Estimation of State-Dependent Utility Functions Using Survey Data
开发了一种计量经济学方法，利用调查数据中关于轻微健康风险的美元风险权衡来估计状态依赖效用函数的结构，发现大多数暂时性健康影响等同于收入下降。
Surveys of individual's risk-dollar tradeoffs illuminate not only the local tradeoff rates but also can be used to address more fundamental questions about the structure of utility functions.This largely unexplored empirical area is investigated by developing an econometric technique to estimate utility functions based on survey data on risk-dollar tradeoffs for minor health effects.The empirical tests indicate that for all but one of the temporary health effects considered, consumers treat injuries as tantamount to a drop in income, implying that the health impact does not alter the structure of the utility function in a fundamental way. I. IntroductionA LTHOUGH a considerable literature exists on the valuation of risks to life and health, its focus has largely been on risk-dollar tradeoffs for small changes in risk.In particular, the most frequent focus is on wage premiums for job risk based on hedonic models of wage determination.1 Available market data do not enable one to expand the scope of such inquiries to include issues such as the underlying shape of the utility function that generated these tradeoffs.However, the increased availability of survey data that provide detailed information on multiple risky decisions can potentially extend the range of issues considered.2In this paper, we indicate how these data can be used to extend the conventional hedonic methodology to estimate the overall structure of utility functions for adverse health effects.This paper utilizes survey data pertaining to four pairs of relatively minor consumer health risks.The essential feature of the data is that it provides information on multiple points on a constant expected utility locus so that one can estimate more than a local risk-dollar tradeoff.In particular, the paper develops a procedure for estimating, for each health outcome, two parameters of the state-dependent utility function that distinguish it from the utility function in good health.This estimation procedure is more general than that proposed in Viscusi and Evans (1990) since it focuses on more than just the relative marginal utility in good and ill health.The novelty of this estimation approach is best understood by comparing it to the more standard methodology based on statistical decision theory.The traditional approach begins by presenting subjects with a succession of hypothetical choices, for which they must give a probability that makes them indifferent between a pair of lotteries with extreme outcomes.This sequence of probabilities is then used to define a utility index.The approach presented here utilizes risk-dollar tradeoffs where subjects are presented with a series of choices and respondents are asked to provide a dollar amount that makes them indifferent between the lotteries.By construction, the series of points provided by the respondent lie along a constant expected utility locus.This paper provides a procedure for estimating the parameters of an explicitly defined utility function by using these risk-dollar tradeoffs.The structure of the estimation procedure introduced here is quite general and can be applied to any data set that contains information on points from a constant expected utility locus.Section II summarizes the alternative perspectives one could take on such health effects, treating them as monetary loss equivalents or as more fundamental changes in utility function structure.The empirical tests reported in section III indicate that for most of the temporary health impacts considered, the welfare effects are tantamount to monetary losses.Section IV explores the implications for the results of different measures of the implicit value of injury, and section V concludes the paper.