We propose a model of narrow framing in insurance and test it using data from a new module we designed and fielded in the Health and Retirement Study. We show that respondents subject to narrow framing are substantially less likely to buy long-term care insurance than average. This effect is distinct from, and much larger than, the effects of cautiousness, risk aversion, or adverse selection, and it offers a new explanation for why people underinsure their later-life care needs.
