Consumers’ collision insurance decisions: a mental models approach to theory evaluation

Abstract

Using interviews with 74 drivers, we elicit and analyze how people think about collision insurance coverage and decide whether to buy coverage, and if so, what deductible level to carry. We compare respondents’ judgments and behaviors to predictions of three models: baseline expected utility (EU) theory, which predicts that insurance is an inferior good, meaning more wealthy people buy less; a modified EU model, which incorporates income constraints and suggests that property insurance is a normal good, meaning more wealthy people buy more; and a mental accounting model which predicts that consumers budget income across consumption categories. The results suggest they purchase insurance as a normal good, guided by a cognitive model that emphasizes budget constraints. Verbal reports reveal a desire to balance two conflicting goals in deductible decisions: keeping premiums ‘affordable’ and keeping deductible level ‘affordable.’ Thus, wealth does not distinguish people by risk aversion, but by ability to pay. In other words, the behavior of less wealthy people is not driven by greater risk aversion, but by their lesser ability to pay, both now and later. We find that a simple heuristic using only vehicle value accounts for most decisions of whether to purchase optional collision coverage: out of 45 respondents who did not have loans on their vehicles, 90% of those with vehicles worth more than $1000 carried collision coverage, while less than 30% of those with lower‐valued vehicles did.

Keywords:

• insurance
• risk aversion
• mental accounting
• decision‐making

Acknowledgments

This research was funded by the National Science Foundation (SES92‐10600) and Allstate Insurance Company. The authors wish to thank an anonymous reviewer for thoughtful comments on an earlier draft of this article. We also thank Claire Palmgren for interviewing and coding assistance.

Notes

1. According to a PA Insurance Department representative at the time of the study, most insurance companies had a $1000 deductible level that they did not actively market. As a result, we treat a $500 as the largest generally available deductive.

2. Limited tort allows lower premiums in exchange for giving up the right to sue for other than out‐of‐pocket expenses, such as pain and suffering, when suffering injuries that are not serious.

3. Reported income and education distributions were somewhat higher than in Pennsylvania generally, likely an artifact of interviewing in a major city and sampling some university graduate students. The percentage of homes with drivers under age 25 seems high, likely an artifact of sampling parents from local schools.

4. This measure was added after the first 12 interviews were completed.

5. Four respondents who provided total policy cost could not determine the cost of collision coverage from reading their policies.

6. The judged expected benefit for five respondents with collision coverage was $4 or less; we assume that their R is at least 5, since any premium difference of $20 or more results in this.

7. Six subjects who said that collision coverage is required (or did not know if it was) were not asked this. Two who said banks require coverage on financed vehicles were mistakenly not asked this question. There are no responses for two subjects whose interviews were accidentally not recorded.

8. Vehicle age might be a surrogate for vehicle value but is listed separately because 28% of the respondents mentioned both factors.