An Open Policy Analysis of “California Billionaires: Wealth, Taxes, and Wealth Tax Revenue Estimates”

class: center, middle, inverse, title-slide

.title[
# An Open Policy Analysis of “California Billionaires: Wealth, Taxes, and Wealth Tax Revenue Estimates”
]
.subtitle[
## A reproduction of the paper by Jasper Boll, Emmanuel Saez and Gabriel Zucman (BSZ, NBER, August 2026)
]

---

name: study-brief

# The study in brief

Boll, Saez and Zucman (NBER WP 35218, May 2026, revised August 2026) document the wealth of
California's billionaires, the taxes they pay, and what a one-time tax on them would raise.

- **Wealth.** California billionaires held **$2.3 trillion** by mid-2026, as much as all the
  income on all Californians' tax returns. Up 144% from 2023 to 2025, driven by the AI boom.
  Since 1982 the real wealth of the top 0.001% has multiplied by 25; average real family income
  has doubled.
- **Taxes.** They pay about **0.2% of their wealth** a year in California income tax
  ($3.2 billion a year over 2023 to 2025; the 2019 to 2025 average the estimate uses is
  $3.06 billion, Table 2), 2.4% of the state's income tax revenue over 2023 to 2025.
- **The top four.** SEC filings for Page, Brin, Zuckerberg and Huang (Ellison through 2020)
  show nearly $1 trillion in business wealth, about 40% of the total, taxed at 0.04% of wealth
  a year.
- **The proposal.** A one-off 5% tax, payable over five years, is small next to recent wealth
  gains and large next to taxes now paid. The authors estimate **about $100 billion**
  (Table 5 prints $104 billion; the workbook gives $103.8 billion),
  with minor effects on income tax.

.small[The figures above are the paper's own, from its abstract, Table 2 and Table 5. Every number from the Deck Map
onward is recomputed by the reproduction pipeline. NBER is the National Bureau of Economic
Research, which circulates working papers.]

---
name: opa-frame
class: center, middle, opa-slide

# What is an Open Policy Analysis?

.opa-intro[
A policy analysis is **open** when a reader can see the one estimate it recommends, follow
every step from the inputs to that estimate, and rerun it. The
[Open Policy Analysis](https://doi.org/10.1093/scipol/scaa067) framework asks for three layers.
]

.opa-legend[
**1. Open Output**: [the explorer](../explorer/), one preferred estimate, the assumptions as dials<br>
**2. Open Analysis**: [the report](../repro.html) and **this deck**, the chain from the Forbes
list to the revenue, step by step, with the guesswork marked<br>
**3. Open Materials**: [the repository](https://github.com/fhoces/opa-prop40/tree/main/bsz-analysis),
the pipeline, its tests and every exported CSV
]

.highlight-box[
These slides are the **second layer**. They do not add new estimates; they make the authors'
own computation legible.
]

.credit[OPA done by [Fernando Hoces de la Guardia](https://fhoces.github.io), using [agentic coding tools](../index.html#how-ai-was-used)]

---

.site-nav[
BSZ analysis of California Proposition 40:
[Open Output (the explorer)](../explorer/) &middot; Open Analysis:
[Report](../repro.html) and Slides (here) &middot;
[Open Materials](https://github.com/fhoces/opa-prop40/tree/main/bsz-analysis) &middot;
[Overview](../index.html)
]

# Deck Map

<table style="width:100%; font-size:85%; border-collapse:collapse;">
<tr><th style="text-align:left; padding:6px;">Part</th><th style="text-align:left; padding:6px;">What it covers</th></tr>
<tr><td style="padding:6px;"><b><a href="#part-policy">1. What is being estimated</a></b></td><td style="padding:6px;">The tax, and the two documents that make BSZ's case</td></tr>
<tr><td style="padding:6px;"><b><a href="#part-steps">2. Five steps to $104 billion</a></b></td><td style="padding:6px;">The base scenario, built one line at a time</td></tr>
<tr><td style="padding:6px;"><b><a href="#part-rows">3. Alternative scenarios</a></b></td><td style="padding:6px;">Three more scenarios, the bottom line, and how far each guess moves it</td></tr>
<tr><td style="padding:6px;"><b><a href="#part-repro">4. What reproduces</a></b></td><td style="padding:6px;">Tests, the notes kept, and what is not checked</td></tr>
<tr><td style="padding:6px;"><b><a href="#backup">Backup</a></b></td><td style="padding:6px;">How the Pareto extrapolation finds the missing billionaires</td></tr>
</table>

.blue-box[
Every number from here on comes from the R pipeline that reproduces the authors' August 2026
workbook cell by cell. The deck reads that pipeline's exported CSVs; nothing is retyped.
]

---
name: part-policy
class: center, middle, inverse

# Part 1: What is being estimated

---

# The tax, in one line

.pull-left[
**Proposition 40** (the Billionaire Tax Act):

- a **one-time** 5% tax on net worth
- on California residents on **1 January 2026**
- payable at once or in **five yearly installments**
- wealth valued on 31 December 2026

Directly held real estate is excluded; business real estate is not.
]

.pull-right[
**Two documents make BSZ's case**

1. Boll, Saez and Zucman (**BSZ**), NBER WP 35218, August 2026, with a public Excel
   workbook. Base scenario: **$104 billion as printed**
   ($103.8 billion reproduced).
2. Galle, Gamage, Saez and Shanske (**GGSS**), expert report, 20 July 2026:
   **$100 billion**.

.highlight-box[
GGSS p.4: "a scoring of $104 billion that we round to $100 billion for simplicity."
**The two are one computation.**
]
]

---
name: part-steps
class: center, middle, inverse

# Part 2: Five steps to $104 billion

---

# Steps 1 and 2: Taxable base and evasion

**Step 1.** Forbes' real-time list on 1 July 2026: `$n_0$` = 250 California billionaires
holding `$W_0$` = $2,307 billion. **Data**, read off Forbes that day.

**Step 2.** Some wealth escapes: under-valued, hidden, or moved out in time. The authors
assume a share `$\alpha$` = 10% of all of it together:

`$$W^{\text{tax}} = (1-\alpha)\,W_0 = 0.9 \times 2,307 = 2,076.3$$`

.highlight-box[
The 10% is <span class="guess">guesswork</span>. GGSS argue it is small: most of the wealth is
listed stock, easy to value, and 250 taxpayers can each be audited. Nobody has measured it,
because the tax has never existed.
]

.small[Details: [report, step 2](../repro.html#step-2-avoidance-and-evasion).]

---

# Step 3: the revenue

`$$T = \tau \, W^{\text{tax}} = 0.05 \times 2,076.3 = 103.8$$`

That is Table 5's row 1: **$103.8 billion reproduced (printed $104 billion), once**, payable in five
yearly installments of about $20.8 billion.

---
name: step-4

# Step 4: selling assets to pay raises income tax

Some billionaires will sell stock to pay the tax. The authors assume a share `$s$` = 1/3 of the
tax is paid by selling.

Not all of a sale is income. Only the **gain** is taxed: the sale price minus what the stock
originally cost (its **basis**). The authors put the gain at `$g$` = 80% of each dollar sold, so the
basis is the other 20%. Example: selling $100 of stock bought for $20 gives an $80 taxable gain.

That gain is taxed at the top California rate on gains, `$t_{cg}$` = 13.3%:

`$$E = s \cdot g \cdot t_{cg} \cdot T = \tfrac13 \times 0.8 \times 0.133 \times 103.8 = 3.68$$`

.small[*E* is extra income tax, collected **once**, while the wealth tax is being paid.]

---
name: step-5-who

# Step 5, part 1: who leaves

Step 2 took a share `$\alpha$` = 10% of the wealth out of the wealth tax. In dollars:
10% of $2,307 billion is **$230.70 billion**.

That 10% mixes two behaviours. The authors split it in half, `$m$` = 50%:

| Part of the escaped wealth | Share of all wealth | Wealth, $B | Still a CA resident? |
|---|---:|---:|---|
| Under-valued or hidden | `$\alpha (1-m)$` = 5% | 115.35 | yes |
| Owner **moves out** of California | `$\alpha \, m$` = 5% | 115.35 | **no** |

.highlight-box[
Step 2 already removed **both** rows from the **wealth tax**. Step 5 asks about a different tax,
the **income tax**. Only the second row matters for it: people who still live in California keep
paying California income tax; people who moved out stop.
]

---
name: step-5-loss

# Step 5, part 2: the income tax leavers take

Today California billionaires pay `$C$` = $3.06 billion a year in California income tax.
That is about $1.33 of income tax for every $1,000 of wealth they hold.

The movers from the previous slide hold $115.35 billion. The authors assume they pay
income tax at that same rate, so their income tax is the same 5% share:

| | Wealth, $B | Income tax, $B a year |
|---|---:|---:|
| All California billionaires | 2,307 | 3.061 |
| The movers (5% of the wealth, `$\alpha \, m$`) | 115.35 | 0.153 |

Once they leave, California loses that income tax every year:

`$$L = -\alpha \, m \, C = -0.10 \times 0.5 \times 3.06 = -0.153 \text{ per year}$$`

.highlight-box[
The minus sign means revenue lost. `$L$` is **per year**, and Table 5 does not say for how many
years. BSZ argue the losses should be temporary and give 5 years as an example (PDF p.27). The
[JR analysis](../../../rjkdc-analysis/site/) (Jaros and Rauh, September 2026; first version by
Rauh, Jaros, Kearney, Doran and Cosso) counts them forever, assuming leavers do not come back
(chapter PDF p.32).
]

---
name: part-rows
class: center, middle, inverse

# Part 3: Alternative scenarios

---
name: row-2

# Missing billionaires scenario, part 1: Forbes misses small fortunes

.pull-left[
<img src="data:image/png;base64,#slides_files/figure-html/pareto-fig-forbes-1.png" style="display: block; margin: auto;" />
]

.pull-right[
<img src="data:image/png;base64,#slides_files/figure-html/pareto-fig-typical-1.png" style="display: block; margin: auto;" />
]

Bars are shares of the 617 billionaires a Pareto tail puts above $1 billion. **Right:** in wealth and tax
data, top fortunes usually follow a Pareto tail, where each step down holds many more people.
**Left:** Forbes lists only part of each band below $4.5 billion. The pale violet is what it misses
(34% of all billionaires in the lowest band alone). Fitted to the list above $4.5 billion, that shape implies
**617** billionaires above $1 billion, not 249.

.small[Method: Pareto extrapolation (BSZ pp.24 to 25, Appendix Figure A4). How it works, in the backup:
[how BSZ detect the gap](#pareto-detail) &middot; [from the gap to 617](#pareto-extrapolation).]

---
name: row-2-revenue

# Missing billionaires scenario, part 2: the revenue

Filling the gap adds 368 people but only
$615 billion: **148%
more people, 28.2% more wealth**. The added fortunes are small,
about $1.7 billion each.

The scenario applies those percentages to the 1 July list: 250 become 620
billionaires, and the added wealth is 28.2% of $2,307 billion
= **$650.35 billion**.

| | $ billion |
|---|---:|
| Base scenario revenue (Steps 1 to 3) | 103.81 |
| + 5% of the added wealth, after a higher 20% avoidance: 0.05 × 0.8 × 650.35 | + 26.01 |
| − phase-in: the rate rises from 0 to 5% between $1.0 and $1.1 billion; 12.1% of the added wealth is in that band | − 1.97 |
| **= Missing billionaires scenario** | **127.85** |

.small[The ratio b is **research** (estimated from the list). The 20% avoidance for these less
visible fortunes is <span class="guess">guesswork</span>.
[Report: the missing billionaires scenario](../repro.html#row-2-adding-the-billionaires-forbes-misses).]

---

# Leavers scenario: who may have left

.pull-left[
<table>
 <thead>
  <tr>
   <th style="text-align:left;"> Person </th>
   <th style="text-align:left;"> Left </th>
   <th style="text-align:right;"> Wealth $B </th>
   <th style="text-align:right;"> Inc. tax lost $B/yr </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> Page </td>
   <td style="text-align:left;"> before 1 Jan </td>
   <td style="text-align:right;"> 294.6 </td>
   <td style="text-align:right;"> 0.103 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Thiel </td>
   <td style="text-align:left;"> before 1 Jan </td>
   <td style="text-align:right;"> 27.3 </td>
   <td style="text-align:right;"> 0.055 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Hankey </td>
   <td style="text-align:left;"> before 1 Jan </td>
   <td style="text-align:right;"> 8.2 </td>
   <td style="text-align:right;"> 0.016 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Kalanick </td>
   <td style="text-align:left;"> before 1 Jan </td>
   <td style="text-align:right;"> 3.6 </td>
   <td style="text-align:right;"> 0.007 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Brin </td>
   <td style="text-align:left;"> later </td>
   <td style="text-align:right;"> 271.7 </td>
   <td style="text-align:right;"> 0.078 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Zuckerberg </td>
   <td style="text-align:left;"> later </td>
   <td style="text-align:right;"> 210.3 </td>
   <td style="text-align:right;"> 0.100 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Fang </td>
   <td style="text-align:left;"> later </td>
   <td style="text-align:right;"> 1.5 </td>
   <td style="text-align:right;"> 0.003 </td>
  </tr>
</tbody>
</table>
]

.pull-right[
Assume four (Page, Thiel, Hankey, Kalanick) left **before** 1 January, which BSZ call aggressive, and escape the tax; three more leave later.
Revenue falls to **$88.8 billion**; the yearly income tax loss
rises to **$0.52 billion**. The **missing + leavers** scenario applies
this and the missing billionaires scenario together.

.highlight-box[
Why so little income tax? SEC filings show the top three pay very little state income tax
relative to their wealth: they rarely sell. Whether any of them has **legally** left is a
<span class="guess">legal judgment</span>, not data.
]
]

---
name: bottom-line

# What the tax collects, and what it costs

<table>
 <thead>
  <tr>
   <th style="text-align:left;"> Scenario </th>
   <th style="text-align:right;"> Wealth tax collected<br>(once) </th>
   <th style="text-align:right;"> Extra income tax<br>from selling assets<br>(once) </th>
   <th style="text-align:right;"> Income tax lost<br>(present value, 5 years) </th>
   <th style="text-align:right;"> <b>Total</b> </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> <b>Base</b> </td>
   <td style="text-align:right;"> <b>$103.8B</b> </td>
   <td style="text-align:right;"> <b>+$3.68B</b> </td>
   <td style="text-align:right;"> <b>&minus;$0.70B</b> </td>
   <td style="text-align:right;"> <span style="display:inline-block; border:1.2px solid #111111; padding:1px 6px; font-weight:700;">$106.8B<br><span style="font-size:70%; color:#111111;">MAIN ESTIMATE<sup>1</sup></span></span> </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Missing billionaires </td>
   <td style="text-align:right;"> $127.9B </td>
   <td style="text-align:right;"> +$4.53B </td>
   <td style="text-align:right;"> &minus;$0.90B </td>
   <td style="text-align:right;"> $131.5B </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Leavers </td>
   <td style="text-align:right;"> $88.8B </td>
   <td style="text-align:right;"> +$3.15B </td>
   <td style="text-align:right;"> &minus;$2.36B </td>
   <td style="text-align:right;"> $89.6B </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Missing + leavers </td>
   <td style="text-align:right;"> $114.8B </td>
   <td style="text-align:right;"> +$4.07B </td>
   <td style="text-align:right;"> &minus;$2.56B </td>
   <td style="text-align:right;"> $116.3B </td>
  </tr>
</tbody>
</table>

.highlight-box[
The income tax loss recurs every year the leavers stay away, so it is a **present value** over a
horizon of 5 years (BSZ's example) at **3% a year**, the middle of the [JR analysis](../../../rjkdc-analysis/site/)'s 1.5% to 4.5% range (BSZ give no rate).
Change any assumption and watch the main estimate move: **[the explorer](../explorer/)**.
]

.small[<sup>1</sup> Not the authors' choice: BSZ's headline is the wealth tax alone, about $100 billion (Table 5 prints
$104 billion; reproduced as $103.8 billion).
This OPA uses the total as its main estimate because it compares more directly with
the [JR analysis](../../../rjkdc-analysis/site/)'s net present value; the [comparison overview](../../../index.html) puts both on one definition.]

---
name: part-guess

# How far does each guess move the main estimate?

.small[The vertical line: the main estimate, $106.8 billion (wealth tax + extra income tax &minus;
income tax lost). Of the five guesses, 4 move it on their own:
avoidance and evasion, share that is leaving, share paid by selling and aggressive leavers.
Counting the missing billionaires is research, and only raises it. Every input and its origin:
[report](../repro.html#every-input-by-origin). Try any combination of the guesses in
**[the explorer](../explorer/)**.]

---
name: part-repro
class: center, middle, inverse

# Part 4: What reproduces

---
name: repro-checks

# Checked, reproduced, and not checked

.pull-left[
**Checked.** 794 test expectations pass, 0 fail
(August workbook, run 2026-10-16). They compare every formula cell the
pipeline re-derives against the workbook's own cached value, check the Python twin and
the SQL queries against R, and pin the explorer to Table 5.

**Reproduces.** The code reproduces the paper's published results. Seven notes are
kept for transparency ([materials page](../materials.html#what-the-checks-found)); none is a
failure to reproduce. Three of them:
- Appendix Table A1 annualizes 1982 to 2026 over 43 years, where the workbook's own Table 1 uses 44
- the missing + leavers scenario skips a deduction the missing billionaires scenario applies
- two printed Table 5 cells are one unit lower in the last digit than the workbook's rounded value
]

.pull-right[
**Not checked.**
- the tax statistics and the authors' hand-filled ownership
  shares, taken as the workbook records them. The Forbes lists,
  SEC Form 4 filings and venture capital panels are rebuilt from
  the authors' raw data in [SQL](https://github.com/fhoces/opa-prop40/tree/main/bsz-analysis/sql)
- GGSS has no workbook; its $100 billion is BSZ's base scenario, rounded

.blue-box[
The reproduction shows the arithmetic from the authors' inputs to their tables is what the
paper says. It cannot show the inputs are right, and it cannot settle the guesses.
]

.small[[Materials page](../materials.html) &middot; [report, what reproduces](../repro.html#what-reproduces)]
]

---
name: backup
class: center, middle, inverse

# Backup slides

---
name: pareto-detail

# Backup: the Pareto test, in numbers

Forbes may miss smaller fortunes. BSZ test this with one ratio, **b** = the average fortune above
a cutoff ÷ the cutoff. For a Pareto distribution, the usual shape of top wealth, **b** is the same at every cutoff.
The Forbes list the workbook used (15 April 2026):

<table style="font-size:80%">
 <thead>
  <tr>
   <th style="text-align:left;"> Cutoff </th>
   <th style="text-align:right;"> People above it </th>
   <th style="text-align:right;"> Their wealth, $B </th>
   <th style="text-align:right;"> Average fortune </th>
   <th style="text-align:right;"> b </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> $10B </td>
   <td style="text-align:right;"> 33 </td>
   <td style="text-align:right;"> 1,469 </td>
   <td style="text-align:right;"> $44.52B </td>
   <td style="text-align:right;"> 4.45 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $4.5B </td>
   <td style="text-align:right;"> 89 </td>
   <td style="text-align:right;"> 1,827 </td>
   <td style="text-align:right;"> $20.53B </td>
   <td style="text-align:right;"> 4.56 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $3B </td>
   <td style="text-align:right;"> 125 </td>
   <td style="text-align:right;"> 1,957 </td>
   <td style="text-align:right;"> $15.65B </td>
   <td style="text-align:right;"> 5.22 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $2B </td>
   <td style="text-align:right;"> 172 </td>
   <td style="text-align:right;"> 2,068 </td>
   <td style="text-align:right;"> $12.02B </td>
   <td style="text-align:right;"> 6.01 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $1B </td>
   <td style="text-align:right;"> 249 </td>
   <td style="text-align:right;"> 2,182 </td>
   <td style="text-align:right;"> $8.76B </td>
   <td style="text-align:right;"> 8.76 </td>
  </tr>
</tbody>
</table>

.highlight-box[
Above $4.5 billion, b stays near 4.48. Below it, b climbs to
8.8: too few people for the wealth above them,
the sign of missing small fortunes. If b stayed at 4.48 all
the way down, there would be **617** billionaires above $1 billion,
not 249.
]

.small[[&larr; Back to the missing billionaires scenario](#row-2)]

---
name: pareto-extrapolation

# Backup: from b to 617 billionaires

In a Pareto tail, the number of people above a cutoff `$w$` falls like `$w^{-a}$`. The exponent `$a$`
and the ratio b describe the same shape: `$b = a/(a-1)$`, so

`$$a = \frac{b}{b-1} = \frac{4.48}{3.48} = 1.29$$`

BSZ hold b at its value above $4.5 billion and extend the rule down from the
89 billionaires above $4.5 billion:

`$$N(w) = 89 \times \left(\frac{4.5}{w}\right)^{1.29}, \qquad N(1) = 617$$`

<table style="font-size:85%">
 <thead>
  <tr>
   <th style="text-align:left;"> Cutoff </th>
   <th style="text-align:right;"> Forbes lists </th>
   <th style="text-align:right;"> Pareto tail implies </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> $4.5B </td>
   <td style="text-align:right;"> 89 </td>
   <td style="text-align:right;"> 89 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $2B </td>
   <td style="text-align:right;"> 172 </td>
   <td style="text-align:right;"> 253 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> $1B </td>
   <td style="text-align:right;"> 249 </td>
   <td style="text-align:right;"> 617 </td>
  </tr>
</tbody>
</table>

.small[The same rule gives the added people's wealth: $615 billion (BSZ p.25).
[&larr; Back to the scenario](#row-2)]