Series: The Secret IRS Files
Inside the Tax Records of the .001%
For the article “Trove of Never-Before-Seen Records Reveal How the Wealthiest Avoid Income Tax,” ProPublica delved deep into the taxes and strategies of the 25 wealthiest people in America. The story’s main finding was that these 25 people saw their worth rise a collective $401 billion from 2014 to 2018 while paying a total of $13.6 billion in federal income taxes. That amounts to what we called a “true tax rate” of 3.4%.
Below, we’ve laid out how we performed this analysis. Our story also included calculations of more typical American households in order to provide context for the ultrawealthy’s numbers, and we explain those here as well.
In addition to the article that measured the taxes of the wealthiest Americans against their wealth growth, we have also published a story that provides a more traditional measurement of taxes — an effective income tax rate — for that same group. You can read that story and its methodology here.
How we calculated wealth growth for the ultrawealthy
For estimates of wealth, wealth growth and the list of the 25 wealthiest, we relied on the annual Forbes Billionaires list. Forbes’ estimates, while not perfect, are the best available, according to experts, and are widely cited in the press and in academic publications. They have occasionally proven to be off, infamously so in the case of Wilbur Ross. But experts said they were likely more accurate for those at the top of the list, particularly people — like Jeff Bezos, Warren Buffett, Elon Musk and others mentioned in our initial article — whose wealth is concentrated in shares of their own companies, which must be publicly disclosed. Our top 25 was composed of the people who led the Forbes list as of March 2021, excluding MacKenzie Scott, who was married to Bezos throughout the period covered by our records.
To calculate wealth growth for the 25 richest Americans, we looked at year-to-year changes in Forbes’ estimates of their wealth. The Forbes Billionaires list is released every March, misaligning it slightly with the tax year, which follows the calendar year. For any given tax year, we treated Forbes’ estimate from March as the wealth figure for the end of the previous year (for example, we treated wealth estimates for March 2019 as end-of-year for tax year 2018). To reduce the chances that temporary financial anomalies would distort results, we generally looked at wealth growth and taxes over a period of five years or longer. However, we sometimes cite individual years for certain people. In those cases, we checked stock prices to ensure that the value of the main asset that the person was known to own — say, Amazon shares, in the case of Bezos — didn’t diverge too widely from the Forbes data.
How we calculated personal federal taxes for the ultrawealthy
The tax numbers came from IRS data obtained by ProPublica. When we cite the “income tax” of an individual or the group, we follow the IRS’ method for calculating this as outlined in Publication 1304; self-employment taxes and other non-income taxes on the 1040 are excluded. When we refer more broadly to “total taxes” or “personal federal taxes,” we are incorporating all taxes on the 1040 as well as employees’ share of wage taxes for Social Security and Medicare. Generally these levies, known as “payroll taxes,” are quite minor for the wealthiest and make little difference whether incorporated or not. (This is true even when including both sides of payroll, as we did for our comparison in “You May Be Paying a Higher Tax Rate Than a Billionaire.”)
In cases where the return was amended, we used the amended amount.
How we calculated comparison figures for “typical” wage-earning American households
To create estimates of wealth and income over time for typical wage-earning American households, we relied on the Federal Reserve’s Survey of Consumer Finances (SCF), which collects data on wealth and income from Americans every three years. We then calculated federal taxes for those households.
In general, we used the word “typical” to mean either “median” or households from the middle of the wealth or income distributions. By “wage-earning” or “working,” we referred to households that were listed as headed by someone who was working in some way and earning 90% or more of their total income from wages. This covers most households headed by a working-age adult.
The SCF is conducted once every three years and covers household finances from the previous year (for instance, the 2019 SCF covers 2018 finances). To calculate wealth and income between survey years, we had to interpolate values. For this, we assumed uniform wealth and income growth (or decline) each year between surveys.
To calculate the taxes for our typical households, we began with a certain set of financial characteristics. Our household was a married couple filing jointly. It had the average number of children for its SCF cohort (we allowed for partial children, even though children are generally recognized to come in whole numbers, rather than fractions). Its income came entirely via wages. Using that profile (and assuming the standard deduction), we calculated the income taxes for our household for each year from 2014 to 2018 using the laws then applicable. To that total, we added employees’ share of payroll taxes (Medicare and Social Security). For simplicity and clarity for the lay reader, we chose to exclude employers’ share of payroll taxes and limit our comparison to taxes directly paid by our household.
We adjusted neither our figures for typical Americans nor those for the 25 wealthiest Americans for inflation.
In our story, we say that the typical wage-earning household paid 14% in federal taxes on its income “in recent years,” by which we meant between 2014 and 2018 (the latter is the most recent year for which the IRS has released information). We based that on median household income for wage-earning households from the 2013, 2016 and 2019 SCFs, imputing income for the years in between and applying income and payroll tax rates in those years.
We also say that the typical wage-earning household’s wealth grew by $65,000 between 2014 and 2018, while it paid about $62,000 in federal taxes. For this calculation, we had to take a slightly different approach.
First, since the typical American household tends to build wealth as those who head it get older, we wanted to find an age cohort that was roughly in the middle among all working-age groups in terms of wealth growth. The typical American household we chose in our example was headed by someone born between 1973 and 1977, making them between 41 and 45 years old at the end of our time period. (For the 2006-2018 period in the graphic “Compare Bezos’ Financial Picture to a Typical American Household,” we employed the same methodology, except we used a cohort of households headed by someone born in the five years around 1964, when Bezos was born.)
Second, we couldn’t use a simple median figure for wealth or income to calculate the taxes for our cohort each year. That’s because households with median wealth don’t necessarily have median income, or vice versa. Instead, to create a large enough sample, we calculated a weighted average of wealth and income (as well as marital status and number of children) for households that fell between the 40th and 60th percentiles of wealth for our age cohort to arrive at final figures for each survey year. We then imputed wealth and income between SCF years and calculated taxes and wealth growth for the typical family in our cohort using these figures.
This method for calculating wealth growth and income for a typical household was informed by discussions with economists who are experts in using the SCF to study household wealth and income.
Finally, we say that it would take about 14.3 million typical wage-earning households to equal the wealth held by the top 25 wealthiest Americans in 2018, and we calculate that those 14.3 million Americans would have paid about $143 billion in taxes that year, compared with $1.9 billion that the top 25 paid.
To do this calculation, we followed a similar approach to the one we used for calculating the income and wealth of a family with typical wealth growth, except we removed the age filter and used only one SCF year. That is to say, we calculated the weighted average wealth and income of wage-earning households falling between the 40th and 60th percentiles of the wealth distribution in the 2019 SCF. Next, we calculated the taxes that someone would have owed in 2018 who made that weighted average income. Finally, we divided the $1.1 trillion in wealth held by the 25 richest families by the weighted average wealth that we calculated to get 14.3 million households, and we multiplied the taxes we calculated for that typical household by 14.3 million to get $143 billion in taxes paid by the entire group.