Government agencies, such as EPA, often use conservative risk estimates (e.g. 95% percentiles)
However, this adds up
If use 95% percentiles for several estimates, actual percentile is above 99% (.95 x .95 x .95)
Consider two chemical hazards:
Chemical A poses a known risk of 2 in 100,000
Chemical B is uncertain
9 out of 10 scientists believe no risk
1 out of 10 believe risk is 6 in 100,000
Government policy says risk of B is greater, since focuses on upper bound – that is, greatest potential risk.
However, chemical A has a higher expected value of risk.