Rates of Time Preferences for Saving Lives in the Hazardous Waste Site Context

Abstract

SummaryWhat is the rate at which people discount future lives saved? The answer to this question has important implications when comparing policies on the grounds of cost per life saved, especially in the context of hazardous waste site remediation, where risk reductions may occur at different times, depending on the permanence of the remedy. We estimate this rate by asking a sample of Italian residents to choose between saving 100 lives now and X lives in T years, where both X and T are varied to the respondents. Assuming constant exponential discounting, the responses to these questions imply a rate of time preference for saving lives of 12%. There is little evidence that this rate is systematically associated with observable individual characteristics of the respondent. There is, however, strong evidence that it declines with the time horizon when the lives would be saved, ranging from 16% for T=10 to less than 4% for T³40. We fit a hyperbolic discount model, finding that it yields a similar value of the discount function for T=10 (the shortest horizon we used in the survey), and that it discounts the future less heavily than the regular exponential discounting model for longer time horizon. We apply our estimated discount functions to two alternate remedial plans for a heavily contaminated area in Italy, and find that-due to the high estimated discount rates-the less permanent solution is found to be more cost-effective.

Evidence of variable discount rates and non-standard discounting in mortality risk valuation

“…Non-exponential discounting has been explored for changes in expected numbers of fatalities prevented. Examples include Cropper et al (1992), reanalysed in Henderson and Bateman (1995) with reference to the implications of real policy decisions; Cairns and Van der Pol (1997a), Cairns and Van Der Pol (1997b), and Alberini et al (2009), all of whom find support for non-constant discounting of lives saved. However, these studies elicit preferences over different numbers of fatalities, and so do not reveal private time preference rates for one's own future fatality risks.…”