Sensitivity analysis EPA’s mercury rule Finalized 2011  ·  Reversed 2020

Is this rule worth it?

A sensitivity analysis asks what happens to your answer when you change the assumptions. Usually it runs at the end, as a robustness check. Run it on this rule and it stops being a check — the answer moves from minus ten billion dollars to plus one hundred, and the whole range comes out of the same government document. Nobody is lying anywhere on this line.

Where you are now
+$58 billion
a year. Benefits beat costs about 7 to 1.
break even
you
Start here This is where EPA landed in 2011, using the assumptions EPA chose. Change one and watch the dot move.

The arithmetic, per year

What the rule buys
Fewer children with lost IQ Kids of anglers who eat what they catch · at 3% $6 million
Fewer people dying of soot 7,600 deaths avoided × $8.9M each $68 billion
Total $68 billion
Real, and worth zero here: 4,700 heart attacks and 130,000 asthma attacks a year, plus every mercury harm except childhood IQ — heart disease among them. A line with no dollar figure counts as nothing in the arithmetic and as something else in the world.
What it costs
Cleaning up the power plants The one number nobody in this fight disputes $9.6 billion
Net +$58 billion
“Read naturally in the present context, the phrase ‘appropriate and necessary’ requires at least some attention to cost.”
Michigan v. EPA (2015) · 5–4 · Scalia, J.

The Court never said what the number had to be. It said the agency had to produce one. This whole page is what producing one looks like.

Five assumptions you can test

What you’re allowed to count co-benefits
This rule was written to cut mercury. Scrubbing the smokestacks also cuts soot, which kills far more people. Those soot deaths are a side effect — so do they belong in the ledger?
Whose kids count standing
EPA put a dollar figure on mercury harm for one narrow group: children whose parents fish for fun and eat the catch. Everyone else exposed to mercury is in the analysis at zero.
† Nobody published these. Widen who counts and you owe the room an argument for the multiplier. Keeping the narrow number doesn’t avoid that argument. It just hides it.
What a life is worth VSL
The soot line is nothing but deaths multiplied by this number. Drag it and watch the whole benefit column move.
$8.9M
The dial opens at $8.9M, roughly what the 2011 analysis leaned on. Every federal agency now uses more than that: DOT $14.2M and HHS $13.6M as published figures, EPA, DHS and OSHA between $10M and $12.8M once their older benchmarks are carried to current dollars. Poorer countries come out lower, because the method prices what people there can afford to spend avoiding risk.
How much the future counts discount rate
A child’s lost IQ becomes lost wages thirty years from now. This number decides how much that is worth today.
A higher rate means waiting hurts more, so a distant benefit is worth less today. EPA’s mercury line falls from $4–6 million at 3% to $0.5–1 million at 7% — same children, same harm, a twelfth of the money. The soot line barely moves, because those deaths are prevented soon. † The government set 2% in 2023 and went back to 3–7% in 2025. The rate changed because an election did.
Which number gets quoted point vs. range
EPA never published one figure for deaths avoided. It published 4,200 to 11,000. All three of these are in the document.

Three people, same document

What the sensitivity analysis found

Which assumption actually moves the answer
The range you have covered
—
Move a dial and the range starts building here. The width of that range, not the number you started with, is the result.
Nothing tested yet.

Every number, and where it came from

The soot line is deaths × value of a life, which reproduces EPA’s published $37–90B closely while folding avoided illness in with avoided death. Built for LPPP 6350, The Politics of Public Policy.