Studies · CA Air Quality · Investigation 01

How Many People Die from California’s Air?

Before you can rank policies, you need a baseline — the size of the annual health burden, the pollutant mix, and the sectors driving it. This investigation quantifies all three across 21,164 census-tract grid cells and 32 million people.

14,665
Deaths / Year
$179.6B
Monetized Burden
4,034
Hospitalizations
$15.2M
EVPI
Sector Decomposition

Ozone Dominates, Wildfire Drives PM2.5

The 14,665 annual deaths split across two pollutant pathways with different policy responses. Ozone accounts for 62% of mortality (9,069 deaths); PM2.5 accounts for 38% (5,596 deaths). Most electrification policies cut PM2.5 directly but can raise ozone through NOx chemistry — a cut in one pollutant is not a cut in the other.

PM2.5 Deaths
5,596
38% of total burden. Dominated by wildfire smoke (77% of PM2.5 mass).
Ozone Deaths
9,069
62% of total burden. Driven by photochemistry across the entire state.
DAC Share
20.8%
Disadvantaged communities bear 20.8% of burden (22.9% of population).

Wildfire accounts for roughly 77% of California’s PM2.5 mass. Transport, buildings, and industry combined cover the remaining 23%. Electrification policies — at any cost — can touch only that 23%. The dominant PM2.5 source is not in the policy toolkit.

Uncertainty Structure

Wide Uncertainty, Robust Decision

The 10,000-draw Monte Carlo produces a wide credible interval: the 5th percentile is 7,679 deaths and the 95th percentile is 23,217 deaths. This 3.0× spread comes primarily from concentration-response function (CRF) uncertainty and the choice between log-linear and supralinear functional forms.

Despite that wide uncertainty, the Expected Value of Perfect Information (EVPI) is only $15.2M. Resolving all uncertainty in the baseline burden would shift the optimal policy by at most $15.2M in expected value — meaning the baseline magnitude is highly uncertain, but the decision riding on it is not.

The surrogate model drives the most EVPPI. Of the $15.2M EVPI, $5.4M is attributable to the ISRM surrogate model uncertainty — the transfer function from emissions to concentrations. CRF structure, emissions inventory, and monetization each contribute less than $1M individually.

Finding
California’s air kills an estimated 14,665 people per year — $179.6B in monetized burden. Ozone is the majority (62%), not PM2.5. Wildfire dominates the PM2.5 that remains. Only 23% of PM2.5 mass is addressable by electrification. The baseline decision is robust: EVPI is $15.2M against a $180B burden.

10,000 Monte Carlo draws · 21,164 census-tract grid cells · 32M population · EPA BenMAP CRFs (log-linear + supralinear) · ISRM source-receptor matrix (2-level for PM2.5, InMAP for NOx) · EPA VSL $11.6M (2024$) · Convergence verified on total deaths, monetized value, DAC share, and benefit ratio

← Back RFAQ Overview Next → Transport Electrification