Report on the EPA Work Group on VSL Meta-Analyses (2006)
Paper Number: EE-0494
Document Date: 07/25/2006
Author(s): Allen, Eilaine; Becker, Betsy Jane; Berlin, Jesse A.; Morton, Sally C.; Olkin, Ingram
Rindskopf, David; Sampson, Allan R.; Wilson, David B.
Subject Area(s): Economic Analysis, Benefits Valuation; Revealed Preferences, Stated Preferences, Value of Statistical Life, Mortality Valuation, Meta-Analysis
Keywords: Economic Analysis, Benefits Valuation; Revealed Preferences, Stated Preferences, Value of Statistical Life, , Mortality Valuation, Meta-Analysis
Abstract:
In December 2005 a work group convened by the Environmental Protection Agency met to discuss the meta-analysis of estimates of the value of statistical life (VSL) and to examine three existing meta-analyses. The following report contains an analysis of the use of meta-analytic procedures to determine an estimate of VSL. Many detailed issues are covered in this report, but several general comments are also highlighted.
Whereas meta-analysis is a reasonable analytical approach to use in obtaining information about VSL, the existing meta-analyses should not be combined in this effort. The key issues the work group uncovered include a high degree of heterogeneity (inconsistency) in the VSL estimates and dependencies (that could not be accounted for) stemming from inclusion of multiple estimates derived from the same underlying data; these issues preclude relying on these meta-analyses as a source of a final VSL estimate.
The work group recommends that in future meta-analyses results of contingent valuation and hedonic wage studies be analyzed separately. A basic consideration for the EPA is the determination of whether a single universal VSL value is applicable to all relevant subpopulations, or whether multiple VSL values should be provided for subpopulations. In particular, meta-analytic methods provide a variety of ways that could be used to characterize a population of VSL values.
This paper is part of the Environmental Economics Research Inventory.