Friday, April 16, 2010

Food for Thought: A Graphical Peek at Income Inequality in the United States

By David L. Griscom

Ever since my post on the the unemployment situation in the U.S. last Sunday, I have been trying to get up to speed on what is known about income equality. My starting point was "15 Mind-Blowing Facts About Wealth And Inequality In America" comprising 15 graphs, the most impressive of which I reproduce below in a slightly modified form to make the comments at the top and bottom legible.

Because the fine print in this figure could scarcely be read even in its web-published version, I reproduce those comments in the quotes below, which refer to the three vertical green bars in the top graph and then the two red bars in the lower graph.
The greater the gap between the rich and everyone else, the more dangerous economics becomes.
In 1928, a year before the U.S. economy nose-dived into depression, the top one-hundreth of 1 percent of U.S. families averaged 892 times more income than families in the bottom 90 percent.
In 1980, the last pre-Reagan year, families in the bottom 90 percent averaged $30,446 in income, after adjusting for inflation, $72 more than the the $30,374 comparable families earned in 2006. The top 0.01 percent in 1980 took home an average $5.4 million, less than one-fifth the $29.6 million average income of the super-rich in 2006.
In 2006 the top 0.01 percent averaged 976 times more income than America's bottom 90 percent.
In 1944 the top marginal tax rate -- the rate on income in the highest tax bracket -- hit 94 percent. In that year, taxpayers making more than $1 million, in 2005 inflation-adjusted dollars, paid Uncle Sam 65 percent of their total income in tax.
In 2005 taxpayers making more than $1 million faced a top marginal rate of 35 percent. These deep pockets paid just 23 percent of their income in federal tax.
Still, this nicely colored and annotated visual is a second-hand product. So I searched the web to find the authors of the two graphs. It turned out that the bottom graph was relatively well known (here for example), whereas the upper one was exclusively the result of relatively recent research by Professor Emmanuel Saez and his coworkers. The main publication on the subject authored by Thomas Piketty and Saez is entitled Income Inequality in the United States 1913--1998 and was published in the February 2003 issue of The Quarterly Journal of Economics. It has since been updated to 2006. These papers are based on study of individual income-tax returns that have been published by the IRS ever since 1913. The authors have carefully studied these returns in order to separate the shares of each source of income, such as wages, business income, and capital income. Their plotted data are for "tax units" defined as a married couple living together or a single adult, including dependents if any. Because of the much larger exemptions prior to 1944, uniting these data with the more recent data restricted them to studying only the top decile. However, they innovatively broke out the top decile into progressively smaller fractiles (see table below), where income is defined as gross income excluding capital gains and before individual taxes.

Inspired, I decided to search for recent tax data covering all deciles. The best I was able to find in the short time I spent looking were published by the U.S. Census Bureau as their Current Population Survery (CPS), and the latest CPS data available were for tax year 2008. These data are based a sampling of 117,181 households, sorted into bins incremented by $2,500 up to $99,999, with all of the higher earners (those studied in great detail by Saez) being binned simply as " $100,000 and over." Columns of data were presented for each of 15 definitions of income (plus 3 variants of Definition 1 and one variant of Definition 14) and were broken out only as quintiles.

I played around with graphing the CPS data and finally settled on a comparison of two of my plots, representing the "Definition 3" pre-tax data (where the rich pocketed all their capital gains while the poor were yet to receive government aid) and "Definition 15" (where the capital gains had been taxed and low-income households had become the benefactors of government transfers).

Note in my graphs below that the tall "spike" on the right (centered at $125,000) represents the percentage of households earning $100,000 OR MORE, rather then the percentage earning between $100,000 and $250,000, which would have been consistent with this style of graph. That is, to have been correct, the data for each $2,500-wide bin should have been plotted all the way out to Bill Gates. But this would have required a graph 10 times as long as those shown merely to pick up the first millionaire and 10,000 times as long to pick up the first billionaire on the present linear scale -- thought it could have been done on the same page by use of a logarithmic scale, had all those data been available (in this case they were not).

However, the left-hand "spike" centered on $1,250 is real and indicates that 12.7% of U.S. households were acutely stressed in 2008. Who can live on $100 a month? And what has become of those who became homeless? Have they been taken away to FEMA internment camps?

The top 2008 CPS quintiles (top 20%) are 96% and 87% comprised of the greater-than-$99,999 share for the Definition-3 and Definition 15 cases, respectively, and as listed in my graphs above they account respectively for 51.9% and 40% of the pre-tax and after-tax household incomes. The 2006 study by Saez found the top decile (top 10%) to include all households earning at least $104,700, and thus was 100% comprised of the greater-than-$99,999 share. Saez found this top decile to equal 49.7% of the total U.S. before-tax household income in 2006, a level higher than any other year since 1917.

In the figure below, Saez has decomposed this top decile income share into 3 progressively smaller, yet richer groups using data from 1913 through 2006.

Although less instructive, the two CPS-based bar graphs for 2008 that I've shown above do give a good impression of the results of progressive taxation in combination with government cash transfers to the impoverished.

So maybe the U.S. government is truly altruistic after all? Well, not exactly ...although in the graph below the U.S. is seen to beat out Italy, Greece, and Turkey.

1 comment:

Michael Collins said...

The Saez study is a classic. The data is undeniable. We're returning to a 1927 distribution of income. "Mission Accomplished!"