David Cay Johnston of the Times presents some "new government data" about the income distribution in America. However, he can't be troubled to provide a source for this data, and he seems to have made some dramatic errors in his presentation, possibly leading to an understatement of the point he is making. Here we go:
Average Incomes Fell for Most in 2000-5
Americans earned a smaller average income in 2005 than in 2000, the fifth consecutive year that they had to make ends meet with less money than at the peak of the last economic expansion, new government data shows.
While incomes have been on the rise since 2002, the average income in 2005 was $55,238, still nearly 1 percent less than the $55,714 in 2000, after adjusting for inflation, analysis of new tax statistics show.
The combined income of all Americans in 2005 was slightly larger than it was in 2000, but because more people were dividing up the national income pie, the average remained smaller. Total adjusted gross income in 2005 was $7.43 trillion, up 3.1 percent from 2000 and 5.8 percent from 2004.
Since it was only last week that the Times editors demonstrated their unfamiliarity with the difference between medians and means, I am sure you are wondering with me - just what is meant here by "average" income?
In 2005 dollars, median individual income was $24,325 (the mean was $35,499). Neither of those figures is anywhere near the $55,238 reported by Johnston.
OK, so maybe he means household income - the Census report shows quintiles but this WH site says $46,326. The mean would be somewhat higher if the chart for individuals is a useful guide, so I suppose a household mean of $55,238 is possible.
But if Mr. Johnston is referring to household income, why does he write things like this:
The growth in total incomes was concentrated among those making more than $1 million. The number of such taxpayers grew by more than 26 percent, to 303,817 in 2005, from 239,685 in 2000.
These individuals, who constitute less than a quarter of 1 percent of all taxpayers, reaped almost 47 percent of the total income gains in 2005, compared with 2000.
"Individuals"? I doubt it. Unless he means "the individuals living in these households", which is a pretty lame rationalization, but maybe not so lame that the Times won't use it.
Pressing on we see this clue:
The group’s calculations showed that 28 percent of the investment tax cut savings went to just 11,433 of the 134 million taxpayers, those who made $10 million or more, saving them almost $1.9 million each. Over all, this small number of wealthy Americans saved $21.7 billion in taxes on their investment income as a result of the tax-cut law.
Hmm - from this site, we see 125 million individual returns in 2000, no doubt covering more than 125 million people, since many file jointly (I know I did.) From which I infer that when Mr. Johnston says "134 million taxpayers", he is talking about 134 million returns and a lot more than 134 million people.
Bah. I don't know why Mr. Johnston can't tell us where he got this report, I don't know why he can't be clear that he is referring to household income, I don't know why I can't find the report, and I can't comment on his numbers without more background. However - it may even be that Mr. Johnston is understating his populist position when he says that average real income fell by 1%. It appears from the household chart that, in 2005 dollars, the 40th percentile (second quintile threshold) fell from $37,408 to $36,000 while the 60th percentile fell from $59,143 to $57,660. Those are declines of 3.9% and 2.6%, for an average decline of 3.2%. The individual chart is less dramatic - the median moved from $24,390 to $24,325, a 0.3% fall.
Well. *IF* he is referring to mean income and not median income, and *IF* the two measures diverge, we may ask Brad DeLong to employ the lash again.
UPDATE: Here we go, thanks to Paul Zrimsek - it was at the IRS website. Mr. Johnston himself tells us it is available under the "Tax Stats" button from the home page and I am sure it is, but I still can't get there from here.
From the study, we are talking about 134 million tax returns, not individuals. And Mr. Johnston is emphatic that he understands medians and means, and is using means here. In which case, a darker story was available from the use of medians.
And here is a clue I overlooked from the original article:
Nearly half of Americans reported incomes of less than $30,000, and two-thirds make less than $50,000.
Clearly, if two thirds make less than $50,000, he must be reporting a mean income of $55K. That said, "half of Americans" ought to read "half of tax filers" - since filers will be a mix of individual and joint returns, I am not at all clear that this can be directly compared to either individual or household data from the Census.
And let's just plug Mr. Johnston's book as a reminder that he is hardly new to these issues. As he notes in his comment, the dead tree version does have space limitations and not every article can be written and targeted to the most anal-compulsive readers among us.
WE NOTE THE HEADLINE CHANGE: The original headline was "Average Incomes Fell for Most in 2000-5"; since Mr. Johnston would not be responsible for that, I did not belabor the absurdity, but apparently the Times has had second thoughts - the new headline is "2005 Incomes, on Average, Still Below 2000 Peak".
I DISPUTE THE AGENDA NOTION: Via Glenn, we see Mr. Johnston being taken to task for focusing on the 2000 income peak rather than the 2002 trough. Well, if he really had an agenda he would have focused on medians, as described above, for which the news is worse. His use of the average obscures a lot, although it is possible that the 2000 average was unduly inflated by the tech bubble.
WHAT I REALLY THINK: The problem with a study comparing cross-sections of the economy at two different times is illustrated by the absurd initial headline, which read "Average Incomes Fell for Most in 2000-5". Set aside the puzzle about whether "most" of us can earn the "average" income and reflect on this:
A group of people filed tax returns in 2000. Five years later, some of those people had died, retired and had no taxable income, or otherwise fled the tax reach of Uncle Sam. Let's say, hypothetically, that the typical person has fifty years as a taxpayer, so that in five years there would be ten percent turnover. In this guess, 90% of the Tax Year 2000 group is still filing in 2005. And in addition to those filers, a whole new group of high school and college grads as well as new arrivals to our verdant shores will be filing.
So - mathematically it is possible that every last manjack (and womanjack) in the 90 percent who filed in both 2000 and 2005 had a higher income in 2005. If the oldsters who retired and died made more than the newbies to the job market, the average income could still fall.
In which case, the headline would be what - people are dissatisfied even though 90% earn more than five years ago? Hey, that could even be true - maybe folks expected ten percent real wage boosts and only got five percent (on average); maybe new job entrants are disappointed by their current station in life relative to their expectations.
However, the IRS numbers are not helpful in gauging people's expectations, are they? Nor are they helpful in telling me what percentage of the 2000 group had a higher income in 2005.
I would have liked to see a study comparing actual filers in 2000 with the same people five years on, and the IRS is uniquely positioned to do that, since they have access to the individual returns. Then we could learn a bit about income mobility and see whether individuals are really progressing.
They did change the headline, fortunately.