Insurance Project Workshop

September 17, 2011
Kenneth Froot of Harvard Business School and Howard Kunreuther of the University of Pennsylvania, Organizers

Levon Barseghyan, Francesca Molinari, and Edward O'Donoghue, Cornell University, and Joshua Teitelbaum, Georgetown University
The Nature of Risk Preferences: Evidence from Insurance Choices

Barseghyan, Molinari, O'Donoghue, and Teitelbaum use data on insurance deductible choices to estimate a structural model of risky choice that permits "standard" risk aversion, loss aversion, and probability weighting. They show that loss aversion and probability weighting -- although not separately identified without strong parametric assumptions -- both imply a distortion of probabilities. They demonstrate that such probability distortions are identified. They further find that probability distortions -- in the form of substantial over-weighting of claim probabilities -- play an important role in explaining the aversion to risk that is manifested in deductible choices. Once the researchers allow for probability distortions, standard risk aversion is relatively small.

Steven Shavell, Harvard University and NBER
A General Rationale for a Governmental Role in the Relief of Large Risks

The government traditionally furnishes relief against large risks, notably disaster assistance, subsidy of flood insurance, and provision of unemployment insurance. Is this justified? Shavell considers a general rationale for the government to relieve significant risks that applies even when private contracting to share risks is perfect (unimpeded by transaction costs, asymmetric information, or other sources of market failure). The rationale, in essence, is that the optimal private sharing of very large risks will not result in complete coverage against them. Therefore, when the risks eventuate, the marginal utility to individuals of relief from the government will be high in a relative sense and may exceed the marginal value to them of public goods. Consequently, social welfare may be raised if the government reduces public goods expenditures and directs these freed resources toward individuals who have suffered losses.

Erwann Michel-Kerjan, University of Pennsylvania, and Jacqueline Wolkman Wise, Temple University
The Risk of Ever-Growing Disaster Relief Expectations

Michel-Kerjan and Wolkman Wise analyze the risk of the vicious cycle that is created by individuals' expectations that the federal government will always come to their rescue in the aftermath of a disaster, and more so, that they can always expect at least as much relief as was given to victims of previous catastrophes. As a result, elected officials might feel that they have no alternative but to provide even more relief the next time if they want to benefit politically from their actions. This cycle is happening in the United States. While previous work has been done on the disincentive that relief creates for individuals to purchase proper insurance coverage (Buchanan, 1975; Coate, 1995), the authors propose here a more systematic political economic analysis of this escalating need for ever more relief and its impact on government revenues. They show that a government will benefit in the long run from simply capping relief payouts. However, this policy will prove more challenging to implement the longer one waits. Eventually, it might be impossible politically for any new administration to do it. The researchers show that the dominating strategy for limiting government liability is to actually use two policy tools simultaneously: a publicly-known cap on relief payouts combined with the right level of public insurance subsidies (or, alternatively, insurance vouchers).

Daniel Bauer and George Zanjani, Georgia State University
The Marginal Cost of Risk, Risk Measures, and Capital Allocation

The Euler (or gradient) allocation technique defines a financial institution's marginal cost of risk exposure by calculating the gradient of a risk measure evaluated at the institution's current portfolio position. However, the technique relies on an arbitrary selection of a risk measure. Bauer and Zanjani reverse the sequence of this approach by calculating the marginal costs of risk exposures for a profit maximizing financial institution with risk averse counter-parties, and then identifying a closed-form solution for the risk measure whose gradient delivers the correct marginal costs. They compare the properties of allocations derived in this manner to those obtained through application of the Euler technique to Expected Shortfall (ES), showing that ES generally yields economically inefficient allocations.

Neil Doherty, University of Pennsylvania, and Christian Laux and Alexander Muermann, Vienna University of Economics and Business
Insuring Non-Verifiable Losses and the Role of Intermediaries

Doherty, Laux, and Muermann analyze optimal risk sharing arrangements when losses are observable by policyholders and insurers but cannot be verified. The optimal contract to insure individual losses can be implemented through a standard insurance contract with a deductible where the policyholder bears all losses below the deductible amount and an upper limit restricts the maximum payment to the policyholder. For a group of policyholders, it is optimal to choose contracts with individual deductibles and a joint upper limit. Insurance brokers can play an important role in implementing these contracts.

Santosh Anagol, University of Pennsylvania, and Shawn A. Cole and Shayak Sarkar, Harvard University
Bad Advice: Explaining the Persistence of Whole Life Insurance

Anagol, Cole, and Sarkar conduct a series of field experiments to evaluate two competing views of the role of financial intermediaries in providing product recommendations to potentially uninformed consumers. The first view argues that financial intermediaries may provide valuable product education, helping consumers to decide which of many complicated products is right for them. Even if commissions influence intermediary recommendations, consumers are sufficiently sophisticated to discount advice. The second, more sinister, view, argues that intermediaries recommend and sell products that maximize the agents well-being, with little regard to the need of the customer. Audit studies in the Indian insurance market find evidence consistent with the second view: agents recommend a product that provides them high commissions, although it is strictly dominated by alternative products. Consumers demonstrating lower levels of sophistication are more likely to be offered the wrong product. Agents also appear to cater to the initial preferences of consumers even when those initial preferences are for products that are not suitable for the consumer. Finally, the authors exploit a natural experiment that occurred during their audits to test how disclosure requirements affect product recommendations. They find that requiring disclosure of commission levels makes agents less likely to recommend the product for which disclosure is required.

Alexander Muermann, Vienna University of Economics and Business
Asymmetric Information in Automobile Insurance: New Evidence from Telematic Data

Muermann provides novel insights into the effects of private information in automobile insurance. His analysis is based on telematic data of insured cars, including detailed information about driving behavior that is unobservable to the insurance company. He finds direct evidence that private information about driving behavior has significant and counteracting effects on the choice of third-party liability and first-party insurance coverage. Yet, tests based on the residual correlation between the level of insurance coverage and risk spuriously allude to different interpretations. These results suggest overlapping and offsetting effects of private information based on risk preferences and driving behavior.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us