Protecting against Low Probability Disasters: The Role of Worry Christian Schade Humboldt-Universität zu Berlin Howard Kunreuther The Wharton School University of Pennsylvania Philipp Koellinger Erasmus University Rotterdam Netherlands Klaus Peter Kaas Goethe-University Frankfurt December 2009 Working Paper # 2009-03-23 Submission to the Journal of Behavioral Decision Making _____________________________________________________________________ Risk Management and Decision Processes Center The Wharton School, University of Pennsylvania 3730 Walnut Street, Jon Huntsman Hall, Suite 500 Philadelphia, PA, 19104 USA Phone: 215‐898‐4589 Fax: 215‐573‐2130 http://opim.wharton.upenn.edu/risk/ ___________________________________________________________________________
49
Embed
Protecting Against Low Probability Disasters: A Large Stakes Experiment
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Protecting against Low Probability Disasters: The Role of Worry
Christian Schade Humboldt-Universität
zu Berlin
Howard Kunreuther The Wharton School
University of Pennsylvania
Philipp Koellinger Erasmus University Rotterdam
Netherlands
Klaus Peter Kaas Goethe-University
Frankfurt
December 2009 Working Paper # 2009-03-23
Submission to the
Journal of Behavioral Decision Making
_____________________________________________________________________ Risk Management and Decision Processes Center The Wharton School, University of Pennsylvania 3730 Walnut Street, Jon Huntsman Hall, Suite 500
THE WHARTON RISK MANAGEMENT AND DECISION PROCESSES CENTER
Established in 1984, the Wharton Risk Management and Decision Processes Center develops and promotes effective corporate and public policies for low‐probability events with potentially catastrophic consequences through the integration of risk assessment, and risk perception with risk management strategies. Natural disasters, technological hazards, and national and international security issues (e.g., terrorism risk insurance markets, protection of critical infrastructure, global security) are among the extreme events that are the focus of the Center’s research.
The Risk Center’s neutrality allows it to undertake large‐scale projects in conjunction with other researchers and organizations in the public and private sectors. Building on the disciplines of economics, decision sciences, finance, insurance, marketing and psychology, the Center supports and undertakes field and experimental studies of risk and uncertainty to better understand how individuals and organizations make choices under conditions of risk and uncertainty. Risk Center research also investigates the effectiveness of strategies such as risk communication, information sharing, incentive systems, insurance, regulation and public‐private collaborations at a national and international scale. From these findings, the Wharton Risk Center’s research team – over 50 faculty, fellows and doctoral students – is able to design new approaches to enable individuals and organizations to make better decisions regarding risk under various regulatory and market conditions.
The Center is also concerned with training leading decision makers. It actively engages multiple viewpoints, including top‐level representatives from industry, government, international organizations, interest groups and academics through its research and policy publications, and through sponsored seminars, roundtables and forums.
More information is available at http://opim.wharton.upenn.edu/risk.
Protecting against Low Probability Disasters: The Role of Worry
Christian Schade Howard Kunreuther Philipp Koellinger Klaus Peter Kaas*
We conducted a large stakes insurance experiment with small probabilities of losses and a
realistic form of ambiguity. Our results demonstrate that worry plays a more important role in the
decision to consider insurance against high losses that are rare than does subjective probability
estimates. For those who do have an interest in buying insurance, worry is also positively related
to the willingness to pay (WTP) for coverage. If faced with an ambiguous risk, an individual is
more willing to consider insurance and pay higher amounts than when the probability of a loss is
specified precisely. An approximately 1,000-fold increase in the ambiguous probability did not
change the percentage of those who consider insurance and had a very small positive impact on
WTP. Our results provide insights into the low probability insurance puzzle where some
individuals are willing to pay too much and others nothing for coverage in relation to the risk
associated with the specific event.
Professor Christian Schade is Director of the Institute for Entrepreneurial Studies and
Innovation Management, School of Business and Economics, Humboldt-Universität zu Berlin, Germany, email: [email protected]
Professor Howard Kunreuther is Director of the Risk Management and Decision Processes Center, Wharton School, University of Pennsylvania, Philadelphia, USA, email: [email protected]
Philipp Koellinger is assistant professor at the Erasmus School of Economics, Erasmus University Rotterdam, the Netherlands, email: [email protected]
* Professor Klaus Peter Kaas, Marketing Department, Faculty of Business and Economics, Goethe-University, Frankfurt/M., Germany, email: [email protected]
2
Introduction
Imagine you are facing a risk that is characterized by a very small probability of
occurrence but, if it occurs, will cause significant damage relative to your total wealth. Examples
of such risks are floods and earthquakes, as well as fire and theft. If you were offered insurance
coverage against such a risk would you try to estimate the probability and multiply this figure by
the value or utility of the potential loss? What role would affect or emotional factors such as
worry play in your decision with respect to specific outcomes?
Understanding insurance decisions with respect to low-probability disasters has been a
challenge for psychologists as well as economists. Field studies and controlled laboratory
experiments have posed the following low probability insurance puzzle: (1) many individuals do
not voluntarily purchase coverage even when premiums are highly subsidized (Kunreuther, 1978;
Slovic, Fischhoff, Lichtenstein, Corrigan, & Combs, 1978). (2) In a controlled experiment con-
sistent with these earlier studies, McClelland, Schulze, and Coursey (1993) showed that most
individuals are either unwilling to pay a penny for low-probability insurance or far too much
when compared with the expected loss from the event. The early version of prospect theory
(Kahneman & Tversky, 1979) takes this feature into account by having a discontinuity in the
probability weighting function close to zero.
Most earlier studies in decision making including the above-mentioned ones have focused
on explaining deviations from the predictions derived from normative models of choice such as
(subjective) expected utility theory (Savage, 1954; von Neumann & Morgenstern, 1947). Only
recently has behavioral decision theory concerned itself with the impact that affect and emotion
have on decision making with respect to protective measures (see, e.g., Hogarth & Kunreuther,
Table 6: Pearson correlation coefficients between probability judgment and worry
Ambiguity and low risk(1 / 5000)
Ambiguity and high risk(1 / 5)
WTP = 0 0.13 (0.54) -0.36 (0.27) WTP > 0 0.03 (0.69) 0.14 (0.25) Average 0.07 (0.36) 0.05 (0.65) Note: Significance levels are reported in parentheses. Table 7: Threshold model estimation results for insurance against disasters with large and small
AIC 209 (df=4) 216 (df=4) 195 (df=5) * denotes marginal significance at 90% confidence ** denotes significance at 95% confidence *** denotes significance at 99% confidence + The number of observations is not twice the number of individuals because there are some missing values for the level of worry and the probability estimates in part A of the experiment. ++ Pseudo R2 for regression on a and adjusted R2 for regression on y. Reference categories: joint policies (x4) and low risk (x5).
38
Table 8: Threshold model estimation results for insurance against disasters with small known
# observations 167 108 R2++ 0.2 0.06 Prob > chi2 0.00 Log likelihood -87 AIC1) 180 (df=3) * denotes marginal significance at 90% confidence ** denotes significance at 95% confidence *** denotes significance at 99% confidence + The number of observations is not twice the number of individuals because there are some missing values for the level of worry and the probability estimates in part A of the experiment. ++ Pseudo R2 for regression on a and adjusted R2 for regression on y 1) The AIC criterion is based on log likelihood but also punishes model complexity. Table 9: Impact of worry on average WTP with small exact and ambiguous probabilities
A.1 Experimental Instructions (translations of parts A and B of the experiment)
Group 1 (separate policies)
Part A: Ambiguity
You inherited a small painting and have received a photograph of it. The photo carries an
individual identification number. You do not know if the painting is an original or a
reproduction. If it is an original it is worth 2,000 DM. If it is a reproduction it is worth
nothing.
One person in the entire group of respondents participating in our experiment (about 260 to
280 people) has an original painting. All others have reproductions. Which one of the
paintings is the original will be determined by a random draw symbolizing the decision of an
art appraiser at the end of the entire experiment. The person who has the original painting will
actually receive the value of the painting: 2,000 DM (in real bills!).
Theft and fire threaten your painting.
Whether or not the painting will be stolen will be determined by the weather conditions in
July. If it will rain on 24 days in July (not more but also not less), a theft will occur. More
precisely, the painting will be stolen if the weather station at the Frankfurt Airport will report
on exactly 24 days of rain. A day is defined as a rainy day if there is at least 1 mm of rain on
this day.
The weather conditions in August determine if a fire will destroy the painting. If it will rain on
23 days in August (not more but also not less), a fire will occur. More precisely, a fire will
40
destroy the painting if the weather station at the Frankfurt Airport will report exactly 23 days
of rain. Here again a day is defined as a rainy day if there is at least 1 mm of rain on this day.
You can buy insurance policies against either each or both of these risks. If you have an
insurance policy against theft or fire and the painting will be stolen or destroyed by fire,
respectively, the insurance will reimburse you for the loss of 2,000 DM. If you have an
insurance policy against fire and the painting will be destroyed by fire, the insurance will
reimburse you for the value of 2,000 DM.
The insurance company will sell the insurance policy and charge the money for it only in case
an art appraiser, represented by the random draw of the experimenter, finds out that your
painting is an original. Thus, for all respondents having the reproduction the payments for the
insurance policies will remain hypothetical. However, for the one having the original painting
they will become true and have to be paid from his or her own money.
The selling procedure for the theft insurance policy is organized in the following way:
Before the experiment, the experimenter selected a secret selling price for the theft insurance
policy. He or she wrote it on a piece of paper and put it into the envelope on the front desk.
You are now required to write a buying price equal to your maximum willingness to pay for
the theft insurance policy on the form in front of you and to put it in the respective envelope.
After the experiment, the experimenter will open the envelope with the selling price. If your
buying price is equal to or higher than the secret selling price you will have bought or are able
to buy the theft insurance policy should you be the person with the original painting.(if you are
the one who has the original painting). If your buying price is lower than the secret price, you
are not able to buy the theft insurance policy.
41
Note that you have no information about the selling price for the theft insurance policy. The
experimenter changes this price every time.
In this situation, the best you can do is to state your true value, your maximum willingness to
pay for the theft insurance policy.
It does not make sense to state a buying price higher than your maximum willingness to pay
since you may end up paying this high price.
It does also not make sense to state a price lower than your maximum willingness to pay. If
your stated price is lower than the selling price but you, in fact, would have been willing to
pay that price you may end up without the theft insurance policy even if you would have liked
to buy it for that price if you are the one who has the original painting.
If you do not want to buy the theft insurance policy please state 0 DM on the respective form.
Please do not announce your buying price to the others and do not raise questions that allow
the other participants to guess your buying price.
Again, note that you only have to actually pay the price for the insurance policy if you are the
one who has the original painting. This is because the insurance company will only sell the
insurance policy if the painting is verified as the original. In this case, the person who has the
original painting is able to buy insurance. He or she has to pay for the coverage of the
insurance policies from his or her own money.
Basically, that means that you are buying insurance for the original and that you only pay for it
in case you have it.
Now, please put the form with your maximum buying price in the appropriate envelope and
hand it over to the experimenter.
42
The selling procedure for the fire insurance policy is organized the following way:
The selling procedure of the fire insurance policy is organized in exactly the same way as the
selling procedure for the theft insurance policy, i.e. again there is a secret selling price in an
envelope, and you again are supposed to state your maximum buying price.
Now, please put the form with your maximum buying price in the appropriate envelope and
hand it over to the experimenter.
Part B: Risk
You inherited a small sculpture and have received a photograph of it. The photo carries an
individual identification number. You do not know if the sculpture is an original or a
reproduction. If it is an original it is worth 2,000 DM. If it is a reproduction it is worth
nothing.
One person in the entire group of respondents participating in our experiment (about 260 to
280 people) has an original sculpture. All others are reproductions. Which one of the
sculptures is the original will be determined by a random draw symbolizing the decision of an
art appraiser at the end of the entire experiment. The person who has the original sculpture
will actually receive the value of the sculpture: 2,000 DM.
Theft and fire threaten your sculpture. Both risks have a known chance of occurrence:
Hazard Chance of occurrence
Theft in one of 10.000 cases
Fire in one of 10.000 cases
43
A bingo cage with 100 balls will be used to determine whether or not the sculpture will be
stolen and whether or not it will be destroyed by fire.
- Whether or not the sculpture will be stolen will be determined by the following two-stage
procedure: If a ball with a number between 2 and 100 is drawn, no theft will have
occurred. We will continue drawing of the bingo cage with 100 balls after the first draw
only if a ball carrying the number 1 is drawn. Otherwise nothing happened. In a second
draw, another ball from the bingo cage with 100 balls will be taken. Theft occurs if the ball
with the number 1 is drawn in the second stage. The chance that both these events occur is
exactly 1 in 10,000.
- Secondly we will determine the case if fire occurs. We will proceed with the same two-
stage procedure as used for the theft situation.
You can buy insurance policies against either each or both of these risks. If you have an
insurance policy against theft or fire and the sculpture will be stolen or destroyed by fire,
respectively, the insurance will reimburse you for the loss of 2,000 DM. If you have an
insurance policy against fire and the fire destroys the sculpture, the insurance will reimburse
you for the value of 2,000 DM.
The insurance company will sell the insurance policy and charge the money for it only in case
an art appraiser, represented by the random draw of the experimenter, finds out that the
sculpture is an original. Thus, for all respondents having the reproduction, the payments for
the insurance policies will remain hypothetical. However, for the one having the original
sculpture they will become true and have to be paid from his or her own money.
44
The selling procedure for the theft insurance policy is organized the following way:
The selling procedure of the theft insurance policy is organized in exactly the same way as the
selling procedure for the theft and fire insurance policies in the first part of the experiment, i.e.
there again is a secret selling price in an envelope, and you again are supposed to state your
maximum buying price.
Now, please put the form with your maximum buying price in the appropriate envelope and
hand it over to the experimenter.
The selling procedure for the fire insurance policy is organized the following way:
The selling procedure of the fire insurance policy is organized in exactly the same way as the
selling procedure for the theft and fire insurance policy in the first and the theft insurance
policy in the second part of the experiment, i.e. there again is a secret selling price in an
envelope, and you again are supposed to state your maximum buying price.
Now, please put the form with your maximum buying price in the appropriate envelope and
hand it over to the experimenter.
45
Group 2 (one policy)
The only difference between groups 1 and 2 was that in group 2 we sold bundled rather than
separate insurance in both parts A and B.
Therefore, the part of the instructions dealing with insurance was written up as follows in part A
(B) of the experiment:
You can buy an insurance policy against each of these two risks. If you have an insurance
policy and your painting (sculpture) will be stolen or destroyed by fire, the insurance will
reimburse you for the loss of 2,000 DM.
Moreover, the selling procedure was described in the following way in part A (B) of the
experiment:
Before the experiment, the experimenter selected a secret selling price for the insurance
policy. He or she wrote it on a piece of paper and put it into the envelope on the front desk.
In the following you are required to write a buying price equaling your maximum willingness
to pay for the insurance policy on the form in front of you and to put it in the respective
envelope.
After the experiment, the experimenter will open the envelope with the selling price. If your
buying price is equal to or higher than the secret selling price, you are able to buy the
insurance policy (in case you are the one who has the original painting (sculpture)). If your
buying price is lower than the secret price, you are not able to buy the insurance policy.
Note that you have no information about the selling price for the insurance policy. The
experimenter changes this price every time.
46
In this situation, the best you can do is to state the true value of your maximum willingness to
pay for the insurance policy.
It does not make sense to state a buying price higher than your maximum willingness to pay
since you may end up paying this high price.
It does also not make sense to state a price lower than your maximum willingness to pay. If
your stated price is lower than the selling price but you in fact would have been willing to pay
that price you may end up without the insurance policy even if you would have liked to buy it
for that price if you are the one who has the original painting (sculpture).
If you do not want to buy the insurance policy please state 0 DM on the respective form.
Please do not announce your buying price to the other participants and do not raise questions
that allow others to guess your buying price.
Again, note that you only have to actually pay the price for the insurance policy if you are the
one who has the original painting. This is because the insurance company will only sell the
insurance policy if the painting is verified as the original. In this case, the person who has the
original painting is able to buy insurance. He or she has to pay for the coverage of the
insurance policies from his or her own money.
Basically that means that you are buying insurance for the original and that you only pay for it
in case you have it.
Now, please put the form with your maximum buying price in the appropriate envelope and
hand it over to the experimenter.
47
A.2: Probability table Please report how probable you have judged „exactly 24 rain days in July“ occurring. Please check an interval that covers the probability you are judging to be correct first. Afterwards please report the exact probability in the right column. Explanation: A chance of 1 in 1.000.000 implies that a July with exactly 24 rain days occurs – on average – every 1.000.000 years. Chance: 1 in Please
check: Exactly:
1 to 5 1 in _______________ 5 to 10 1 in _______________ 10 to 50 1 in _______________ 50 to 100 1 in _______________ 100 to 500 1 in _______________ 500 to 1.000 1 in _______________ 1.000 to 5.000 1 in _______________ 5.000 to 10.000 1 in _______________ 10.000 to 50.000 1 in _______________ 50.000 to 100.000 1 in _______________ 100.000 to 500.000 1 in _______________ 500.000 to 1.000.000 1 in _______________ 1.000.000 to 5.000.000 1 in _______________ 5.000.000 to 10.000.000 1 in _______________ Less probable Exactly 1 in _____________