Excess mortality & VSL$

Loading…
⚠ CONDITIONAL What is being compared: the rule stays in force vs. the rule is repealed. A modelled counterfactual, not an enacted outcome — the rule is still under White House regulatory review and the final rule is not yet published in the Federal Register.

Loading…

What was computed, and what the three anchors are

Loading…

Loading…

Loading…

Peak added PM2.5, downwind

Where the peak is, and its spread

Loading…

Excess deaths per year

How this death count is worked out

Loading…

Health damages, illustrative

What a statistical life is priced at

Loading…

This cluster is not the fleet — read this before quoting a number.
How it compares with EPA's fleet-wide figure

Loading…

How these numbers were made, checked and reproduced
Dose–response curve
GEMM (the Global Exposure Mortality Model, Burnett et al. 2018) — a curve that turns a change in PM2.5 into a change in death rate. It gives higher estimates than the simpler curves EPA uses, so treat these figures as the upper end of the plausible range.
Baseline death rate
0.01182 deaths per person per year — the US national rate for non-communicable disease plus lower-respiratory infection, ages 25 and over (the causes GEMM is fitted to), applied to the adult (25+) population. The national counterpart of the California-specific rate used in the Bay-Area demos.
Emissions denominators
Shares are taken against Climate TRACE’s 2025 coal-fleet totals — 613,000 t SO2 and 687,000 t NOx per year. The NOx total runs about 41% above the publicly reported coal-fleet NOx figure (486,000 t/yr); SO2 is within 4.8%. Sulfate formed from SO2 dominates the modelled health effect, so these figures are not sensitive to the NOx gap.
Value of a statistical life
Loading…
Where the effect is counted
Loading…
What is being compared
Loading…
Which plants
Loading…
Emissions-to-air model
Loading…

Checks we ran before trusting these numbers

Loading…

Raw output from each check

How to reproduce these figures

Loading…

The exact read contract

Column layout of the matrix file, and the single ranged request used to pull one source column: