Updated December 13, 2024, and February 28, 2025

August 2, 2024

Updated December 13, 2024, and February 28, 2025

Estimating Tax Burdens by Wealth Groups

Akcan Balkir, Emmanuel Saez, Danny Yagan, and Gabriel Zucman

UC Berkeley

Abstract:

This paper estimates income and tax burdens by wealth groups with a particular focus

on top wealth groups. Taxes include individual income taxes (federal and state), payroll

taxes, estate and gift taxes (federal and state), and corporate taxes (federal, state, and

foreign). At the very top of the wealth distribution, corporate taxes (paid by businesses

owned by the wealthiest) are the largest tax followed by individual income taxes (with

over half of reported individual income taking the form of realized capital gains). Other

taxes and in particular estate and gift taxes are relatively minor. Charitable contributions

are on the same order of magnitude as the sum of all taxes paid.

This paper was developed in the context of the external research contract TIRNO16E00013 with

the Statistics of Income (SOI) Division at the US Internal Revenue Service. We are grateful to

Bob Gillette, Barry Johnson, and Mike Weber for helpful comments and guidance in using the

tax data. We also thank the OTA and IRS-SOI team for a stimulating discussion of preliminary

results.

1

1. Research project description

The goal of this project is to estimate tax burdens by wealth groups with a particular

focus on top wealth groups. Taxes include individual income taxes (federal and state),

estate taxes (federal and state), payroll taxes, property taxes, and corporate taxes

(federal and state). This question is of direct tax administration interest in light of various

proposals to “tax wealth like work”. Several bills in Congress and state legislatures have

also been proposed to condition taxes on high net worth, sometimes focusing

specifically on billionaires, as well as reforms to corporate income taxation. Creating

statistics on the tax burden by wealth groups all the way up to very top can help

illuminate the effects of such proposals on tax burdens by wealth group.

Measuring tax rates by wealth groups requires estimating wealth. There are four

strategies that can be used to estimate wealth in tax data. First, estate tax data provide

a direct measure of wealth at death for wealthy decedents and can be linked to previous

income tax data following a long tradition within the Statistics of Income (SOI) division at

IRS (see e.g., Bourne et al. 2018). Second, capital income from income tax data can be

capitalized to estimate wealth (as done by Saez and Zucman 2016). Third, business

wealth can be estimated using the detailed balance sheets reported on business tax

returns (as done recently by Smith, Zidar, and Zwick 2023). Fourth, the top 400

billionaire list created for Forbes magazine since 1982 can be matched to tax data (as

done in Raub, Johnson, and Newcomb 2010 who compare Forbes wealth with reported

wealth on estates of Forbes decedents).

For each wealth group, we measure wealth, income, and taxes broken down by

category of tax. Our key innovation relative to the Piketty, Saez, and Zucman (2018)

distributional national income statistics is that we rank by wealth, match individual tax

data to estate and gift tax data (to obtain better estimates of estate and gift taxes paid)

and (in progress) to business tax data (to obtain better estimates of corporate income

taxes paid), and to Forbes 400 data (to obtain better estimates of the tax burden for this

2

very top group). Our results can inform simulations of how various capital tax proposals

would affect total tax burdens by wealth groups.

Methodology

We rank tax units by wealth. Percentiles are defined relative to the full population (tax

filers and non-filers). We break down the full population of tax units (including non-filers)

by wealth groups using capitalized wealth estimated from Distributional National

Accounts (Piketty, Saez, and Zucman 2018 and most recent updates) up to the top

.001%, the highest wealth group that can be reliably considered using the capitalization

method. For groups above the top .001%, we match the Forbes 400 lists (from 2010 to

2020) to tax data following the methods of Raub, Johnson, and Newcomb (2010).

Matching methodology. The match to tax data is done using the publicly available

exact date of birth, first 4 letters of the last name, and state of residence of the Forbes

400 wealthiest from year 2010 to 2020. These variables are matched to the INSOLE

individual income tax files for years 2005-2019 restricted to large incomes (AGI in

absolute value in excess of $5m). We start from the finest matches using all variables,

and then move to coarser matches that use only the year of birth (as opposed to the

exact day of birth). We were able to match about 98% of individuals from the Forbes

400 lists from years 2010 to 2020. Once a match is obtained, we can compute individual

income and taxes for years 2010 to 2020 by matching to individual tax data and using in

priority the edited SOI INSOLE files and then, if necessary, the unedited full population

CDW data to complete the data. For estate and gift taxes, we match to estate and gift

tax returns in the CDW database. For corporate taxes (in progress), we match owners

to the C-corporations they own using the SOI corporate study file for 2019 along with

the CDW data for additional variables, years, and firms. The match of C-corporations is

done using two strategies.

3

We then use Forbes wealth to rank billionaires into three groups in: top .0002%,

top .0001%, and top .00005%. In 2019, the top .0002% corresponds to approximately

the richest 360 tax units on the Forbes list, the top .0001% to approximately 180 tax

units, and the top .00005% to approximately the richest 90 tax units. This fine division is

informative. Most of the increase in the Forbes wealth share is due to the top 100 within

the top 400 as shown earlier by Kopzcuk and Saez (2004). Moreover, the results below

find that there are substantial variations in tax burden relative to income and to wealth

across these the very top groups.

We report statistics averaged over groups of years: 2010-3, 2014-7, and 201820. These temporal divisions allow us to compare results before the TCJA tax reform

(2010-3, and 2014-7) and after the TCJA tax reform (2018-2020).

Computation of taxes.

Individual income taxes. Individual income taxes are directly measured from individual

income tax return 1040 information. We directly compute federal Social Security and

Medicare taxes on wage income based on that 1040 information. State income taxes

are obtained from Schedule A of form 1040 (such taxes are reported without cap even

after TCJA capped such deductions). Foreign income tax is derived from the foreign tax

credit Form 1116.

Transfer taxes. Transfer taxes include federal estate and gift and analogous state taxes

for next year decedents (e.g., members of the 2019 Forbes list who die in 2020). Tax

data is obtained from the federal estate tax returns forms 706, as well as gift taxes for

gifts made during the year and reported on the federal gift tax form 709. Estate tax

return data includes both the federal tax paid as well as state level taxes. Gift tax data

includes only the federal gift tax. Because there are relatively few decedents in any

given year, to smooth year to year variation, transfer taxes are averaged across all

years 2010 to 2020. A large share of the Forbes 400 list make gifts in any given year, so

all statistics are based on a large number of returns with non-zero values.

Corporate taxes.

4

We employ two methodologies to estimate corporate taxes for the very top groups: a

simpler methodology based on macroeconomic tax rates, and a refined methodology

based on matching the Forbes 400 to public and internal tax data on the businesses

they own.

Simpler corporate tax methodology. The simpler methodology is based on an

extrapolation entirely using pre-existing public data, based on the Distributional National

Account data for the top group .001%. We compute the ratio of corporate taxes paid by

this group to wealth owned by this group, and we then apply this ratio to the wealth of

the very top groups .0002%, .0001%, and .00005% in order to estimate the corporate

taxes paid by each very top group. This procedure amounts to assuming that the ratio of

corporate taxes to wealth remains the same within the top .001%. This assumption will

hold, for example, when the fraction of corporate equity in wealth remains constant

across the very top groups and the corporate tax relative to equity is also constant

across the very top groups. Once we have estimate the corporate taxes, we break them

into federal, state and local, and foreign based on aggregate data.

Refined corporate tax methodology (in progress).

We create a refined corporate tax computation for 2019-owned-corporations only to

assess the assumptions made in the simpler methodology just described. We match the

Forbes 400 to public and internal tax data on the businesses they own, based on SEC

Schedule 13-D and related filings. For publicly traded businesses, the Forbes 400 data

provide the company ticker and stake owned by the person. We merge this information

to the publicly available Computstat database that provides information on the book

income and corporate taxes paid by these large businesses. The Compustat database

is created by compiling the publicly available 10-K forms that all publicly traded

companies must file with the Security Exchange Commission. We use Compustat

current federal taxes, current state taxes, and current foreign taxes and prorate them

according to the ownerships shares.

For private businesses, the Forbes data provide the business name and

sometimes information on stakes on many private businesses owned by the Forbes

400. We first obtain EINs for these private businesses from online searches. We then

match EINs to internal corporate tax data, both the corporate study (which is a stratified

5

sample of corporations with many variables and quality checks developed and

maintained by SOI at IRS) and the CDW corporate database (the complete raw

administrative tax data that includes the universe of US corporations but without

additional cleaning of the variables). From these matches, we use book income

information from the schedules L and M3 to obtain the comparable information to the

Compustat data used for publicly traded corporations. For stakes of private businesses,

we use both the information from Forbes supplemented by information from schedule

1125-E of the 1120 tax form, which provides information on stakes owned by officers of

the company (as most of the large shareholders of private businesses are also officers

of the company).1

The refined methodology may yield different results from the simpler

methodology, or may yield similar results. Key determinates of difference or similarity

will be the share of top-owned private business that are C-corporations and thus face

the corporate income tax, and the effective tax rates of those top-owned private

businesses. The results reported below are based on the simpler method; we will

present sensitivity analysis based the refined method in a future updated draft.

Results

Table 1 reports the main results. The table reports capitalized wealth, adjusted

gross income (AGI), and various subcomponents of AGI, as well as various taxes. All

amounts are the average per tax unit in the group and expressed in thousands of

current dollars.

For wealth percentiles up to the top .001%, the statistics are based entirely on

pre-existing data. Wealth estimates and the non-corporate-tax statistics are drawn from

(Saez and Zucman 2016). The corporate tax estimates are based on the distributional

national account method as in Piketty, Saez, and Zucman (2018, updated). These

statistics are reported for year 2019. For very top percentiles .0001% and above, the

non-corporate-tax values are based on this paper’s matching of the Forbes 400 list to

tax data, reporting means over years 2018, 2019, and 2020. For these very top groups’

1 The definition of officer of a company varies depending on State level corporate laws.

6

corporate tax estimates, we report values from the simpler method described above

based entirely on preexisting public data.

For the very top groups, estimated corporate taxes are the largest category of

tax, followed by individual income taxes (with over half of reported individual income

taking the form of realized capital gains). Other taxes and in particular estate and gift

taxes are relatively minor. At the very top of the wealth distribution, charitable

contributions are about as large at the sum of all taxes paid. Note also that current

charitable contributions are larger than deducted charitable contributions, due to the

50% AGI limit on charitable deductions (with carry-forward). It is also of note that the tax

to wealth ratio declines throughout the wealth distribution, including the very top

percentiles within the top 400.

Table 2 follows the data sources of Table 1 and considers the evolution of

income and taxes for the very top wealth percentiles over time from 2010 to 2020. We

start by computing income and taxes for each percentile and tax year from 2010 to

2020, and then present averages across 3 groups of years: 2010-3, 2014-7, and 20182020 (already presented in Table 1). A notable result is that the tax to wealth ratio

declines over time, consistent with both the rise in wealth at the very top and the

decrease in tax rates after TCJA (period 2018-2020).

In Table 3, we consider cumulative income and taxes paid by the richest in the

10 years after (in the left-panel columns) or the 10 years before (in the right-panel

columns). The goal is to compare the income and taxes accrued during a decade

compared with the level and change in wealth over the same period. This provides a

longer-term perspective that complements the shorter-term evidence presented in Table

1. The statistics are overall consistent with the shorter-term statistics and also similar for

the left-panel and right-panel, consistent with a relatively modest turnover at the very

top of the US wealth distribution.

Table 4 presents supplementary means. Panel A presents means of 2019

corporate taxes in three subsets of the Corporation Study File: private C-corporations

with a 2019 Forbes owner, large private C-corporations (those with at least

$250,000,000 in gross receipts or $2,500,000,000 in assets), and large private Scorporations. The matched C-corporations have a ratio of mean taxes to mean pre-tax

7

book income that comparable to the analogous ratio of large private C-corporations

generally. These values can be used to implement versions of the refined corporate tax

methodology discussed above. Panel B presents estate tax ratios among matched

decedents that contribute to the Table 1’s Estate, inheritance, and gift taxes values.

Taxable estates are a fraction of gross estates and Forbes private wealth, consistent

with Table 1’s relatively small Estate, inheritance, and gift taxes values. See the table

note for further details.

Table 5 presents additional means of pass-through business income. Panel A

presents means of positive and negative partnership and S-corporation income among

Forbes tax units matched to INSOLE. The second, fourth, sixth, and eighth columns

apply the tax-unit-level definition from Table 1 while the other columns apply an

alternative definition at the tax-unit-by-pass-through-business-type level. Differences

across the definitions reflect tax units that earn positive business income from some

firm(s) and negative business income from other(s) that can offset each other and lead

to the relatively small net business income amounts listed in Table 1. Panels B-C

present means of the 2019 partnership and S-corporation income among such

businesses that were able to be matched to 2019 Forbes tax units. The panels show

substantial differences between ordinary business income and book income that

coincide with substantial capital gains and dividends. See the table note for further

details.

8

References

Bourne, Jenny, Eugene Steuerle, Brian Raub, Joseph Newcomb, and Ellen Steele.

2018. “More than They Realize: The Income of the Wealthy.” National Tax Journal 71,

no. 2: 335–56.

Leiserson, G., and Yagan, D. (2021), ‘What is the Average Federal Individual Income

Tax Rate on the Wealthiest Americans?’ Council of Economic Advisors.

Parisi, Michael, and Michael Strudler. 2003. "The 400 individual income tax returns

reporting the highest adjusted gross incomes each year, 1992-2000: data release."

Statistics of Income. SOI Bulletin 22, no. 4: 7-10.

Piketty, Thomas, Emmanuel Saez, and Gabriel Zucman. 2018. “Distributional National

Accounts: Methods and Estimates for the United States.” Quarterly Journal of

Economics 133 (2): 553-609.

Raub, Brian, Barry Johnson, and Joseph Newcomb. 2010. “A Comparison of Wealth

Estimates for America’s Wealthiest Decedents Using Tax Data and Data from the

Forbes 400.” National Tax Association 103rd Annual Conference on Taxation,

November 18–20. https://ntanet.org/wp-content/uploads/proceedings/2010/020-raub-acomparison-wealth-2010-nta-proceedings.pdf

Saez, Emmanuel, and Gabriel Zucman. 2016. “Wealth Inequality in the United States

since 1913: Evidence from Capitalized Income Tax Data.” Quarterly Journal of

Economics 131, no. 2: 519–78.

Saez, Emmanuel and Gabriel Zucman. 2019. “Progressive Wealth Taxation.” Brookings

Papers on Economic Activity, Fall 2019, 437-511.

Slemrod, Joel. “The Fortunate 400” 100 Tax Notes 935 (Aug. 18, 2003)

Smith, Matthew, Owen Zidar, and Eric Zwick. 2023. “Top Wealth in America: New

Estimates under Heterogeneous Returns,” Quarterly Journal of Economics 138(1): 515573.

Statistics of Income. 2016. “The 400 Individual Income Tax Returns Reporting the

Largest Adjusted Gross Incomes Each Year, 1992–2014” https://www.irs.gov/pub/irssoi/14intop400.pd

Yagan, Danny. 2023. “What is the Average Federal Individual Income Tax Rate on the

Wealthiest Americans?” Oxford Review of Economic Policy 2023, volume 39 pp. 438450.

9

10

11

12

13

14