California Proposition 40 billionaire tax calculator
Interactive calculator for California's Proposition 40 (the 2026 Billionaire
Tax Act, Initiative 25-0024A1) on the November 3, 2026 ballot: a one-time tax
of 5% of the entire net worth of California residents (as of January 1, 2026)
worth $1 billion or more, valued on December 31, 2026. Between $1.0 billion
and $1.1 billion the rate ramps from 0% to 5% (RTC §50301(b) as proposed).
Live app: https://policyengine.org/us/california-wealth-tax
What the model does
One model, scored person by person, in two stages.
- Stage 1: one-time wealth tax. The residency roster is everyone Forbes
listed in California on January 1, 2026. Each person is valued at their
latest Forbes worth, wherever Forbes lists them now; people Forbes has
added to its California list since are added, with their January 1
residency assumed; people Forbes no longer lists at all are carried at the
last positive value in the stored daily history. Directly held real property
is excluded, then the $1 billion threshold and rate ramp apply to each
observed wealth-tax unit. Documented separately listed spouses share one
threshold and rate (Jay-Z/Beyoncé and Lynda/Stewart Resnick); other family
assets remain unobserved. The model then applies an optional erosion haircut,
the legal assumptions, and the payment election (lump sum due in 2027, or
five installments with a 7.5% charge on the unpaid balance). The legal
assumptions are a probability that the measure survives legal challenge,
which scales expected receipts and nothing else, and two switches that take
people out of the base without moving them: people who left after
January 1, and non-US citizens. Each defaults to the measure as written.
With stage 2 off, the headline is nominal receipts.
- Stage 2 (optional): future California income tax. The cohort's total
California income tax is an input ($4.3B a year by default; Rauh et al. and
Boll, Saez and Zucman overlap on it). It is divided among people either by
wealth (Rauh et al.'s f × C) or, by default, with filings-based estimates for
the four largest fortunes (
data/income_tax_filings.json, from Boll, Saez and
Zucman's Table 3) and the rest by wealth. A mover's share is the loss; the
stream is attributed, grown, discounted, and netted against receipts, both in
present value as of 2026. Income-tax departures remain person-based: a
departing spouse can lose California income tax while the unit's wealth-tax
liability remains because the other spouse was a resident on January 1.
An optional total share of the cohort's income tax lost to departures
(Jaros and Rauh's f) counts documented and modeled departures first; any
remainder is a proportional share of the people still paying who leave after
January 1, owing the wealth tax but no longer California income tax.
The baseline takes no position on contested residency. Eleven documented claims
that a roster member was not a resident on January 1, 2026 are individual
adjustments, each shown with its evidence and source
(lib/residencyAdjustments.js, data/billionaire_metadata.json). Larry
Ellison is the one exclusion from every base: Forbes lists him in Florida and
both published estimates now exclude him.
The "Berkeley assumptions" and "Hoover assumptions" starting points apply the
assumptions of Galle, Gamage, Saez and Shanske and of Jaros and Rauh to this
calculator's roster. They are settings inside the same model, not replications
of those papers' rosters. Since October 21, 2026 the Hoover set follows Jaros
and Rauh's NBER Working Paper 35899 (October 2026), as in their central
revenue estimate: their departures and pre-filing exits out of the base, Mark
Zuckerberg and non-US citizens left out, 7% a year wealth growth to the
valuation date, a 50% chance the measure survives challenge, no unannounced
departures in the wealth-tax base, and 45% of a $4.55B cohort income tax lost,
divided by wealth. Their NPV section instead centers revenue on half of the
$72B ceiling with Zuckerberg and the pre-filing exits in (their formula gives
a point NPV of −$32.25B at f = 0.45, C = $4.55B and r − g = 3%) and reports a
mean over draws of revenue, f, C and r − g (−$38.9B), so
the set's point value is not their published mean. Before, the set followed
Rauh et al.'s March 2026 paper; that set stays in lib/presets.js as
HOOVER_MARCH_2026_ASSUMPTIONS, and the tests still score it.
Related work
Two other analyses compare versions of the Berkeley and Hoover estimates. The
page lists them at the bottom, with summaries kept in lib/relatedWork.js:
This is the beginning of the README. Read the rest at the original. Read the full README ↗
