1. Insurance and Behavioral Economics: Improving Decisions in the Most Misunderstood Industry (with Mark Pauly and Stacey McMorrow) Howard Kunreuther kunreuther@wharton.upenn.edu James G. Dinan Professor of Decision Sciences & Public Policy Co-Director Risk Management and Decision Processes Center The Wharton School University of Pennsylvaniakunreuther@wharton.upenn.edu Insurance Economics: New Risks, New Regulation, New Approaches SIFR Conference, Stockholm, August 24-25, 2015 1

2. Some insurance markets work well Term life insurance Auto collision insurance Homeowners’ insurance But Low Probability-High Consequence (LP-HC) events puzzle consumers, insurers and politicians/regulators Consumers: Very limited personal experience with events Insurers: Ambiguous risks and correlated losses pose insurability challenges Politicians/Regulators: Concerned with re-election and fairness/equity What is Great and Not-So-Great about Insurance 2

3. Three Motivating Examples 3 Example 1: Purchase of Disaster Insurance by Homeowners Most homeowners in flood prone areas do not voluntarily purchase flood insurance – even when it is highly subsidized – until after they suffer damage from a disaster. Those who do not experience losses in the next few years are likely to cancel their policy. Similarly, demand for earthquake insurance in California increased significantly after the Northridge earthquake of 1994 – the last severe quake in the state; today relatively few homeowners have coverage.

4. Three Motivating Examples 4 Example 2: Insuring Against the Terrorism Risk Prior to the terrorist attacks of September 11, 2001, actuaries and underwriters did not price the risk associated with terrorism nor did they exclude this coverage from their standard commercial policies. This was surprising given the World Trade Center bombing in 1993, the 1995 Oklahoma City bombing and other terrorist-related events throughout the world. Following 9/11, most insurance companies refused to offer coverage against terrorism, considering it to be an uninsurable risk.

5. Example 3: Insurance Regulation of Insurance State insurance regulators sometimes have restricted insurers from setting premiums that reflect risk. Following Hurricane Andrew in 1992, the Florida insurance commission did not allow insurers to charge risk-based rates and restricted them from canceling existing homeowners’ policies. After the severe hurricanes of 2004 and 2005 in Florida, the state-funded company, Citizens Property Insurance Corporation offered premiums in high-risk areas at subsidized rates, thus undercutting the private market. Today, Citizens is the largest provider of residential wind coverage in Florida. Three Motivating Examples 5

6. These three examples indicate that insurance today is not effectively meeting two of its most important objectives: Providing information to those residing in hazard prone areas Incentivizing those at risk to invest in loss reduction measures Factory mutual companies in the 19 th century played these roles very effectively Required inspections of factories prior to issuing a policy Poorly-monitored factories had their policies canceled Premiums reflected risk and were reduced for factories that instituted loss prevention measures Key Roles of Insurance 6

7. How Can Insurance Encourage Loss Prevention? 7 Questions to be addressed: What are the decision processes that explain the actions taken by each of the interested parties based on the above three examples? What are two guiding principles for insurance to encourage loss prevention prior to a disaster? What long-term roles can the private and public sectors play if these principles are implemented? How can the National Flood Insurance Program be modified to serve as a model for linking insurance and mitigation?

8. A New Era of Catastrophes Linking Intuitive and Deliberative Thinking for Dealing with Extreme Events Guiding Principles for Insurance Insurance Voucher and Mitigation Loan Program Challenges and Questions for Discussion 8 Outline of Talk

9. 9 WORLDWIDE EVOLUTION OF CATASTROPHES, 1980-2014

10. 10 NORTHERN EUROPE IF GREENLAND’S ICE SHEET MELTS

11. A New Era of Catastrophes Linking Intuitive and Deliberative Thinking for Dealing with Extreme Events Guiding Principles for Insurance Developing Long-term Strategies for Extreme Events 11 Outline of Talk

12. KNOWLEDGE FOR ACTION Linking Intuitive and Deliberative Thinking for Dealing with Extreme Events 12

13. 13 System 1 operates automatically and quickly with little or no effort Individuals use simple associations including emotional reactions Highlight importance of recent past experience Basis for systematic judgmental biases and simplified decision rules System 2 allocates attention to effortful and intentional mental activities Individuals undertake trade-offs implicit in benefit-cost analysis Recognizes relevant interconnectedness and need for coordination Focuses on long-term strategies for coping with extreme events Intuitive Thinking (System 1) & Deliberative Thinking (System 2)

14. Behavior Triggered by Intuitive (System 1) Thinking Availability Bias – Estimating likelihood of a disaster by its salience Threshold Models – Failure to take protective measures if perceived likelihood of disaster is below threshold level of concern Imperfect Information – Misperceives the likelihood of event occurring and its consequences. Myopia – Focus on short-time horizons in comparing upfront costs of protection with expected benefits from loss reduction 14

15. The Lowland family is considering whether to invest $1,500 in flood proofing their house so it is less susceptible to water damage. Hydrologists have estimated that the chances of storm surge from hurricanes affecting their home is 1/100, and that if it occurs, the savings from flood proofing will be $27,500. If premiums reflect risk their annual insurance cost will be reduced by $275 (i.e., 1/100 $27,500) if they undertake this investment. Failure to Invest in Flood Adaptation Measures 15

16. Responses by the Lowland family: Imperfect information: Lowland family misperceives flood risk, thinking that it is 1/1000 rather than 1/100 Threshold model: Flood risk is below their level of concern Myopic behavior: Failure to consider long-term benefits of flood protection Cancellation of flood insurance: Consider it to be a poor investment since they have not suffered any flood-related damage Many banks do not enforce the flood insurance requirement Many states do not enforce building codes 1/3 of the damage from Hurricane Andrew (1992) could have been avoided if Florida had enforced its building codes Today, Florida has well-enforced codes (learning from a disaster) Illustrations of Intuitive (System 1) Thinking 16

17. Many homeowners cancel their flood policy if they have not experienced a flood for several years. Reason: Flood insurance was not a good investment. Data: Of 1,549 victims of a flood in August 1998 in northern Vermont, FEMA found 84% of residents in SFHAs did not have flood insurance. 45% were required to purchase it. 17 Lack of Interest in Protection Against Disasters: Cancellation of Flood Insurance Even When Required

18. New Business Year 20012002200320042005200620072008 Housing Units 841,000876,0001,186,000986,000849,0001,299,000974,000894,000 1 year 73%67%77%78%76%73%74%73% 2 years 49%52%65% 63%59%58% 3 years 39%44%57%55%53%48% 4 years 33%38%50%48%44% 5 years 29%33%44%38% 6 years 25%30%33% 7 years 22%26% 8 years20% Sources: Michel-Kerjan, Lemoyne de Forges and Kunreuther – Data from NFIP/FEMA Note: our analysis of the American Community Survey reveals that the median length of residence was about 6 years over this period. Dynamic Analysis of Flood Insurance Tenure 18

19. Prior to the Loma Prieta earthquake (1989) only 22.4 percent of the homes had earthquake insurance. Four years later, 36.6 percent had purchased earthquake insurance—a 72 percent increase. One year after the Northridge earthquake of 1994 more than two-thirds of the homeowners surveyed in Cupertino County had purchased earthquake insurance. Consumer Behavior: Purchase and Cancellation of Earthquake Insurance 19 There have been no severe earthquakes in California since Northridge and only 10 percent of those in seismic areas of the state currently have coverage. If a severe quake hits San Francisco in the near future, the damage could be as high as $200 billion, and it is likely that most homeowners suffering damage will be financially unprotected.

20. Prior to 9/11, insurers did not charge anything for terrorism coverage despite the attempted bombing of the World Trade Center in 1993, the 1995 Oklahoma City bombing and terrorist attacks throughout the world. After 9/11, most insurers refused to offer terrorism insurance, or if they did provide coverage they charged extremely high premiums. Insurer Behavior: Terrorism Insurance 20

21. Responses by Insurers Threshold Behavior: Prior to 9/11 insurers treated the likelihood of a terrorist attack in the U.S. as below their threshold level of concern so ignored potential consequences. Availability Bias: After 9/11 insurers focused on enormous potential claim payments from another terrorist attack. As a result they felt terrorism was an uninsurable risk. Imperfect Information: Insurers failed to take into account the likelihood of a future terrorist attack when determining premiums they would have to charge for coverage, and how much firms would be willing to pay for protection. Target of Opportunity for Insurers: 6 months after 9/11 a brokerage firm negotiated an insurance policy where an industrial company paid $900,000 for $9 million in coverage for damage to their building next year from a terrorist attack. Terrorism Insurance: Insurer Behavior Triggered by Intuitive Thinking 21

22. After the severe hurricanes of 2004 and 2005 in Florida, the state-funded company, Citizens Property Insurance Corporation provided homeowners in hurricane-prone areas with highly subsidized insurance policies. If Florida had experienced a severe hurricane in the next few years, Citizens would have been insolvent. Question: Who would bail out Citizens should such a storm had occurred? Decision processes of politicians/regulators Threshold Behavior: Severe hurricane is below threshold level of concern of the public Myopic Behavior: Focus on short-term benefits of economic development without considering long-term consequences of a severe hurricane to the affected area and financial stability of the state Politicians’/Regulators’ Behavior: Forming the Citizens Property Insurance Corporation in Florida 22

23. A New Era of Catastrophes Linking Intuitive and Deliberative Thinking for Dealing with Extreme Events Guiding Principles for Insurance Developing Long-term Strategies for Extreme Events 23 Outline of Talk

24. Principle 1: Premiums reflecting risk Insurance premiums should be based on risk to provide individuals with accurate signals as to the nature of the hazards they face and to encourage them to engage in cost- effective mitigation measures to reduce their vulnerability. Principle 2: Dealing with equity and affordability issues Any special treatment given to those deserving special treatment (e.g. low-income individuals) currently residing in hazard-prone areas should come from general public funding and not through insurance premium subsidies. Funding could be obtained from several different sources (e.g. general taxpayer revenue, state government or taxing insurance policyholders) depending on the response to the question “Who should pay?” 24 Guiding Principles for Insurance

25. Cost of Adaptation Measure: $1,500 to flood-proof their home Nature of Disaster: –1/100 chance of disaster –Reduction in loss ($27,500) Expected Annual Benefits: $275 (1/100 * $27,500) Annual Discount Rate: 10% 25 Lack of Interest in Mitigation by Lowland Family with Risk-Based Premiums

26. Upfront cost of adaptation Expected benefits over time 26 Expected Benefit-Cost Analysis of Adaptation (Annual Discount Rate 10%)

27. Illustrative Example: The Lowland Family Cost to flood-proof their home: $1,500 Expected annual benefit of partial roof adaptation: $275 (1/100 * $27,500) Annual payments from a 20-year $1,500 loan at 10% annual interest rate: $145 Reduction in annual insurance payment: $275 Reduction in annual payments due to adaptation: $275-$145= $130 Linking Insurance and Mitigation Via Multi-Year Loans 27

28. 28 Everyone is a Winner Homeowner: Lower total annual payments Insurer: Reduction in catastrophe losses and lower reinsurance costs Financial institution: More secure investment due to lower losses from disaster General taxpayer: Less disaster assistance

29. A New Era of Catastrophes Linking Intuitive and Deliberative Thinking for Dealing with Extreme Events Guiding Principles for Insurance Developing Long-term Strategies for Extreme Events 29 Outline of Talk

30. Long-term Strategies for Dealing with Extreme Events 30 Choice architecture Frame the problem so that the risks are salient Develop scenarios so people recognize the importance of insurance and investing in loss reduction measures prior to a disaster Public-private partnerships Assist those who cannot afford to invest in protective measures Public sector provides financial protection to private insurers against catastrophic losses Multi-year insurance Provide premium stability to policyholders Lower marketing costs to insurers Reduction in the cancellation of coverage by those at risk

31. Choice Architecture 31 Provide better information to consumers on the role of insurance The best return on an insurance policy is no return at all Use availability bias to focus on consequences to consumers Highlight financial problems if disaster occurred and the property were destroyed because it was unprotected and uninsured Overcome threshold model used by insurers Construct worst-case scenarios before a disaster Assign likelihoods to worst-case scenarios after a disaster to show that risk is insurable Stretch time horizon Example: Likelihood of 100-year flood Next year: 1 in 100 25 years: greater than 1 in 5 chance of experiencing at least 1 flood during this period

32. Encourage investment in loss reduction measures Risk-based premiums Home improvement mitigation loans tied to property Premium reductions for undertaking mitigation measures Address affordability issue Means-tested vouchers for current residents Covers insurance premium and mitigation loan Condition for a voucher: You must mitigate Public-Private Partnerships: Dealing with Affordability 32

33. Private sector (insurers/reinsurers) cover losses against all non-catastrophic events Public sector (state/federal) provides reinsurance against catastrophic losses Insurers pay a premium upfront if there is good data to estimate likelihood of losses (e.g. natural disasters) Federal government pays for catastrophic losses and recoups these payments from insurers over several years (e.g. TRIA) Well enforced building codes to reduce catastrophic losses Public-Private Partnerships: Catastrophic Protection 33

34. Encourages investment in cost-effective mitigation measures Avoids cancellation of policies by insured Reduces search costs to consumers Provides budget planning to insured by having stable premiums Reduces marketing costs of insurer Reduces variance by spreading risk over time Reduces insurance cycles due to catastrophic losses Multi-Year Insurance: Positive Features 34

35. Regulators allowing insurers to set risk-based prices Inability to change premiums or non-renew policies Sufficient protection against catastrophic losses through risk transfer instruments and public sector protection Systematic changes in risk (e.g. climate change) Multi-Year Insurance: Challenges 35

36. Modifying the National Flood Insurance Program 36 Premiums would reflect risk based on updated flood maps so that private insurers would have an incentive to market coverage. Means-tested vouchers would be provided by the public sector to those who undertook cost-effective mitigation measures. A multi-year insurance policy tied to the property would prevent policyholders from canceling their policies if they did not suffer losses for several years Reinsurance and risk-transfer instruments marketed by the private sector could cover a significant portion of the catastrophic losses from future floods. Federal reinsurance would provide insurers with protection against extreme losses.

37. Stretch time horizon on likelihood of disasters occurring Highlight expected benefits of loss reduction measures to key interested parties Tie loans and insurance to the property (not to the individual) through assumable mortgage contracts or via property taxes Examine role of multi-year insurance contracts tied to the property in encouraging investment in loss reduction measures 37 Encouraging Investment in Loss Reduction Measures

38. Insurance markets can help spread the risk of unavoidable disasters and offer incentives to mitigate risk. But they cannot work miracles, especially in LP-HC settings. Insurers can encourage mitigation by being allowed by regulators to set premiums that reflect risk and partnering with banks and financial institutions to provide long-term loans. The premium reduction will be greater than the annual cost of the loan for cost-effective measures. In this way insurers can return to their 19 th century roots. Conclusions 38

39. 39 The Challenges of Linking Flood Insurance with Adaptation Measures

40. Part I: Contrasting Ideal and Real Worlds of Insurance Chapter One: Purposes of this Book Chapter Two: An Introduction to Insurance in Practice and Theory Chapter Three: Anomalies and Rumors of Anomalies Chapter Four: Behavior Consistent with Benchmark Models Part II: Understanding Consumer and Insurer Behavior Chapter Five: Real World Complications Chapter Six: Why People Do or Do Not Demand Insurance Chapter Seven: Demand Anomalies Chapter Eight: Descriptive Models of Insurance Supply Chapter Nine: Anomalies on the Supply Side Part III: The Future of Insurance Chapter Ten: Design Principles for Insurance Chapter Eleven: Strategies for Dealing with Insurance-Related Anomalies Chapter Twelve: Innovations in Insurance Markets through Multi-Year Contracts Chapter Thirteen: Publicly-Provided Social Insurance Chapter Fourteen: A Framework for Prescriptive Recommendations 40 Insurance and Behavioral Economics: Improving Decisions in the Most Misunderstood Industry