Skip to main content

Table 1 Income elasticities of the WTP for a reduction in mortality risks—results of meta-analyses

From: Differential discounting in the economic evaluation of healthcare programs

Author and year

Type of included studies and data base [(a): number of VSL estimates; (b): number of studies; (c): number of countries]

Results

Liu et al. 1997 [55]

WR; (a) 17;(b) 17; (c) 3

0.53, ns

Miller 2000 [56]

WR, PR, SP; (a) 68; (b) 27; (c) 13

0.85–1.00

Bowland et al. 2001 [57]

WR; (a) 33; (b) na; (c) na

1.52–2.27

Mrozek et al. 2002 [58]

WR; (a) 203; (b) 33; (c) na

0.37–0.49

De Blaeij et al. 2003 [59]

WR, SP, CPLS; (a) 95; (b) 30; (c) 10

0.50

Viscusi et al. 2003 [60]

WR; (a) 49; (b) 49; (c) 11

0.46–0.60

Kluve et al. 2008 [61]

WR, PR, SP; (a) 42; (b) 37; (c) 10

− 0.26, ns

Bellavance et al. 2009 [62]

WR; (a) 32; (b) 32; (c) 8

0.84–1.08

Lindhjelm et al. 2011 [63]

SP; (a) 856; (b) 76; (c) 24

1.17–1.31

Reduced data seta: (a) 405; (b) na; (c) na

0.77–0.88

Exclusion of VSL estimates according to authors’ recommendations: (a) 350; (b) na; (c) na

0.74–0.89

Passed internal or external scope test: (a) 297; (b) na; (c) na

0.69–0.74

Passed internal and external scope test: (a) 79; (b) na; (c) na

0.25–0.30

Studies with identical questionnaire: (a) 169; (b) na; (c) na

0.30–0.44

Doucouliagos et al. 2012 [64]b

WR; (a) 32; (b) 32; (c) 8; re-analysis of the dataset used in [60]

0.20–0.38

Doucouliagos et al. 2014 [54]b

WR; (a) 101; (b) 14 meta-analyses; (c) na

0.25–0.63

Viscusi 2015 [65]b

WR; (a) 550; (b) 17; (c) 1

0.53–0.59

Viscusi et al. 2017 [66]b

WR; (a) 953); (b) 68; (c) 14

0.78

Majumder et al. 2017 [67]

WR; (a) 34; (b) 30; (c) na

0.26

Viscusi 2018 [68]b

WR; (a) 1025; (b) 68; (c) 14

0.53–0.85

Masterman et al. 2018 [69]b

SP; (a) 1145; (b) 85; (c) 27

0.69

  1. WTP willingness to pay, VSL value of a statistical life, WR wage risk, PR product risk, SP stated-preference, CPLS cost per life saved, ns not significant, na not available
  2. aVSL estimates were excluded if (a) no value for the risk change has been reported, (b) a subsample was smaller than 100 observations or the main survey sample had less than 200 observations, (c) the sample was not representative of a broad population
  3. bCovariates of meta-regression include a measure of precision of the VSL estimate (usually its standard error) to control for publication selection or reporting bias