Academic journal article Economic Inquiry

Heterogeneity in State-Dependent Utility: Evidence from Strategic Surveys

Academic journal article Economic Inquiry

Heterogeneity in State-Dependent Utility: Evidence from Strategic Surveys

Article excerpt

I. INTRODUCTION

An implication of standard life-cycle models is that optimizing individuals will equate the expected marginal utility of consumption across states and periods if insurance is available at actuarially fair rates. Yet a number of empirical regularities--such as consumption declines at older ages or the apparent underinsurance of many consumption risks, including long-term care (e.g., Brown and Finkelstein 2009, 2011) and longevity risk (e.g., Benartzi, Previtero, and Thaler 2011; Brown 2001)--have led some to suggest that the marginal utility of consumption may depend on one's health status. Direct empirical evidence of state dependence is rare, due in large part to the difficulty in measuring how a consumer values, ex ante, consumption in future, uncertain states. (1) In theory, one could infer state dependence from observing insurance purchase decisions, but given that most insurance markets suffer from multiple market failures (e.g., adverse selection, moral hazard, imperfect competition, public policy-induced demand distortions, etc.), it would be hazardous to measure state dependence by observing behavior in incomplete insurance markets.

In this article, we propose a novel mechanism to assess the extent of state-dependent utility using a "strategic survey" approach, in the spirit of Ameriks et al. (2011, 2015a, 2015b). They note that a key advantage of surveys is that one can design questions that "represent natural thought experiments concerning behaviors in contingencies selected for high information content."

We designed questions to elicit how an individual would value marginal consumption in healthy versus disabled states, being careful to isolate the state dependence from other factors--such as wealth shocks--that would normally confound empirical inference in "real world" data. These survey questions, which we field in the American Life Panel (ALP), ask individuals to allocate resources across various health states, while holding fixed the household's health expenditures. A second advantage of using a strategic survey is that it allows an examination of within-person differences in resource allocations based on expected future health status, and thus also addresses concerns about unobservable factors that would typically afflict studies that rely on cross-sectional variation in responses to shocks. A third advantage is that we can examine responses for the same person across different types of health shocks to assess whether the behavior varies depending on the type of illness. Partially offsetting these advantages is that stated preferences in response to hypothetical questions are an imperfect proxy for true underlying preferences. Thus, although we consider our approach informative as to the presence and relative size of state dependence, we urge caution against using our findings to precisely parameterize state dependence. We hope that other researchers will refine and advance this methodology to a point where such parameterization becomes possible.

For several reasons, our scenarios focus on disabilities. First, we hypothesize that a health shock that leads to a permanent disability is more likely to be associated with changes in preferences than is a temporary shock. Second, disabilities are an interesting setting to examine the possibility that marginal utility could rise or fall with the onset of a disability (Finkelstein, Luttmer, and Notowidigdo 2009). For example, a severe physical injury that requires expensive modifications to one's life-style might increase one's marginal utility whereas the onset of cognitive impairment that diminishes one's enjoyment of consumption goods might decrease one's marginal utility. Thus, this is an interesting context to explore heterogeneity, both across the population and across different types of disabilities. Finally, this is a highly policy-relevant context. A significant risk that people face during their lifetimes is that of becoming disabled--either early in life when the disability decreases one's earnings capacity, or later in life when chronic/custodial care in a nursing facility may be needed--yet most Americans have little private insurance to protect against these risks, even though public insurance is incomplete. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.