Mortality Risk Valuation in Policy Assessment

Annex C. Descriptive statistics of VSL and coded variables

The four tables below provide full descriptive statistics for the VSL estimates reported in Section 3.3 of Chapter 3 on the estimate and study levels, including (Table A ‎C.1 and Table A ‎C.3) and excluding far-outs (Table A ‎C.2 and Table A ‎C.4).

Descriptive statistics of VSL estimates at the estimate level

Table A ‎C.1. Summary statistics for VSL estimates at the estimate level

VSL estimates in USD million, including far-outs

Datasets	N Estimates	Mean	SD	Q25	Q50	Q75	Min	Max	N Studies
New SP	1027	60.8	1627.9	1.3	5.1	10.9	-0.9	52162	99
New RP	1470	8.6	19.6	0.8	4.2	11.2	-89.6	378	55
Old SP	917	12.9	31.3	1.3	4.2	11.0	0.0	307.2	72
Old RP	694	15.3	21.9	4.5	12.0	21.0	-83.0	173	51
Subtotal SP	1944	38.2	1183.3	1.3	4.7	11.0	-0.9	52162	171
Subtotal RP	2164	10.8	20.6	1.1	6.3	14.1	-89.6	378	106
Subtotal New	2497	30.1	1044.1	0.9	4.6	11.2	-89.6	52162	154
All	4108	23.8	814.2	1.2	5.3	12.7	-89.6	52162	277

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2008-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Table A ‎C.2. Summary statistics for VSL estimates at the estimate level

VSL estimates in USD million, far-outs excluded

Datasets	N Estimates	Mean	SD	Q25	Q50	Q75	Min	Max	N Studies	% Far-outs
New SP	1013	7.3	7.8	1.2	4.9	10.6	-0.9	46.9	99	1.4%
New RP	1436	6.5	7.9	0.8	4.0	10.9	-21.7	45.8	55	2.3%
Old SP	861	6.8	8.0	1.2	3.8	9.0	0.0	45.9	71	6.1%
Old RP	653	12.9	11.9	4.2	11.4	19.0	-29.6	47.2	51	5.9%
Subtotal SP	1874	7.1	7.9	1.2	4.4	9.9	-0.9	46.9	170	3.6%
Subtotal RP	2089	8.5	9.8	1.0	5.8	13.3	-29.6	47.2	106	3.5%
Subtotal New	2449	6.9	7.9	0.9	4.4	10.8	-21.7	46.9	154	1.9%
ALL	3963	7.8	9.0	1.1	5.0	12.0	-29.6	47.2	276	3.5%

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Descriptive statistics of VSL estimates at the study level

Table A ‎C.3. Summary statistics for VSL estimates at the study level

VSL estimates in USD million, including far-outs

Datasets	N Studies	Mean	SD	Q25	Q50	Q75	Min	Max
New SP	99	26.1	177.5	1.0	4.6	7.8	0.0	1751
New RP	55	11.1	16.2	1.7	6.6	13.8	0.1	109
Old SP	72	11.6	24.5	2.0	4.6	9.6	0.0	163
Old RP	51	12.1	13.7	5.1	10.3	17.1	-32.6	68
Subtotal SP	171	20.0	135.9	1.3	4.6	8.4	0.0	1751
Subtotal RP	106	11.6	15.0	3.2	8.6	15.7	-32.6	109
Subtotal New	154	20.8	142.5	1.3	5.1	9.2	0.0	1751
ALL	277	16.8	107.1	1.7	5.7	11.0	-32.6	1751

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Table A ‎C.4. Summary statistics for VSL estimates at the study level

VSL estimates in USD million, far-outs excluded

Datasets	N Estimates	Mean	SD	Q25	Q50	Q75	Min	Max	% Far-outs
New SP	99	5.3	5.6	1.0	4.6	7.7	0.00	29.6	0.0%
New RP	55	8.1	7.1	1.7	6.6	12.8	0.05	37.4	0.0%
Old SP	71	6.4	5.9	1.9	4.6	9.5	0.02	22.1	1.4%
Old RP	51	11.3	6.9	6.0	10.3	17.1	0.73	30.0	0.0%
Subtotal SP	170	5.8	5.7	1.3	4.6	8.1	0.00	29.6	0.6%
Subtotal RP	106	9.6	7.1	3.7	8.6	14.0	0.05	37.4	0.0%
Subtotal New	154	6.3	6.3	1.3	5.0	8.7	0.00	37.4	0.0%
ALL	276	7.2	6.6	1.8	5.7	10.4	0.00	37.4	0.4%

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Cumulative distribution functions (CDF) of VSL estimates

A CDF is defined as the probability (Y-axis) that a random variable X, in this case a VSL estimate, will take a value less than or equal to x (X-axis).

Figure A ‎C.1 and Figure A ‎C.2 show the cumulative distribution functions for the four datasets and specific RP and SP methods, respectively. As can be seen from Figure A ‎C.1, the distributions of Old SP, New SP and New RP follow each other relatively closely, while the cumulative distribution function (CDF) of the Old RP data has a more right-skewed distribution. Figure A ‎C.2 shows that CM studies and CM and HW have relatively different distributions, while the CDFs of the SP methods (CE and CV) are located between the two RP methods.

Figure A ‎C.1. Cumulative distribution function (CDF) of VSL estimates per dataset

VSL estimates in USD million, far-outs excluded

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Figure A ‎C.2. Cumulative distribution function (CDF) for VSL estimates per specific methods

VSL estimates in USD million, far-outs excluded

Log-distribution of VSL for the different datasets

Figure A ‎C.3 shows the statistical distributions for the natural logarithm of VSL estimates for the four datasets. It shows that the Old RP data follows a distribution that is somewhat different compared to the other three datasets.

Figure A ‎C.3. Statistical distribution of log(VSL), by data source

Note: Study Level = Weighted by the inverse of the number of estimates from each study. Outliers (“far-outs”) are excluded, defined as $\notin [Q25-3(Q75-Q25), Q75+3(Q75-Q25\left)\right]$. VSL estimates are expressed in 2022 USD. New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.

Descriptive statistics of coded variables

Table A ‎C.5 presents the mean and number of estimates (N) for coded variables for the four datasets. These variables are divided into six categories:

• Article information and estimates

• Country characteristics

• Risk types by cause (including cancer if mentioned in the study, and acute, chronic or mixed risk)

• Elicitation methods (how data is derived, specific SP or RP method)

• Sample data characteristics

• Methodological characteristics (some of which may be related to quality of the study)

Table A ‎C.5. Coded variables, mean and number of observations (N) for the four datasets

	All data		New RP and SP		New SP		New RP		Old SP		Old RP
	Mean	N	Mean	N	Mean	N	Mean	N	Mean	N	Mean	N
Article information and estimates
Journal	0.69	4108	0.71	2 517	0.74	1027	0.69	1470	0.40	917	1.00	694
Publication Year	2010	4108	2016	2 517	2015	1027	2017	1470	2004	917	1998	694
Year of VSL estimate	2004	4108	2009	2 517	2010	1027	2009	1470	2000	917	1989	694
VSL estimates (2022 USD million)	23.76	4108	30.1	2 517	60.81	1027	8.55	1470	12.91	917	15.30	694
Standard errors of VSL estimates (2022 USD million)	14.62	4108	17.14	2 517	31.15	1027	7.32	1470	8.02	917	14.25	694
Country characteristics
G7	0.61	4108	0.61	2 517	0.40	1027	0.75	1470	0.36	917	0.93	694
EU	0.19	4108	0.19	2 517	0.41	1027	0.04	1470	0.30	917	0.01	694
EEA	0.20	4108	0.2	2 517	0.43	1027	0.04	1470	0.33	917	0.01	694
OECD	0.81	4108	0.79	2 517	0.77	1027	0.81	1470	0.72	917	0.97	694
USA	0.40	4108	0.43	2 517	0.13	1027	0.64	1470	0.07	917	0.75	694
WB (High-income countries)	0.83	4108	0.83	2 517	0.76	1027	0.87	1470	0.73	917	0.97	694
WB (Upper-middle income countries)	0.05	4108	0.05	2 517	0.08	1027	0.04	1470	0.07	917	0.01	694
WB (Low-middle income countries)	0.09	4108	0.08	2 517	0.16	1027	0.04	1470	0.18	917	0.00	694
WB (Low-income countries)	0.03	4108	0.03	2 517	0.00	1027	0.05	1470	0.02	917	0.01	694
GDP per capita (thousands of 2022 USD)	47.0	4108	50.62	2 517	46.56	1027	53.46	1470	36.67	917	47.60	694
Risk types based on causes
Climate	0.03	3413	0.04	2 516	0.08	1026	0.02	1470	0.00	917	.	.
Crime	0.00	3413	0	2 516	0.01	1026	0.00	1470	0.00	917	.	.
Disaster	0.01	3413	0.01	2 516	0.01	1026	0.01	1470	0.00	917	.	.
Environment	0.08	3413	0.05	2 516	0.06	1026	0.04	1470	0.17	917	.	.
Health	0.31	3413	0.24	2 516	0.49	1026	0.06	1470	0.49	917	.	.
Job	0.26	3413	0.36	2 516	0.00	1026	0.61	1470	0.00	917	.	.
Military	0.08	3413	0.12	2 516	0.00	1026	0.20	1470	0.00	917	.	.
Natural Disaster	0.04	3413	0.06	2 516	0.14	1026	0.00	1470	0.00	917	.	.
Not Specified	0.00	3413	0	2 516	0.00	1026	0.00	1470	0.00	917	.	.
Suicide	0.00	3413	0	2 516	0.00	1026	0.00	1470	0.00	917	.	.
Transportation	0.16	3413	0.1	2 516	0.16	1026	0.05	1470	0.35	917	.	.
Virus	0.02	3413	0.03	2 516	0.05	1026	0.01	1470	0.00	917	.	.
Cancer (if addressed in the paper)
Cancer	0.22	3326	0.26	2 429	0.47	1003	0.11	1406	0.12	917	.	.
Risk acuteness
Acute	0.39	1874	0.39	1 894	0.66	498	0.29	1376	.	.	.	.
Chronic	0.13	1874	0.13	1 894	0.32	498	0.07	1376	.	.	.	.
Mixed/job-related	0.47	1874	0.47	1 894	0.00	498	0.64	1376	.	.	.	.
Elicitation methods
Consumer Market valuation method	0.07	4108	0.11	2 517	0.00	1027	0.19	1470	0.00	917	0.00	694
Hedonic Wage method	0.46	4108	0.48	2 517	0.00	1027	0.81	1470	0.00	917	1.00	694
Contingent Valuation method	0.27	4108	0.2	2 517	0.49	1027	0.00	1470	0.67	917	0.00	694
Choice Experiment	0.15	4108	0.2	2 517	0.48	1027	0.00	1470	0.13	917	0.00	694
Other SP method	0.05	4108	0.01	2 517	0.02	1027	0.00	1470	0.20	917	0.00	694
Sample data characteristics
Average sample income (USD thousand)	44.3	2676	45.37	1 457	53.3	563	40.3	879	35.5	540	49.0	694
Disposable household income	0.31	2660	0.34	1 441	0.71	492	0.15	934	0.60	540	0.00	694
Nationwide	0.63	3297	0.71	2 400	0.4	910	0.90	1470	0.41	917	.	.
General Population	0.66	3264	0.56	2 367	0.86	877	0.39	1470	0.92	917	.	.
Representative sample	0.57	3199	0.58	2 302	0.74	812	0.50	1470	0.55	917	.	.
Sample Average Age	41.6	2325	39.53	2 372	44.6	548	37.1	1165	47.4	612	.	.
Only male	0.20	4108	0.19	2 372	0.00	1027	0.32	1470	0.01	917	0.47	694
Only female	0.06	4108	0.06	1 728	0.04	1027	0.08	1470	0.03	917	0.09	694
Methodological characteristics
% of SP studies with representative sample	0.64	1729		829	0.74	812	.	.	0.55	917	.	.
% of SP studies including baseline risk of respondents	0.65	1919		1 019	0.67	1002	.	.	0.62	917	.	.
% of SP studies focusing on scope sensitivity	0.66	1925		1 025	0.61	1008	.	.	0.72	917	.	.
% of SP studies using visual aids	0.71	1943		1 043	0.65	1026	.	.	0.78	917	.	.
% of RP studies using Census of fatal occupational injuries (CFOI) data	0.48	2164		801	.	.	0.49	1470	.	.	0.46	694
% of RP studies using IV estimates	0.14	2123		1 473	.	.	0.16	1429	.	.	0.10	694
% of RP studies controlling for non-fatal injury risks	0.35	2164		1 432	.	.	0.29	1470	.	.	0.47	694
% of Hedonic wage RP studies	0.10	2164		1 473	.	.	0.10	1470	.	.	0.12	694

Note: New SP and New RP studies refer to Stated Preferences and Revealed Preferences studies published in the period 2009-2023. Old SP and Old RP studies refer to Stated Preferences and Revealed preferences studies published in the period 1970-2008.