New risk rates, inter-industry differentials and the magnitude of VSL estimates

Abstract

The Census of Fatal Occupational Injury (CFOI) provides data for creating objective measures of workplace risk used in estimations of wage-risk premia for value of statistical life (VSL) calculations. This relatively new data set enables a more theoretically defensible measure for use in hedonic wage equations. However, constructing these rates from the CFOI data necessarily involves creating an industry-occupation matrix defining the “jobs,” deciding whether or not to include the self-employed, and selecting a denominator. These choices in the construction of the risk measure alone, as shown here, result in variations of VSL estimates ranging from $8 million to $18 million. Further, risk measures based on the CFOI data, regardless of construction, are sensitive to simple changes in the specification for the hedonic wage equation. In particular, fixed effects describing the industry in which a worker is employed, as well as the worker’s occupation, are primary influences on the magnitude of the VSL estimates.

Valuing Reductions in Fatal Illness Risks: Implications of Recent Research

“…The highlighted VSLs range from $2.1 million to $20.8 million, with most between $6.8 million and $12.0 million. The lowest values ($2.1 million to $4.1 million) are from a working paper that experiments with the use of Occupational Safety and Health Administration inspection data (Lee and Taylor ), while the highest ($9.2 million to $20.8 million) are from a paper where the author notes, ‘[t]he intent of this study is not to posit a particular value for the VSL; rather, it is to demonstrate how the construction of the fatal risk rate measure impacts the magnitude of the VSL estimate’ (Scotton , p. 65).…”

“…13 Rather than selecting a single estimate from each study, Viscusi 10 Previous meta-analyses of the wage-risk literature have been criticized in part for not applying carefully developed, explicit criteria for selecting studies for inclusion (EPA 2006, Cropper et al 2007). 11 The 10 excluded studies are Jennings and Kinderman (2003), Evans and Smith (2008), Viscusi and Hersch (2008), Evans and Schaur (2010), Kniesner, Viscusi, and Ziliak (2010), Scotton andTaylor (2011), Lavetti (2012), Kniesner, Viscusi, Woock, and Ziliak (2012), DeLeire, Khan, and Timmins (2013), and Kniesner, Viscusi, and Ziliak (2014). We generally exclude these studies because they address only a subset of workers and/or do not control for occupation as well as industry; one (DeLeire et al) does not report a full sample VSL.…”

“…12 As noted earlier, the Kniesner et al study is not included in Table II because it addresses only men but has the advantage of relying on longitudinal data. 13 The 17 studies include Aldy and Viscusi (2008), Evans and Schaur (2010), Hersch and Viscusi (2010), Kniesner and Viscusi (2005), Kniesner et al (2012), Kniesner, Viscusi, and Ziliak (2006, 2014, Kochi and Taylor (2011), Scotton (2013), Scotton and Taylor (2011), Viscusi (2003, 2004, Viscusi and Aldy (2007), Viscusi and Hersch (2008), and Viscusi and Philip (2014). Some of these studies do not meet all of our selection criteria, but the controls that Viscusi (2015) included for study characteristics address many of the concerns that led to their exclusion.…”

The value of statistical life for adults and children: Comparisons of the contingent valuation and chained approaches

“…Using wages as an estimate of a VSL wrongly equates prices and values (Rice, 2015). Revealed preference techniques using either wage premiums or expenditure on safety equipment (Bellavance et al, 2009) require strong assumptions regarding the information available on the risk associated with particular jobs or behaviours (Dolan et al, 2008), and values are very sensitive to the exact nature of risk estimation (Scotton, 2013).…”

Effect of Income Heterogeneity on Valuation of Mortality Risk in Taiwan: An Application of Unconditional Quantile Regression Method

“…On the contrary, the HWM estimates VSL while using the wage data of the labor market, which are often readily available, thereby resulting in a low operating cost. Thus, HWM is currently the dominant approach in estimating VSL, which has been extensively used [18,19,20,21,22,23,24,25,26,27,28,29].…”