Decision Making for Low Probability Events: A Conceptual Framework

(1)

Working Paper

Howard Kunreuther

November 1980 WP-80-169

DECISION MAKING FOR LOW PROBABILITY EVENTS: A CONCEPTUAL FRAMEWORK

International Institute for Applied Systems Analysis

A-2361 Laxenburg, Austria

(2)

NOT FOR QUOTATION WITHOUT PERMISSION OF THE AUTHOR

DECISION MAKING FOR LOW PROBABILITY EVENTS

:

A CONCEPTUAL FRAMEWORK

Howard Kunreuther November

1

980 WP-80-169

Working Papers

are interim reports on work of the International Institute for Applied Systems Analysis and have received only limited review. Views or opinions expressed herein do not necessarily repre- sent those of the Institute or of its National Member Organizations.

INTERNATIONAL INSTITUTE FOR APPLIED SYSTEMS ANALYSIS

A-2361 Laxenburg, Austria

(3)

ABSTRACT

Recent empirical evidence from field surveys and controlled laboratory experiments reveal anomalies with r e s p e c t t o decisions by individuals to p r o t e c t themselves against low probability, high loss events. In particular, behavior is frequently a t odds with what would be p r e d i c t e d by s t a n d a r d models of choice which involve benefit-cost comparisons.

This p a p e r develops a framework for analyzing decisions for low probability events and discusses their policy implications. The framework highlights t h e following four i n t e r r e l a t e d components:

(1) Type of information collected by individuals in making their decisions (i.e., a c c u r a c y of d a t a on losses, probabilities and protective options);

(2) The decision process of individuals (e.g., expected utility maximization, threshold models);

(3) Implications of policies on specific groups (e.g., affected individuals.

general taxpayers); and

(4) Welfare implications (e.g., equity and efficiency considerations).

Examples from studies on n a t u r a l hazards, health and szfety problems will be used t o illustrate how this framework synthesizes descriptive models of choice with policy prescription. The p a p e r concludes by suggesting directions for future r e s e a r c h .

(4)

TABLE OF CONTENTS

INTRODUCTION

11. DEVELOPING A GENERAL FRAMEWORK Ill. ILLUSTRATNE EXAMPLES

Safer Power Mowers Motor Vehicle Safety Cigarette Smoking Siting of LNG Facilities

IV. INSURAYCE AGAINST NATURAL HAZARDS Problem Formulation

Interested Parties Decision Processes

Formulating Alternative Policies Evaluating Strategies

V.

TOWARDS A DESCRIPTIVE MODEL OF CfiOlCE Performance of a Market System

Role of Government

NOTES REFERENCES

(5)

DECISION MAKING FOR

LOW

PROBABILITY EVENTS: A CONCEPTUAL FRAMEWORK

Howard Kunreuther

I. INTRODUCTION*

\

Society has become increasingly concerned with developing appropriate measures for mitigating t h e consequences of low probability events which have potentially large losses. It should be recognized a t t h e outset t h a t what is a low probabihty event for one interested party may be viewed a s a high probability event for another. Similarly, the relative magnitude of t h e losses is also a function of where one sits. For example, the chances of suffering a severe property loss from a natural disaster or a severe injury from a n automobile accident may be viewed as very small by a single individual but t r e a t e d a s relatively high by a government agency concerned with national losses. Property damage from a fire may appear staggering t o t h e affected family but s e e m relatively small a t a more aggregate level because of t h e different bases used to evaluate consequences.

This paper proposes a conceptual framework for dealing with events which a r e perceived to have a small chance of occurrent? by a t least one of t h e interested parties. The approach emphasizes the importance of undertaking descriptive analysis a s a critical input for prescriptive recommendations.

After outlining the framework (Section II), I will illustrate its applicability in Section 111 with several examples which have both personal significance (e.g., safety of power mowers and motor vehicles) as well a s societal importance (e.g., siting of LNG facilities). The importance of understanding decision processes a s .Iwould like to thank Uday Apte for helping to gather material on the illustrative examples

discussed in section II of the paper.

(6)

a critical input t o policy is underscored by empirical data on individual decision making with respect t o insurance protection against natural hazards. Section

N

summarizes key results from a large-scale field survey and controlled laboratory experiments which comprised this four-year study and illustrates t h e possible roles t h a t the public and private sectors can play in providing better protection against future losses. In t h e concluding section, a more formal model is proposed which incorporates the decision processes and the role of information as critical inputs for developing prescriptive measures.

F u u r e 1 depicts a conceptual framework for structuring the analysis. An appropriate starting point is Problem Formulation (Box 1). Before undertaking a detailed analysis one needs t o identify and define the problem. What a r e the goals a n d objectives corresponding to the particular a r e a of concern? Can one gain insight into the nature of the problem through an historical perspective?

This initial definitional phase is critically important as it enables one t o under- take a detailed descriptive or behavioral analysis which can then be linked t o alternative strategies. Furthermore, it helps limit the types f policies or plans that a r e relevant and provides guidelines for evaluating them.

P

Figure 1. Conceptual framework for analysis s t r u c t u r ~ .

Let us t u r n now t o the descriptive phase. We need to define and describe explicitly the i n t e r e s t e d p a r t i e s (Box 2 ) impacted by the problem. Three sectors are considered in the illustrative examples which follow: consumers (i.e., those who demand the particular products or a r e directly or indirectly &ected by it);

firms or enterprises (i.e.. the organizations or business or supply the product);

and government (i.e.. public sector agencies or bureaus who interact with the private sector-consumers and enterprises). For each problem there a r e a set of legal and political constraints which determine how information currently flows between the three sectors and the groups within each sector. It is important t o understand t h e dynamics of this interaction: who interacts with whom, and w h e n this interaction takes place.

The other key element of the descriptive phase is the decision processes (Box 3 ) of each of the involved interested parties. By decision processes we

(7)

mean the collection and processing of information relevant to the problem being analyzed. Recent empirical evidence from field and experimental studies have revealed systematic biases with respect to the processing of information and simplified rules in combining data in making decisions (Fischhoff, e t al. 1980;

Kunreuther. e t al. 1978; and Tversky and Kahneman 1974). These findings shed considerable light on the relative importance of external events. such as past experience, as well a s internal dynamics, such as discussions with others. in influencing decisions on low probability events. It is thus clear that the collection and processing of information a r e likely to be closely tied t o the relevant constraints and t h e interactions between t h e interested parties.

Turning now to t h e prescriptive phase, there is a need t o formulate a l t m a - tive p l a n s or courses of actions (Box 4) for coping with a particular problem. The generation of goals and objectives for any problem w i l l suggest a s e t of plans t o be considered. Two types of institutional arrangements between the interested parties circumscribe t h e types of plans which can be considered. One extreme is for consumers and firms to interact through a market system without any government involvement. The other extreme is for government to impose strict regulations which gives t h e private sector no freedom of choice. Most strategies a r e between these two extremes: the government sector utilizes incentives such as subsidies and taxes along with some regulations and information exchange to guide consumer and firm market interactions.

Finally, there is a need to evaluate p l a n s (Box 5). How well different policies perform will be influenced by the decision processes of the interested parties. The ranking of M e r e n t policies is also contingent upon the relative importance given to the interested parties. If residents in hazard-prone areas are deemed important enough to merit special treatment after a disaster, then Strategy A may be much more appealing than Strategy B. On the other hand, if disaster victims a r e deemed responsible for their o m recovery, then Strategy B may be seen as preferable to Strategy A. In evaluating different measures one has to include the compliance costs which must be p

9

by the sellers as well a s t h e enforcement costs which utilize government funds.

111. ILLUSTRATIVE EXAMPLES

The framework depicted in Figure 1 can be applied to a set of problems which involve protective measures to reduce the probability of an event or miti- gate its consequences. The section begins with those which involve personal safety and conclude with broader societal issues. The purpose of these illustra- tions is to show how the framework can structure analysis; no detailed evaluation of alternatives is undertaken.

Safer Power Mowers

Should power mowers be made safer? Each year approximately 75,000 people come in contact with moving power mower blades which can cause severe injuries. Nearly 10,000 of the blade-contact injuries involve amputations of fingers or toes (Washington Post 1979a). The problem involves a tradeoff between the costs of producing a safer mower and the reduction in injuries which presumably would result. In this case, the relevant interested parties are

(8)

t h e homeowner or g a r d e n e r who has or desires a power mower; the lawn mower industry; a n d the Consumer Products Safety Commission (CPSC), the regulatory agency with the responsibility for approving safety s t a n d a r d s in this area.

The decision processes of consumers plays a critical role in evaluating any policy. If individuals a r e careless because they feel t h a t nothir,g c a n happen to them when they utilize a mower, then it may be necessary t o make power mowers safer. In addition, or a s a n alternative, warnings could be provided on t h e dangers of the mower (e.g., not to use it on wet grass). How well this information is actually processed by individuals determines how well such a policy works.

With r e s p e c t t o a l t e r n a t i v e plans, y e CPCS has proposed mandatory safety standards in designing power mowers. Industry claims t h a t this regulation.

which would increase t h e cost of a power mower by $35, is too strong. In e v a l u a t i n g t h e s e plans questions of product liability arise: Is t h e manufacturer responsible should t h e r e be an injury from a mower? A r e c e n t case awarded

$6000 t o a man who lost p a r t of his foot in a lawn mower. The company claimed t h a t t h e accident. which occurred because the person's foot slipped on wet grass. could have been avoided had he read the user manual which warned : "Do not use this mower on Wet Grass." In this case ignorance was considered an excuse and t h e claim was upheld (Business Week, February 12, 1979).

Motor Vehicle S a f e t y

What a r e t h e appropriate safety measures for reducing deaths and injuries from motor vehicles? This question has some significance when one studies the statistics for the United States: "In 1977, motor vehicles caused 47,700 deaths, 1,900,000 disabling injuries, and approximately $12 billion in property damage"

(Bick and Hohenemser 1979). A t present less t h a n 20% of t h e drivers or passengers in private vehicles protect themselves by wearing s e a t belts even though they a r e installed in all cars. Here, t h e p r o b l e m involves the tradeofis between personal freedom a n d possible adverse consequences to individuals and society when people do not voluntarily protect themselves. The relevant i n t e r e s t e d p a r t i e s a r e t h e drivers and passengers, t h e automobile industry. and t h e National Highway Traffic Safety Administration, the regulatory agency empowered t o deal with motor vehicle safety.

Echoing t h e same t h e m e a s above, the decision processes of consumers are critically important for designing prescriptive measures. Empirical evidence from laboratory studies suggests t h a t one reason people do not voluntarily take protective action s u c h as wearing safety belts, is because they feel t h a t the probability of a n accident is so small t h a t they don't have t o worry about it (Slovic, e t al. 1978). A survey conducted by National Analysts (1971) for t h e Department of Transportation revealed t h a t those most likely t o wear belts a r e ones who have been asked by o t h e r s t o wear them. This raises the question of the irnpor- t a n c e of personal influence in the decision making process.

A t a policy level t h e r e a r e several options which can be considered. Market mechanisms s u c h as lower insurance premiums for c a r s equipped with passive restraints (e.g., automatic belts or air bags) could encourage people t o voluntarily adopt these measures. Some countries do not pay insurance claims for injuries if it is shown t h a t the individual has not protected himself with a safety belt, thereby providing economic incentives for individuals t o use them. A stronger measure, utilized in some countries is t o impose a fine for those not

(9)

wearing t h e belt. An e x t r e m e m e a s u r e would b e t o r e q u i r e t h a t all a u t o s be equipped with a passive r e s t r a i n t . Each of t h e s e m e a s u r e s h a s t o be evaluated o n a number of dimensions, t h e m o s t i m p o r t a n t being t h e c o s t s of imposing t h e particular a p p r o a c h a n d t h e potential benefits. As in all t h e e x a m p l e s in t h i s section s o m e p a r t i e s will b e helped while o t h e r s will suffer depending on which alternative is chosen.

C i g a r e t t e S m o k i n g

Should o n e impose r e s t r i c t i o n s on c i g a r e t t e s t o d e t e r individuals from smoking and if s o how should this b e done? This question is s t i m u l a t e d by empirical d a t a which s u g g e s t s t h a t annually 350,000 lives a r e lost a n d approxi- m a t e l y $18 billion in hospital bills a r e i n c u r r e d f r o m diseases c a u s e d by smoking (Washington P o s t 1979b). The relevant a n t e ~ e s t e d p a r t i e s a r e s m o k e r s , non- s m o k e r s , t h e t o b a c c o i n d u s t r y a n d t h e Office of Smoking and Health, a regula- t o r y agency c o n c e r n e d with t h e effects of c i g a r e t t e smoking.

The d e c i s i o n p r o c e s s e s of s m o k e r s a r e critical t o t h e design of alternative policies. If individuals a r e aware t h a t smoking is h a r m f u l t o t h e m b u t ignore t h e s e potential effects, e i t h e r b e c a u s e t h e y feel "nothing will h a p p e n t o m e , "

t h e n additional information c a m p a i g n s a r e unlikely t o c h a n g e behavior. There is also t h e question a s t o how sensitive t h e s m o k e r is t o p r i c e c h a n g e s in c i g a r e t t e s should additional t a x e s be imposed.

The s p e c t r u m of a l t m a t i v e p l a n s r a n g e from m a r k e t solutions (do nothing a n d let people suffer t h e consequences) t o s t r i c t regulation (banning cigarettes). R e c e n t proposals have involved a s e t of incentive s y s t e m s , s u c h a s increasing t a x e s a n d using t h e revenue t o help s m o k e r s quit (Harris 1980), o r prohibiting smoking i n c e r t a i n public places (e.g.. hospitals, t h e a t e r s , and retail s t o r e s ) (Washington P o s t 1979b). In evaluating t h e s e plans one recognizes t h a t different i m p o r t a n c e weights o n t h e relevant i n t e r e s t e d p a r t i e s m a y l e a d t o different rankings. For e x a m p l e , a policy of "do nothing" favors t h e s m o k e r s a n d t h e tobacco i n d u s t r y while banning cigarette: h a s t h e r e v e r s e effect. Taxation policies a n d fines for smoking in c e r t a i n places falls somewhere between t h e above two e x t r e m e s .

S i t i n g ^3fLhrG F a c i l i f i e s

Liquefied n a t u r a l g a s

(LNG)

is a potential s o u r c e of energy which requires a fairly complicated technological p r o c e s s t h a t has t h e potential, albeit with very low probability, of c r e a t i n g s e v e r e losses. To i m p o r t LNG t h e g a s h a s to be con- verted t o liquid form a t a b o u t 1/600 t h e volume. I t is shipped in specially con- s t r u c t e d t a n k e r s a n d received a t a t e r m i n a l where it u n d e r g o e s regasification a n d is t h e n d i s t r i b u t e d . The e n t i r e s y s t e m (i.e., t h e liquefication facility, t h e

LNG

t a n k e r and the receiving t e r m i n a l a n d regasification facility) c a n cost m o r e t h a n $1 billion t o c o n s t r u c t (Office of Technology Assessment 1977). The siting problem of i n t e r e s t is w h e t h e r one should locate facilities for regasifying and shipping LNG a n d if so where would be t h e b e s t place. The i n t e r e s t e d p a r t i e s a r e t h e residents of a r e a s c o n s i d e r e d as potential s i t e s , t h o s e benefiting from t h i s additional s o u r c e of e n e r g y , t h e g a s companies or c o n s o r t i u m who a r e willing t o invest in a proposed p r o j e c t a n d g o v e r n m e n t agencies a t t h e Federal. s t a t e , and

(10)

local level who have responsibility for trading off the costs (including potential losses from an accident) and benefits of any decision.

Turning now to the decision p r o c e s s associated with siting, t h e r e are questions as to how each of t h e groups utilize information on the probability of any accident to a n LNG terminal and the resulting consequences. One of the contro- versies emerging in the siting debate is whether one c a n or should specify an acceptable level of risk. Some risk assessments of a particular site focus on the chances of a catastrophic accident and conclude t h a t it is acceptable if the probability is below some critical level. Others have utilized worst case scenarios and paid attention to the consequences without paying much attention t o the chances of its o c ~ u r r e n c e . ~ There is also a need t o understand how t h e different interested parties weigh t h e safety issue in relation t o other concerns of a siting policy such a s t h e economic impacts, effects on the environment and how LNG serves national energy policy.

The formulation of alternative strategies will be greatly impacted by t h e decision process of t h e different parties. One way of clarifying differences between the groups is t o specify who is responsible for damages should an accident occur. If the location of a n LNG facility is viewed primarily a s a private venture, then some form of insurance should be offered t o gas companies t o pro- t e c t them against catastrophic losses. If this type of coverage is not available on the private market. then government may have to provide this protection. A complementary s e t of plans may involve compensating residents of a proposed siting area for decreases in their real estate value a n d perhaps provide them with lower energy r a t e s in r e t u r n for their increased risk in the future. An alternative is to pass regulations such as the one by the Department of Energy which requires t h a t new sites be in remote areas or in locations with relatively small population densities.

IV.

I N S U W CE AGAINST NATURAL H A Z A R D S ~

Let us now t u r n to a more detailed st^-dy of homeowner decisions on whether to protect themselves against the consequences of natural hazards.

The results raise a s e t of policy-related issues. They also shed light on the decision processes individuals a r e likely to use when dealing with situations such as those discussed in the previous section.

P r o b l e m F o r n v l a t i o n

The problem of interest is the appropriate role of the public and private sectors in providing insurance protection against t h e consequences of natural hazards and relief in the aftermath of a disaster. An historical perspective with respect t o this problem is relevant here. Annual losses from natural disasters in the United States is frequently over $1 billion dollars. Relatively few homeowners have voluntarily purchased insurance against t h e consequences of floods and earthquakes, even though coverage is easily available and in the case of floods highly subsidized by the federal government. In the past, the

U.S

government has responded to the financial plight of the uninsured victims by providing liberal relief in the form of low interest loans and grants to aid the recovery efforts.

(11)

Evidence on increased federal disaster relief is provided by comparative data on the Small Business Administration (SBA) disaster loan program. The growth of the program is easily s e e n in Figure 2; the increase is particularly significant in the case of home loans where both t h e total number and total dol- l a r values in t h e 1966-76 period were more than 25 times what they were in t h e first 12 years of t h e program. It is striking t h a t t h e $1.2 billion approved by t h e SBA for victims of Tropical Storm Agnes represented almost four times t h e entire amount allocated by the SBA for all disasters between fiscal years 1954 and 1965. Over $540 million of the amount approved by t h e SBA for victims of this disaster were in t h e form of forgiveness g r a n t s which did not have t o be repaid.

~ 1 0 ' 1

VALUE 3.411.1

ISM

n

Kw: 19s-6Y 0 !W&IS

Figure 2. Small business administration disaster loans.

Interested Parties

Insurance against floods is provided by the Federal Insurance Administra- tion with hqhly subsidized r a t e s on existing property; new property is charged a premium based on estimated risk. For insurance t o be offered t o residents and businesses in a hazard-prone region, the community must agree to adopt land use regulations and building codes t o reduce future losses from t h e hazard.

Earthquake coverage is offered t o t h e public by private companies. Even though coverage is not expensive ($2 p e r $1000 coverage on wood-frame homes in Cali- fornia with a 5% deductible), less than 3% of the homeowners in this earthquake- prone s t a t e have bought this insurance.

The interested parties for this problem a r e thus the Federal Insurance Administration ( a government agency), the private insurance industry (i.e..

companies and agents), the Small Business Administration, t h e residents in hazard-prone areas, a n d the general taxpayer who covers the subsidized portion of flood coverage and t h e subsidized portion of disaster relief.

(12)

Decision Processes

What a r e t h e factors which influence individuals to purchase insurance protection against relatively low probability events such as floods and earthquakes?

To answer this question field survey questionnaires and controlled laboratory experiments were undertaken. The field survey involved face-to-face interviews with 2,055 homeowners residing in 43 areas throughout the United S t a t e s sub- ject t o coastal and riverine flooding, and 1,006 homeowners Living in 18 earthquake-prone areas of California. Half t h e respondents had purchased flood or earthquake insurance, t h e other half had not. The controlled laboratory experiments undertaken by Paul Slovic. Baruch Fischhoff, and Sarah Lichten- stein, a t Decision Research, shed Light on the causal relationships between variables entering into t h e insurance decision. A few of t h e key findings from this study which r e l a t e t o individual decision processes a r e now summarized.

Although most uninsured homeowners interviewed were aware t h a t flood and earthquake coverage existed, the majority were unaware t h a t they were eli- gible t o purchase a policy. Those who were aware had no reliable knowledge of t h e costs of a policy. The subsidized flood r a t e is between $2.50 and $3.50 p e r

$1000 coverage depending on the proportion of coverage devoted t o s t r u c t u r e and contents. The earthquake premium on wood-frame homes in California averages $2 p e r $1000. Hence any homeowner estimating the respective r a t e s between $2 and $4 for flood coverage and $1 and $3 for earthquake insurance was classified a s reasonably accurate. Figure 3 shows t h a t most of the insured homeowners were a c c u r a t e in their estimate or underestimated the premium.

Few uninsured individuals had accurate information and a large proportion overestimated t h e premium. This finding suggests t h a t t h e uninsured individuals had not made any conscious effort to obtain information on r a t e s from their insurance agent even if they knew coverage was available.

Figure 3. Subjective estimates of insurance premium.

(per $1,000 coverage)

(13)

With respect to the hazard itself, both insured and unir~sured individuals had imperfect information on the probability and consequences of a severe flood or earthquake causing damage t o their property and contents. When homeown- e r s were asked to estimate the chance of a severe flood o r earthquake damaging their property in t h e next year, 15 percent of the respondents in flood areas and 8 p e r c e n t of those in earthquake areas, were unable t o provide any sort of estimate. Some people thought the probability of a disaster hitting t h e m was quite high-1 chance in 10-yet they said they had purchased no disaster insurance.

Others believed the chances of a disaster affecting t h e m was almost nil-1 in 100,000-yet they had purchased disaster insurance. It seems clear t h a t a number of individuals participating in t h e field survey do not understand t h e concept of probability. The findings a r e consistent with t h e heuristics and biases implied by controlled laboratory experiments over t h e past decade (Fischhoff, e t al. 1979; Tversky and Kahneman 1974).

It is tempting t o a t t r i b u t e this casual attitude about the risks of natural hazards and protective activities to homeowners' beliefs t h a t the federal govern- m e n t will bail them out in a crisis. But Figure 4 indicates t h a t t h e majority of uninsured residents anticipate no aid a t all from the government even when they expected t o suffer large losses from a disaster. Most of these people were aware t h a t t h e SEA provides aid to the victims, but they had little knowledge of t h e loan t e r m s or whether they could receive forgiveness grants. On the basis of these results, one can conclude that most homeowners in hazard-prone a r e a s have not even considered how they would recover should they suffer flood or earthquake damage. Instead they t r e a t such events as being s o unlikely t h a t they ignore t h e consequences altogether.

Figure 4. Impact of federal aid on uninsured homeowners.

What variables influence a person's decision to purchase insurance? A key factor is a belief that the hazard is a serious problem. This concern is found primarily among people who have had past experience with the hazard. "You ask m e why I didn't have insurance before the June 1972 flood" said one homeowner in Norristown, Pennsylvania. "We had the flood in September of '71 and I had two

(14)

f e e t of water in m y b a s e m e n t . And I felt this I c a n tolerate a n d this is probably a s high a s it will ever get." To his chagrin, this man suffered severe property damage in 1972. Only t h e n did he decide t h a t he needed insurance. Another uninsured flood victim, said t h a t his rationale was t h a t "the $60 in premiums t h e y could use for something else. But now they don't c a r e if t h e figure was

$600. They're going t o t a k e insurance because they've been through it twice a n d they've learned a lesson from it."

As shown in Figure 5, another important factor in influencing t h e purchase of a policy a p p e a r s t o b e knowing someone who has purchased coverage o r dis- cussing insurance with a friend, neighbor o r relative. The following example graphically illustrates this point. In a p r e t e s t of t h e questionnaire in San Fran- cisco, a homeowner r e s p o n d e d t o o n e of t h e questions by saying t h a t h e did not have e a r t h q u a k e insurance. A friend of his who was listening t o t h e interview c o m m e n t e d t h a t he had himself p u r c h a s e d s u c h insurance a few y e a r s before.

The r e s p o n d e n t was dumbfounded and asked his friend about t h e availability of e a r t h q u a k e coverage a n d how m u c h it cost. "I'm going t o have t o look into e a r t h q u a k e insurance myself," he added.

Figure 5. Interpersonal communication between insured and uninsured.

The controlled laboratory experiments on insura c e undertaken a t Decision

8

Research provides f u r t h e r insight into t h e s e results. Subjects were exposed t o a variety of risks t h a t had different losses and probabilities associated with t h e m . By keeping the p r e m i u m c o n s t a n t for all risks and varying t h e losses and probabilities in such a way t h a t t h e expected loss (loss multiplied by probability) was t h e s a m e , i t was possible t o t e s t t h e i m p o r t a n c e of probability a n d loss on i n s u r a n c e purchase decisions.

One would e x p e c t t h a t individuals should prefer to insure themselves against events having a low probability of occurrence but a high loss r a t h e r t h a n against those having a high probability and low loss. The reverse was found t o b e t r u e for a variety of experimental formats. These results suggest t h a t if t h e c h a n c e s of a n event a r e s ~ ~ c i e n t l y low, people do not even reflect o n i t s consequences. In o t h e r words, people a r e primarily i n t e r e s t e d in buying insurance if

(15)

t h e y feel t h e probability of a d i s a s t e r is high enough for t h e m t o s t a n d a good c h a n c e of g e t t i n g a r e t u r n . They t h u s view i n s u r a n c e a s a n .rnvestment r a t h e r t h a n a!: p r o t e c t i o n .

F o r m u l a t i n g A l t e r n c r t i v e PoIicies

There a r e a s e t of allernative policies for dealing with t h e n a t u r a l h a z a r d s problem outlined above. The c u r r e n t institutional a r r a n g e m e n t s for floods and e a r t h q u a k e s i l l u s t r a y s t h e role of incentives and regulations t o s u p p l e m e n t m a r k e t m e c h a n i s m s .

In t h e c a s e of t h e flood h a z a r d , t h e f e d e r a l g o v e r n m e n t offers subsidized p r e m i u m s as a n incentive for r e s i d e n t s t o p u r c h a s e coverage. They also a r e imposing specific land use regulations on c o m m u n i t i e s who p a r t i c i p a t e in t h e flood p r o g r a m . More recently b a n k s have r e q u i r e d homeowners t o p u r c h a s e flood i n s u r a n c e a s a condition for obtaining a mortgage. Those who apply f o r f e d e r a l relief a f t e r a d i s a s t e r a r e also r e q u i r e d t o p u r c h a s e coverage a s a condition for obtaining a low i n t e r e s t loan. F o r t h e s e g r o u p s , flood i n s u r a n c e is m a n - d a t o r y r a t h e r t h a n voluntary.

P r o t e c t i o n against e a r t h q u a k e d a m a g e has b e e n m o r e of a private r a t h e r t h a n public afTair. No o n e is r e q u i r e d t o p u r c h a s e i n s u r a n c e a s a condition for a m o r t g a g e o r a d i s a s t e r loan. Even t h o u g h coverage is available, t h e r e h a s b e e n n o g r e a t e f f o r t m a d e by i n s u r a n c e c o m p a n i e s o r t h e i r agents t o actively m a r k e t policies. The i n s u r a n c e industry c l a i m s t h a t i t d o e s ' n o t have enough rein- s u r a n c e c a p a c i t y t o cover t h e d a m a g e f r o m a c a t a s t r o p h i c q u a k e in a populated a r e a of California if m o s t r e s i d e n t s a n d businesses were p r o t e c t e d with i n s u r a n c e . Today t h e principal g o v e r n m e n t role with r e s p e c t t o t h e h a z a r d is t h r o u g h local building codes a n d o r d n a n c e s on t h e design of s t r u c t u r e s a n d t h e provisions of f e d e r a l aid t o cover t h e u n i n s u r e d portion of a n e a r t h q u a k e loss.

O t h e r p r o g r a m s for coping with t h e p r o b l e m a r e s t i m u l a t e d by t h e following questions:

(1) What t y p e s of information would enable peop!e t o m a k e b e t t e r decisions for coping with t h e risk? How c a n e i t h e r t h e i n s u r a n c e i n d u s t r y , g o v e r n m e n t a t all levels (i.e., federal, s t a t e , a n d local) a n d / o r public i n t e r e s t g r o u p s aid in t h i s effort?

One c o u r s e of action is t o m a k e flood a n d e a r t h q u a k e coverage m o r e a t t r a c t i v e by presenting information t h r o u g h n o r m a l channels. The i n s u r a n c e a g e n t m a y serve a n i m p o r t a n t and useful function in this r e g a r d . To t h e e x t e n t t h a t he h a s t h e t r u s t of his clients, h e c a n stirnu- l a t e t h e i r awareness of t h e h a z a r d by telling t h e m t h e c h a n c e s of a h s a s t e r occurring and t h e p o t e n t i a l losses t h a t could r e s u l t . One way for t h e agent t o i n c r e a s e t h e client's c o n c e r n with t h e h a z a r d m a y b e t o p r e s e n t information o n t h e probability of a d i s a s t e r o n a d i f f e r e n t t i m e interval t h a n t h e t r a d i t i o n a l o n e y e a r period. For e x a m p l e , in describing t h e c h a n c e s of a 100 y e a r flood, t h e a g e n t could n o t e t h a t for s o m e o n e living in a h o u s e for 25 y e a r s , t h e c h a n c e s of s u f f e r i n g d a m a g e a t least once will b e .22. He c a n also provide details a s t o what coverage i s available a n d how m u c h it costs. Since m o s t individuals s e e m t o t r e a t insurance a s a n investment, t h e agent should e d u c a t e his cl' n t s t h a t t h e biggest r e t u r n on t h e i r policy 1s t o have no r e t u r n a t all.

kr

(16)

What is the balance between the use of market mechanisms for equat- ing supply and demand, developing appropriate incentives (e.g., taxes and subsidies) a s well a s regulatory measures (e.g.. required insurance coverage) in the design of a hazards strategy?

Financial institutions may play a key role h e r e by requiring some type of natural hazard insurance as a condition for a mortgage on residen- tial property. Several types of policies deserve consideration. One option would be a broader form of homeowners insurance which com- bines flood and earthquake. A less extreme proposal would be to add only earthquake coverage t o a standard homeowners policy and maintain t h e current flood insurance program. A third option would be to maintain the current insurance coverage m d provide disaster relief to special groups o r for special situations. Distributional cost considerations may suggest that special treatment be given t o low-income or elderly residents.

Evaluating Strategies

Any strategy or program impacts on the interested parties in different ways. The evaluation phase forces policy makers to come to grips with the question as t o the appropriate role of the public and private sectors in hazard management. To illustrate, consider two contrasting scenarios. In scenario 1.

acts of God, such as floods and earthquakes, are viewed a s a public responsibility; then liberal disaster relief should b e provided to all victims and/or highly subsidized insurance offered to residents in hazard-prone areas. In scenario 2, individuals a r e expected t o assume t h e responsibility for protecting themselves against the hazard; then private insurance should be offered and those who decide not t o purchase coverage voluntarily will be forced t o suffer t h e consequences. Scenario 1 is equivalent t o assigning a high weight t o potential victims and a low weight to the general taxpayer. Scenario 2 gives increased importance t o the general taxpayer. In this case, policies which require individuals faced with a risk to bear the cost of potential losses a r e viewed a s being attractive. How this evaluation process currently takes place and should t a k e place in the f u t u r e is an important topic for discussion.

V. TOWARDS A DESCRIPTIVE MODEL OF CHOICE

The examples presented above suggest the need for an understanding of the decision processes of the interested parties before one can recommend different policies. A first step in this direction is depicted in Figure 6 where the three interested parties-consurners. firms, and government-are linked to a s e t of events (e.g., catastrophes, accidents) each of which has a probability and loss associated with it. To make the problem more concrete and realistic assume t h a t there a r e n diderent consumer groups some of which have diEerent possible losses and probabilities associated with a particular event. For example, there m a y be m e r e n t exposures to a certain hazard so that the chances of incurring a specified loss will differ between individuals. Assume t h a t there are m identical firms each providing the same type of protection (e.g.. insurance) against the consequences of these events.

(17)

, l k F

CDNSUTERS

(PROBABILITY

LOSS)

Figure 6. Descriptive component.

P e r f o m a n c e of a Market S y s t e m

Given this simplified world it should t h e n be possible t o analyze how well a m a r k e t s y s t e m o p e r a t e s u n d e r a variety of different a s s u m p t i o n s regarding t h e a c c u r a c y of information by c o n s u m e r s a n d firms on t h e distribution of events.

For example, suppose c o n s u m e r s and firms have p e r f e c t information on t h e probability and loss distribution of events. What type of i n s u r a n c e policies w i l l b e offered t o c o n s u m e r g r o u p s ? How does the situation c h a n g e when t h e r e is i m p e r f e c t information b y e i t h e r or both of t h e s e p a r t i e s ?

A similar analysis c a n b e u n d e r t a k e n i f one p o s t u l a t e s difierent types of decision rules u s e d by c o n s u m e r s o r firms. For example, suppose t h a t e a c h con- s u m e r evaluates t h e benefits a n d costs of purchasing i n s u r a n c e a n d chooses a n a m o u n t (possibly no coverage) which maximizes e x p e c t e d utility. How m u c h i n s u r a n c e will e a c h c o n s u m e r g r o u p p u r c h a s e a n d what t y p e s of coverage will firms offer? Suppose, o n t h e o t h e r hand, t h a t c o n s u m e r s utilize a threshold model of choice: if t h e probability of the event is perceived t o b e below s o m e critical level t h e n t h e p e r s o n ignores its consequences a n d d o e s not consider a n y t y p e of p r o t e c t i o n ; otherwise they p u r c h a s e t h e a m o u n t of coverage which maximizes e x p e c t e d utility. What i m p a c t will s u c h a behavioral model have o n t h e t y p e s of i n s u r a n c e policies offered by firms and t h e d e g r e e of protection a d o p t e d by c o n s u m e r s ? In a similar vein one c a n investigate t h e l m p a c t of a m o d e l where f a c t o r s s u c h a s p a s t experience anc! discussions with friends and neighbors t r i g g e r s e a r c h for new information a n d i n t e r e s t in protective rneas- u r e s s u c h a s insurance.

The i m p a c t of different assumptions regarding t h e a c c u r a c y of information a n d alternative decision r u l e s c a n be investigated e i t h e r a t one point of t i m e o r in a dynamic context. When o n e looks a t t h e situation over time t h e n t h e r e is a n e e d t o specify t h e different rules t h a t c o n s u m e r s a n d firms a r e likely t o u"l' ~1 ize f o r updating information on t h e probability a n d consequences of specific events.

As shown by Arrow (1963) and Akerlof (1970) t h e r e a r e p r o b l e m s of adverse

(18)

selection when firms have misinformation or imperfect information on the risks each of the n consumer groups a r e facing. For example, if firms cannot distin- guish between high a n d low risk groups they may s e t a premium based on the average probability of a loss. If consumers have a c c u r a t e information on t h e hazard, high risk groups will find this policy to be much m o r e attractive t h a n low risk groups. and will purchase a proportionately larger s h a r e of t h e total coverage. Over time, claims experience leads firms t o s e t higher and higher premiums, thus making insurance less and less attractive t o those in t h e lower risk classes. Eventually t h e only group who finds insurance t o be attractive a r e those in the highest risk class.

The above example illustrating market failure is important for prescriptive purposes because it indicates t h a t the private sector may not provide satisfac- tory protective solutions t o potentially disastrous events, either because of misinformation and/or because of the decision processes of t h e interested parties. The lack of protection may then be very costly t o both the disaster victims (who may not be able t o g e t protection o r a r e unaware of t h e consequences of t h e hazard) a s well a s t h e general taxpayer (who may have t o foot t h e bill after a disaster occurs). The example also suggests the importance of determining what information c o n s u m e r s and firms have available, how a c c u r a t e these d a t a a r e and how they a r e actually utilized in the decision-making process.

Role of C o v e m m e n t

If consumers and/or firms have misinformation, one of t h e important roles t h a t t h e third p a r t y , government, can play is t o provide b e t t e r data on the hazard itself (e.g., losses, probabilities of its occurrence) as well a s ways of protecting oneself (e.g., available insurance, type of coverage and its cost). I t can also provide m o n e t a r y incentives t o encourage c e r t a i n actions (i.e.. subsidies) a s well a s disincentives (i.e.. fines, taxes) to inhibit o r discourage certain types of behavior. Finally, i t c a n regulate or require c e r t a i n types of actions.

The success of e a c h of t h e s e approaches depends on t h e decision processes of the interested p a r t i e s and t h e objectives of d f l e r e n t policies. Thus, if consu- m e r s are maximizing e x p e c t e d utility then a subsidized insurance premium would lead t o an increase in demand for coverage. This type of incentive system would have no effect for any consumer who behaved according to a threshold model and perceived t h e probability of a event t o be below his critical level. In t h e latter case t h e only way t o induce interest in insurance is to provide information on t h e hazard so t h a t t h e probability is perceived t o be above the critical level(s) or t o require t h e person t o have insurance coverage.

From a dynamic viewpoint t h e r e is a need t o u n d e r s t a n d differences in ez- a n t e estimates a n d ex-post valuations and their effect on policy. Prior t o a disaster a n individual is likely to behave with one s e t of estimates of the probability and losses. After a n event occurs he may revise his estimate considerably, partly on t h e basis of t h e new information (i.e.. updating his prior estimates of probabilities and losses) but also because of the n a t u r e of this decision process (e.g., he now views the probability to be above a critical threshold level and hence is concerned with possible losses). If government policy responds t o these ex post perception changes in a way that was unanticipated prior t o the disaster, this process has to be understood before one designs policies. A clear example of this behavior is in t h e natural hazards field: government provided liberal d s a s t e r after the o c c u r r e n c e of a disaster because few people protected

(19)

themselves prior to the event. If crises normally trigger unanticipated reac- tions due to political and social pressures (c.f.. the Three Mile Island response), then this process must be taken into account in designing appropriate s t r z - tegies for dealing with low probability-high consequence events.

Finally, t h e r e a r e a set of philosophical znd ethical issues that have t o be addressed directly when evaluating the role of government as p a r t of any alternative plan. Given our increased understanding of t h e imperfect information and simplified rules t h a t people use in making decisions, there is t h e open question as to "when should we protect individuals from themselves?" If policy mak- e r s have learned from experience t h a t t h e r e is e x post regret by uninsured con- s u m e r s a f t e r an event, what type of incentives o r regulations, if any, should be taken e x ante? There is no- easy answer t o this query but it should be explicitly addressed as an issue regarding the appropriate role of government in dealing with t h e consequences of low probability events. I t also illustrates the interaction between the descriptive and prescriptive components depicted in the conceptual framework (Figure 1) which has motivated this paper.

(20)

For an excellent discussion of how one can specify goals and objectives for s o c i e t a l p r o b l e m s , s e e Keeney and Raiffa 1976, C h a p t e r s 1 & 2.

For an analysis of alternative remedies proposed by t h e Federal Trade Com- mission in t h e context of t h e s e a n d o t h e r c o s t s , s e e Federal Trade Comrnis- sion 1979.

The specific s t a n d a r d s a r e t h a t t h e foot c a n n o t r e a c h t h e blade of t h e mower a n d t h a t t h e blade m u s t s t o p within 3 s e c o n d s of release of t h e power switch s o t h e hand c a n n o t r e a c h t h e t u r n i n g blade.

These differences a r e clearly s e e n in the LNG siting d e b a t e in California.

For m o r e detail on this c a s e , s e e Ahearn 1980, in p r e s s ; Deutsch in p r e s s ; K u n r e u t h e r 1980; and Linnerooth 1980.

The m a t e r i a l in t h e next section s u m m a r i z e s t h e findings from a four-year s t u d y s u p p o r t e d by funds f r o m t h e National Science Foundation. R e a d e r s i n t e r e s t e d in m o r e detail a r e r e f e r r e d t o K u n r e u t h e r , e t al. 1978.

More d e t a i l s on t h e i n s u r a n c e e x p e r i m e n t s c a n b e found in Slovic, e t al.

1977. The m a t e r i a l also a p p e a r s in Kunreuther, e t al. 1979, Chap. 7.

A

m o r e detailed discussion of policy options a p p e a r s in Ginsberg and Kun- r e u t h e r (in p r e s s ) .

Kunreuther a n d S c h o e m a k e r (in press) provide a m o r e detailed discussion of t h e role of the a g e n t and the insurance i n d u s t r y in promoting t h e sale of flood coverage.

(21)

REFERENCES

Ahern. William. 1980. "California Meets t h e LNG Terminal." Coastal Zone Management Journal, 7: 195,221.

Ahern, W. (in press). "The Role of Technical Analyses in California Energy Facil- ity Siting Decisions." Proceedings of t h e Liquefied Energy Cases Facil- ity Siting Task Force Meeting, IIASA. Laxenburg, Austria. S e p t e m b e r

1980.

Akerlof, C. 1970. "The Market for 'Lemons': Quality U n c e r t a i n t y and t h e Market Mechanisms." Quarte7-ly Journal of Economics, 84:480-500.

Arrow, K. 1963. " U n c e r t a i n t y a n d t h e Welfare Econo,nics of Medical Care."

American Economic Review, 53: 94 1-973.

sick, T., and C. Hohenemser. 1979. "Target: Fighway Risks. I. Taking Individual h m . " E n v i r o n m e n t , 21:i'-15, 29-38.

Deutsch, RW. (in press). "Siting a n L.N.C. Fncility in California: The Regulatory Framework a n d t h e Factors Involved in t h e Decision Making Process."

Proceedings of t h e Liquefied Energy Gases Facility Siting Task Force Meeting, IIASA, Laxenburg, Austria, S e p t e m b e r 1980.

Federal Trade Commission. 1979. Consumer I n f o r n a t i o n R e m e d i e s . Washing- ton, D.C.: U.S. Government Printing Office.

Fischhoff. B., P. Slovic, S. Lichtenstein, 3. Read, a n d B. Combs. 1978. "How S a f e is Safe Enough? A Psychometric Study of Attitudes Towards Technolog- ical Risks a n d Benefits." Policy Sciences 9: 127-152.

Ginsberg. R.. and H. Kunreuther. (in press). "Behavioral Determinants of

(22)

Market Failure: The Case of Disaster Insurance," in S. Fiddle, ed., U n c e r t a i n t y : S o c i a l a n d B e h a v i o r a l D i m e n s i o n s . New York: Praeger.

Harris, J. 1980. "Taxing Tar and Nicotine." A m e r i c a n E c o n o m i c R w i e w , 70:300- 311.

Keeney, R., and H. Raiffa. 1976. D e c i s i o n s w i t h M u l t i p l e O b j e c t i v e s . New York:

John Wiley.

Kunreuther, H., R. Ginsberg, L. Miller, P. Sagi, P. Slovic, B. Borkan. and N. Katz.

1978. D i s a s t e r I n s u r a n c e P r o t e c t i o n : P u b l i c P o l i c y Lesso7~,9. New York:

Wiley.

Kunreuther, H., and P. Schoemaker. (in press). "Decision Analysis for Complex Systems," K m w l e d g e

Kunreuther, H. 1980. "Societal Decision Making for Low Probability Events:

Descriptive a n d Prescriptive Aspects." IIASA Working Paper. WP-80-164, Laxenburg, Austria.

Linnerooth, J. 1980. "A Short History of t h e California LNG Terminal." IIASA Working P a p e r , WP-80-155. Laxenburg, Austria.

National Analysts Inc. 1971. M o t i v a t i n g F a c t o r s in t h e U s e of R e s t r a i n t S y s t e m s . Final Report, c o n t r a c t FH 11-7610. Philadelphia: prepared for the U.S Department of Transportation.

Office of Technology Assessment. 1977. T r a n s p o r t a t i o n of L i q u e f i e d N a t u r a l G a s . Washington. D.C.: Office of Technology Assessment.

Slovic. P., B. Fischhoff, a n d S. Lichtenstein. 1978. "Accident Probabilities and Seat Belt Usage: A Psychological Perspective." A c c i d e n t A n a l y s i s z n d P r e v e n t i o n , 10, 281-295.

Slovic, P . , B. Fischhoff, S. Lichtenstein, B. Corrigan, and B. Combs. 1977.

"Preference for Insuring Against Probable Small Losses: Implications for the Theory a n d Practice of Insurance." J o u r n a l of R i s k a n d I n s u r a n c e , 4 4 , 237-258.

Tversky, A , and D. Kahneman. 1974. "Judgment Under Uncertainty: Heuristics and Biases." S c i e n c e 185:1124-1131.

Washington Post. 1979a. "Safety Rules Eyed for Power Mowers." January 20.

Washington Post. 1979b. "Prince George County's Smoking Ban Becomes Law,"

March 2.