c a n generalize, often takes a markedly different approach to coping with uncertainty t h a n t h a t assumed by the expected utility risk analyst. Three brief exam- ples illustrate this. When t h e Office of Technology Assessment did a study on LNG in 1977, it concluded t h a t , in t h e face of t h e large analytic uncer- tainties.

"...

decisions about LNG systems shoultl be m a d e on t h e basis of nonquantitative approaches

...

" (USOTA 1977). When t h e FPC staff as- s e s s e d t h e risks of two California LNG sites (Oxnard and Point Concep- tion), it found t h e two resulted in different b u t very low risks. I t decided t h a t both sites resulted in an acceptable level of risk, t h e n dropped risk a s a factor a n d favored the more risky Oxnard site on o t h e r grounds.

When ttie California s t a t e legislature drafted t h e LNG Terminal Act of 1977, it s e t a stringent r e m o t e siting const.raint intended to preclude a worst c a s e scenario anything like the ones SES published concerning Ox- nard.

All of t,he above examples point to t h e s a m e conclusion: t h e s e political decision making processes do n o t t a k e the probabilistic perspec- tive of t h e rlsk analyst, a r e not ready to iricorporate uncertainty into risk evaluation in any way resembling the expected utility approach, and re- fuse t o explicitly t r a d e off risk t o life and limb against any o t h e r dimen- sion. I t i s misleading to r e f e r t o a political process a s risk averse in t h e expected utility sense, since t h e thinking evidenced in the processes cit- e d above has very little t o d o with t h e expected utility model. The think- ing is basically non-probabilistic. In reference t o t h e California r e m o t e siting provision arid the SES worst c a s e s for Oxnard, one legislative staff m e m b e r s t r e s s e d t h a t t h e California legislature could not allow a site t h a t could kill 40,000 people. The r e m o t e siting provision was a constraint

(13)

used t o av0i.d any explicit trading off of safety with environmental quali- ty, cost, etc. The FPC s t a f f employed similar reasoning in determining t h a t the risk was less t h a n some level, t h e n dropping it from considera- tion. In e a c h case, t h e setting of a constraint was effectively a n heuristic employed t o simplify a difficult decision. Of course, t h e level a t which t h e constraint was s e t involved implicit tradeoffs between safety, environ- m e n t , a n d cost, a s well a s judgments concerning t h e n a t u r e of s i t e s avail- able on the California coast, but those tradeoffs a n d judgments were nev- e r considered in a n explicit m a n n e r .

As with any heuristic, the r e m o t e siting constraint imposed a probable cost. Ahern, representing himself and not his agency, has main- tained t h a t i t almost certainly precluded sites t h a t would be d e e m e d more desirable by any reasonable evaluation (personal communication 1979). More generally, a desire t o preclude t h e possibility of a disaster, no m a t t e r how improbable, c a n lead to decisions t h a t impose financial and environmental costs. Ahern (1980b) describes two c a s e s of this in r e - c e n t California decisions.

IV.

THE ROLE O F ANALYSIS IN THE POLITICAL PROCESS

It is interesting to consider various modes of risk a s s e s s m e n t on their own merits. But n o ana.lysis exists in a vacuum. I t s effectiveness depends critically on t h e process of which it is a p a r t . Three critical features of t h a t process which c a n cause problems for t h e use of risk as- s e s s m e n t s a r e briefly reviewed here. We first consider t h e disaggregated n a t u r e of the political process, t h e n examine t h e two typical ways in which analyses a r e used in siting procedures: in drafting legislation o r r e - gulations, and in hearings.

A. DISAGGREGATED NATURE OF THE PROCESS

The m o s t serious problem in the use of a decision analytic siting evaluation model, such a s t h e one proposed by Keeney (1980c), is t h a t t h e decision s t r u c t u r e assumed by t h e model does not m a t c h t h e institution- al s t r u c t u r e making t h e siting decision. For example, t h e m o s t basic as- sumption made by the above model is t h a t t h e r e is some single, self- aware process somewhere making tradeoffs between conflicting objec- tives, such as between safety, cost, and environmental quality. In fact, t h e a c t u a l process may be such t h a t those tradeoffs a r e n o t rnade by any identifiable agency, b u t a r e made in the interaction between agencies.

with each agency only looking a t a p a r t of the problem.

The California LNG siting process is a good example of disaggre- gation. Three agencies were centrally involved in t h e s t a t e level decision making: t h e s t a t e legislature. t h e California Public Utilities Commission (CPUC) and t h e California Coastal Commission (CCC). The s t a t e legisla- t u r e s e t u p t h e siting process, including a single-agency licensing provi- sion intended t o e n s u r e a timely decision, and a r e m o t e siting require- ment to ensure public safety. As p a r t of t h a t process, t.he CCC r a n k e d al- ternative sites, considering primarily environmental quality subject t o

(14)

t h e rernote siting constraint. Then the C P U C s e l e c t e d a site. not neces- sarily t h e site ranked highest by t h e CCC, considering primarily cost and how quickly the site could be developed, relative t o the s t a t e ' s need for LNG. The major relevant objectives were cost, safety, environmental qual- ity, and need for LNG a s a function of time. Each agency would maintain t h a t i t considered all factors, y e t the m a n d a t e s of e a c h a r e s u c h t h a t e a c h paid special attention to a subset of those objectives. Complicating t h e picture still further, two agencies within the federal government were also p a r t of the process: one looking a t all of the major objectives, t h e o t h e r attending t o questions of national import policy. Add t o this the s t a t e and national courts, and it becomes very h a r d t o identify any single self-aware decision making process where all t h e tradeoffs were made. To be sure, t h e overall process results in some decision, which c a n be analyzed a s being consistent with particular implicit tradeoffs, b u t any similarity between tradeoffs consciously m a d e and those inferred is likely t o be simply fortuitous. I t is not clear where a decision analytic evalua- tion model would fit into s u c h a process, whose tradeoffs should be used to s e t t h e p a r a m e t e r s , or even t o whom t h e analysis should be delivered.

B. ROLE OF ANALYSIS IN FORMING SITING LEGISLATION, REGULATION

One of t h e observed roles of analysis in siting p r o c e d u r e s is in guiding the drafting of t h e relevant legislation a n d regulations. Consider, once again, t h e California LNG Terminal Act. That a c t established several interesting aspects of the siting process, b u t h e r e we will focus on the re- m o t e siting constraint mentioned above. While it could be described as a n heuristic, it is s e t in very technical language ( a t most 10 people per square mile within one mile. at most 6 0 p e r square mile within four miles), with numbers t h a t were based on analyses. A decision analyst would be inclined to m e a s u r e t h e probabilistic costs a n d benefits of each of various levels of t h e remoteness constraint, and pick t h e m o s t favor- able. But in this case, while t h e legislative staff h a d a n entire r a n g e of analyses t.o look at, from e a c h of several different studies and experts, none of t h e m assessed in any direct way t h e costs and benefits of any lev- el of constraint. Instead, each analysis indicated t h e maximum distance a flammable m e t h a n e cloud could travel from a n LNG plant. The s t a t e of t h e a r t in underst.anding t h e physical process was weak enough t h a t any of a n u m b e r of acceptable s e t s of assumptions could drive a n analysis t o any of a number of results. In fact t h e results of t h e analysis varied from 1 mile to 50 rniles for maximum cloud distances. I t was not clear which s e t of assumptions was "neutral", or which was most appropriat.e for use in drafting legislation. No analysis was used in t u r n t o aggregate this range of analyses into s u m m a r y measures to guide t h e setting of the re- mote siting constraint. The drafting of t,he legislation was a m a t t e r of ex- amining t h e range of analyses, t h e n making d i r e c t intuitive judgments a s to t h e most appropriate constraint. In this case, then, analyses were used in the drafting of legislation, but only for very low-level inputs t h a t were n o t very directly r e l a t e d to t h e costs a n d benefits involved in t h e ac- tual decision.

(15)

C. ROLE OF ANALYSIS IN HEARINGS

The second observed role of analysis in siting procedures is in presenting t h e various cases before a hearing process. There have been several hearings in t h e California LNG siting process, ranging from quasi- judicial t o legislative in n a t u r e , a t t h e local, s t a t e , and national levels. The hearings considered every aspect from energy need t o safety t o cost t o t h e relative m e r i t s of t h e various sites. Take a s one example t h e CPUC hearings to d e t e r m i n e the need for LNG in California. Several different agencies p r e s e n t e d testimony, each supported by a n analysis of a slightly different c h a r a c t e r , a s described by Ahern (l980b). But once again, a s in t h e previous subsection, the several analyses showed several different conclusions. a n d they were aggregated into the CPUC decision t o approve t h e t e r m i n a l without any traceable analytic link back to t h e testimony analyses.

D. SECTION CONCLUSIONS

The key feature in e a c h of the previous two subsections is t h a t in t h e California siting example, t h e r e was not an analysis, b u t many ana- lyses, all suspected of one bias o r another, and those analyses were ag- gregated into risk management decisions in a non-analytic way. There did not s e e m to be a single objective risk, b u t r a t h e r m o r e than one analysis indirectly assessing risk. There a r e analytic tools t h a t could be used t o combine several analyses into a single source of advice for t h e decision making process. However, those tools, based on Bayesian statistical inference, would require subjective judgments t h a t would explicitly s e t t h e weight given t o each analysis, o r t o e a c h s e t of assumptions adopted.

While this would be a promising avenue t o pursue, it may not b e politically feasible t o collect those judgments. Beyond t h a t , it m a y not b e s e e n as appropriate t o extend analysis t o t h a t level, a s it would involve political considerations t h a t a r e s e e n as properly t r e a t e d only in legislative and hearing processes.

V. CONCLUSIONS: FUTURE KESEARCH DIRECTIONS

The previous sections of this p a p e r have p r e s e n t e d basic prob- lems with t h e interface between risk assessments and the political siting process. Those problems involve differences in how risk is defined and evaluated, how u n c e r t a i n t . ~ is coped with, how analyses a r e used in t h e political process, and the nature of t h e political process itself. While t h e discussions of those problerns a r e interesting in a n d of themselves, they also provide insight into the most promising directions for r e s e a r c h t o improve t h e usefulness of risk a s s e s s m e n t in t h e political risk manage- m e n t process. To begin with, t h e r e a r e of course two sides t o t h e inter- face between risk assessments a n d t h e political risk nianagement pro- cess. However, t h e problematic features on t h e political side of t h e inter- face s e e m e i t h e r extremely difficult to change or n o t desirable t o change.

I t s e e m s most. appropriate, then, to address the problematic features on t h e assessment side of t h e interface. A review of t h e previous sections in-

(16)

d i c a t e s t h e m o s t promising a r e a s for improvement:

Extend t h e s c o p e of risk evaluation m e a s u r e s t o a c c o u n t a s m u c h a s possible for societal concerns.

Section 11 described basic differences in t h e t e c h n i c a l and societal perspectives on risk, a n d t h e implications of t h o s e differences. Risk a s s e s s m e n t s typically a d o p t a technical perspective, a n d s o provide very straightforward, single-number evaluations of risk t h a t d o n o t in- c o r p o r a t e societal c o n c e r n s very well. Political p r o c e s s e s a r e sensi- tive t o t h e s o c i e t a l perspective, b u t r e s e a r c h describing t h a t p e r s p e c - tive is n o t o r i e n t e d toward evaluations of risk t h a t provide c l e a r gui- d a n c e for risk m a n a g e m e n t decisions. F u t u r e r e s e a r c h should s e e k t o close t h a t g a p by developing hybrid risk evaluation a i d s t h a t a r e sensitive t o societal c o n c e r n s , y e t provide clear risk m a n a g e m e n t de- cision aids. Work h a s begun in t h a t direction (Bodily 1980; Keeney 1980 a , b ) , b u t should b e extended.

Develop risk m a n a g e m e n t decision aids t h a t a r e compatible with t h e essentially non-probabilistic orientation of t h e political decision m a k e r.

Section I11 d e s c r i b e d basic differences in the technical a n d political a p p r o a c h e s t o risk m a n a g e m e n t . The m o s t significant difference is t h e essentially non-probabilistic orientation of m a n y political deci- sions. That is very p r o b l e m a t i c for most risk a s s e s s m e n t s , which t e n d t o b e probabilistic in n a t u r e . Yet if risk a s s e s s m e n t s a r e t o b e useful, ways m u s t be found for t h e i r r e s u l t s t o b e e x p r e s s e d in a language t h a t t h e political decision m a k e r c a n u n d e r s t a n d .

Examine analytic techniques for aggregating t h e r e s u l t s of several different analyses, a n d s e l e c t and a d a p t t h e m o s t promising ones f o r use in t h e political process.

Section IV identified t h e n e e d for developing analytic to3ls with which t o a g g r e g a t e r e s u l t s of analyses, then went on t o e n u m e r a t e t h e prob- l e m s involved in implementing those tools. Yet n o t enough is known at t h i s time t o a s s e s s t h e feasibility of those t o o l s . That feasibility m u s t be t e s t e d by applying t h e m o s t promising aggregation tech- niques i.n a c t u a l c a s e studies.

Certairily, none of t h e r e s e a r c t i t a s k s d e s c r i b e d h e r e is easy.

However, t h e potential usefulness of i m p r o v e m e n t s in any of t h e a r e a s would justify t h e r e s e a r c h effort many t i m e s over.

(17)
(18)

REFERENCES

Ahern, W. 1980a. "California Meets t h e LNG Terminal", C o a s t a l Z o n e M a n a g e m e n t J o u r n a l 7, p. 185-221.

Ahern, W. 1980b. "The Role of Technical Analyses in California Energy Sit- ing Decisions". P a p e r p r e p a r e d for t h e LEG Case S t u d y Task Force Meeting, IIASA. Laxenburg, Austria, S e p t e m b e r .

Bodily, S.E. 1980. "Analysis of Risk t o Life a n d Limb", O p e r a t i o n s R e s e a r c h 28(1), Jan.-Feb.

Deutsch, R.W. 1980. "Siting a n LNG Facility in California: The Regulatory Framework a n d t h e F a c t o r s Involved in t h e Decision Making Pro- cess". P a p e r p r e p a r e d for t h e LEG Task force Meeting, IIASA, Laxenburg, Austria. S e p t e m b e r .

Fischhoff, El., P. Slovic, S. Lichtenstein, e t al. 1978. "How Safe is Safe Enough? A Psychometric Study", P o l i c y S c i e n c e s 8.

FPC. 1976. Pacific Indonesia Project, Final Environmental I m p a c t S t a t e - m e n t , F e d e r a l Power Commission. December.

Hazelwood, R.N. a n d L.L. Philipson. 1977. Survey of LNG Risk Assessment, Report f o r t h e California Public Utilities Commission,Socio- Economic S y s t e m s , Inc., Los Angeles.

Keeney, R. L. 1980a. "Evaluating Alternatives Involving Potential Fatali- ties", O p e r a t i o n s R e s e a r c h , 28 ( I ) , Jan.-Feb.

(19)

Keeney, R. L. 1980b. "Equity and Public Risk", O p e r a t i o n s R e s e a r c h 28(3), May-June.

Keeney, R. L. 1980c. S i t i n g E n e r g y F a c i l i t i e s , New York: Academic Press (in press).

Keeney, R.

L

and H. Raiffa. 1976. Decisions w i t h M u l t i p l e Objectives, New York: Wiley and Sons.

Kunreuther, H. 1980. "Societal Decision Making for Low Probability Events: Descriptive a n d Prescriptive Aspects", (forthcoming IIA- SA working paper).

Lichtenstein, S., P. Slovic, B. Fischhoff e t al. 1978. "Judged Frequency of Lethal Events", J o u r n a l of E z p e r i m e n t a l P s y c h o l o g y : H u m a n L e a r n i n g a n d M e m o r y , 4 (6).

Linnerooth. J. 1978. "Reevaluating the Value of Life: Practical Considera- tions", invited paper presented a t ORSA/TIMS Conference, Los Angele s, November.

Otway, H. J. and P.D. Pahner. 1976. "Risk Assessment". F u t u r e s , 8 (2).

Pat$, M.-E. 1978. "Public Policy in Earthquake Effects Mitigation", Techn- ical Report 30, Blume Earthquake Engineering Center, Stanford.

SAI. 1975. LNG Terminal Risk Assessment Study for Oxnard, California.

Prepared for Western LNG Terminal Company, SAI-75-6 15-LJ. La Jolla. California: Science Applications, Inc.

SES 1976. Environmental Impact Report for the Proposed Oxnard LNG Facilities, Socio-Economic Systems, Inc., Los Angeles, California.

September.

Slovic, P. and B. Fischhoff. 1979. "How Safe is Safe Enough? Deter- minants of Perceived and Acceptable Risk", in L. Gould and C. A.

Walker (Eds.). Too Hot to H a n d l e : Social a n d P o l i c y I s s u e s in t h e M a n a g e m e n t of R a d i o a c t i v e Wastes, New Haven: Yale University Press.

Slovic, P., B. Fischhoff and S. Lichtenst,ein. 1980. "Perc:eived Risk", in R.

C. Schwing and W. A. Albers (Eds.), S o c i e t a l R i s k A s s e s s m e n t : How S a f e is S a f e Enough? New York: Plenum Press.

Starr. C. 1969. "Social Benefit Versus Technological Risk", S c i e n c e , 165:

1232- 1238.

USNRC. 1975. R e a c t o ~ S a f e t y S t u d y ,

as as muss en

~ e p o r t ; " ) , WASH-1400, Washington. D.C.: U.S. Nuclear Regulatory Commission.

(20)

USOTA. 1977. Transportation of Liquefied Natural Gas, Washington, D.C.:

U.S. Office of Technology Assessment.

von Neurnann, J. and 0. Morgenstern. 1947. Theory of Games and Economic Behavior, Princeton: Princeton University Press.

Referenzen

ÄHNLICHE DOKUMENTE

Priska Müller ist ein Beispiel für viele, die nach der Familienzeit eine befriedigende Aufgabe oder einen Einstieg in die Pflege suchen.. Ein Quereinstieg in eine

Porcelain stoneware is an ultra-compact technical material, extremely resistant to wear, tread, shock, scratching and heavy loads. Per maggiori informazioni consultare il

Operations to pursue proper deduction and evaluation of mass movement disposition as well as the identification of endangered areas with physical based models need comprehensive area

Based on the detailed evaluation of the specific processes and the resulting natural haz- ards for the investigated section a comprehensive assessment of the natural hazard potential

Fluconazol ist möglich bei Patienten, die nicht kritisch krank sind, keine Azol- Vorbehandlung erhalten haben oder bei denen empfindliche Erreger (C. albicans) nachgewiesen

technisches System oder Verfahren zu verwenden ist (S. 123 des Planfeststellungsbeschlusses). Als Systeme, welche die in der Nebenbestimmung spezifizierten Anforderungen gewährleisten,

Stadtvertreter Ehlers antwortet, dass er diese Fragen im Tagesordnungspunkt 6.1 aufgreifen werde.. 4 1.3 Frau Meike Störtenbecker, Ortsteil Gollendorf, nimmt Bezug

The broadest biographical details of Elizabeth’s life can be traced through the Journals and Correspondence of Lady Eastlake, compiled and edited by her