|Person's life||$2,000,000||The Value of a Statistical Life: A Critical Review of Market Estimates [free] tables 2 and 3 review 40 estimates of the value of statistical life. The vast majority of the results fall in the range US2000$1-10m. Taking the geometric mean of this range, and then adding 30% inflation from 2000-2011, gives an estimated value of a statistical life of $4,000,0000. However the vast majority of these estimates are based on labor market analysis, and the non-labor market estimates are significantly lower. So, for the time being we conservatively adopt the value of a life of $2,000,000. Some organizations, such as the Copenhagen Consensus Center, take the view that the value of a life of someone in the developing world is worth less, than someone in the developed world, by assigning the statistical value of life as 2-4 times annual income. This makes sense for local planning purposes, the value placed on life is lower, but it doesn't make sense for global planning purposes, people are more than what they produce, and the life of an Australian isn't equivalent to that of fifty Tanzanians.|
|Child's life||$2,500,000||The value of a life assumes a person of average age. A child will live twice as long and therefore their life should be more valuable. We follow the WHO DALY definition in discounting the future at 3% per year. The value of a child's life then is the value of a person's life * (1 + 97% ^ 69/2), where 69 is the HALE (Healthy Life Expectancy) of (67.0 + 71.0) / 2 from WHOSIS HALE for U.S. 2002.|
|DALY (Disability Adjusted Life Year)||$90,000||Value of one year at present value of person's life discounted at 3% over remaining HALE half life. ie. Value of person's life = DALY value * (1 + 97% + 97%^2 + .. 97%^(HALE/2-1)).|
|Extreme pain and suffering per day||$200||When pain is so severe and the outlook so prolonged that people choose euthanasia or suicide they are valuing their life as less than worthless. This means the pain imposes a cost greater than the value of a DALY / 365, or $140, but since suicide is always an option it can't be that much greater.|
|Individual lifted out of extreme poverty||$400,000||In high income countries we are willing to pay perhaps $5-10k/year to keep an individual from extreme poverty through welfare. Capitalizing this cost over 40 years gives $200k-400k. We thus value lifting an individual out of extreme poverty as worth at least this much.|
|Primary education||$50,000 - $500,000||Presumably education represents value for money, or we wouldn't fund it. This then allows us to compute a lower bound on the value of education. In the US the cost of education per pupil per year for primary and secondary education is $10,441 in 2007-2008. For 6 years this comes out to $62k, possibly slightly less because primary education is slightly cheaper than secondary education, let us say $50k. This represents a lower bound on the value we place on primary education. For an upper bound we employ another reasonableness argument. It seems highly unlikely that the value of education (at least at the margin) exceeds the US GDP, indeed even 50% of GDP seems quite excessive. Let us roughly assume that the upper bound value of education is 1/3 primary, 1/3 secondary, and 1/3 tertiary of 50% US GDP. US GDP was $14,369b in 2008. This gives $2,400b as an upper bound on the value of primary, secondary, or tertiary education in the US. Expenditure on primary and secondary education totaled $597b in 2007-2008, let us assume again this represents roughly $250b primary. The upper bound is then 9.59 times the cost, or $480k.|
|Secondary education||$70,000 - $500,000||Similar arguments apply as to primary education. This time we assume secondary education is slightly more expensive than primary education, and costs $70k for 6 years. This is our lower bound. For our upper bound we divide $2,400b by $350b and multiply by the cost to once again get $480k.|
|Tertiary education||$200,000 - $700,000||Not everyone in the US completes tertiary education, and data on the resulting wage differential allows us estimate the value of tertiary education. BLS median weekly earnings (April 19, 2011) for high school graduates $633 versus Bachelor's degree or better $1150. $27k annual income differential capitalized over 20 years is $540k. This estimate is far from perfect. It will be an overestimate since those that complete college are more likely to be stronger academically to begin with. It will be an underestimate since it doesn't include any societal benefits that might accrue. And it will be an overestimate because it ignores crowding out effects incurred by non-college graduates. Now, performing the same methodology as we performed for primary education. $40,476/year for tuition, room, and board at private non-profit(?) college in US (state university costs are subsidized); 4 year Bachelor degree takes 5 years to complete to incorporate extra time or not graduating at all. Total cost is thus $202k. This is our lower bound. For our upper bound we divide $2,400b by an estimated cost of (17.5m students x $40,476) or $708b to give 3.39 and multiply by the cost to get $680k as an upper bound. Looking at these three estimates gives us the estimated range we use.|
|$1 to someone living in a least developed country||$20||UN Human Development Report 2007/2008 GDP per capita (PPP US$) High income countries / Least Developed Countries ($33,082 / $1,499)|
We make the following assumptions:
- We perform all calculations in rich country terms
- A life is equally valuable whether it be that of a person from a poor country or a rich country
- When in doubt, estimate conservatively
- When the payoff extends over multiple years (unless there is a good reason to do otherwise) we capitalize the payoff over a 20 year period with no discounting; this corresponds to capitalizing over 30 years with 3% discounting; 53 years with 4% discounting; or to infinity with 5% discounting; there are financial and psychological arguments for limiting the time over we assess returns; an orthogonal justification is that the future is hard to foresee and over longer time periods an alternative solution to the problems faced may be developed; for payoffs and costs far into the future, such as global warming, we discount at 3% consistent with the discount rate used by the WHO. Were we to redo this site we would look critically at the issue of discounting.
- These are rough calculations and Leverage Factors are rounded to 1.5 significant figures as follows; allowed values: 10, 11, 12, 13, 14, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90.
- We are currently inconsistent in our use of ranges to indicate uncertainty associated with an estimate. When we do use ranges we are inconsistent in the likelihood they represent. We are standardizing on using ranges to denote roughly a 68% likelihood, or plus or minus one standard deviation.
- For most ranges we simply multiply upper or lower bound estimates. This produces a broader range than is warranted. So for some of the more uncertain estimates that involve combining multiple uncertain sub-estimates, we assume independence and a lognormal distribution, and combine them more tightly using the following uncertain.py Python script.