Do Publications Have Any Responsibility to Screen Their Editorials? (Part 4)

Do Publications Have Any Responsibility to Screen Their Editorials? (Part 4)

At the end of my last post, I said that it would be interesting to see if the Wall Street Journal finally corrects the online version of Brett Stephens's column, "Obama's Envy Problem". As it happens, they have finally done so. I first noticed it when I rechecked the site after my last post late on January 20th. According to a blog post by Paul Krugman titled "Department of Corrections, and Not", the column was still uncorrected when he checked it a few minutes before his posting at 1:40 PM on January 20th. Hence, the Journal may have corrected it shortly after that. I may have initially missed it because they posted the correction at the end of the column. The correction reads as follows:

Corrections & Amplifications

Inflation-adjusted income data from the U.S. Census Bureau show that incomes declined by 2.6% for the bottom quintile between 1979 and 2012, increased by 5.7% for the middle class (third quintile), and by 42% for the top quintile. This column used non-inflation adjusted income figures.

An interesting blog post titled "Bret Stephens and Paul Krugman: What Should a Correction Look Like in the Digital Era?" discusses what an online correction should look like. The Journal did in fact follow the advice not to do a "silent edit" after a column's errors become the basis for other people’s posts. However, there may have been additional reasons that the Journal did not do a silent edit. Once the nominal figures are corrected for inflation, the entire argument put forth by the column pretty much falls apart. For this reason, it might be better if the Journal puts the correction before the column rather than after it. That will save future readers the time and effort of reading it. Coming at the end, the correction should likely begin with the words "Never mind!". Still, it is good to see that the Journal did correct the online column. Hopefully, this might help lead to a policy of doing this on all corrected articles posted by the Journal and other publications. However, that does leave open the question in the title of this post. To what degree do publications have the responsibility to screen their columns and catch these obvious errors to begin with?

Do Publications Have Any Responsibility to Screen Their Editorials? (Part 3)

In my last post, I pointed out that the Wall Street Journal website is still displaying the same editorial with the same non-inflation-corrected numbers for which they had posted a correction. I concluded that it will now be interesting to see if the Wall Street Journal corrects it.

As of this posting, the Wall Street Journal has still not corrected it and I was therefore happy to see Paul Krugman also take up this issue. On January 19th, a Krugman column titled "The Undeserving Rich" was posted on the New York Times website. In it, he stated:

For an example of de facto falsification, one need look no further than a recent column by Bret Stephens of The Wall Street Journal, which first accused President Obama (wrongly) of making a factual error, then proceeded to assert that rising inequality was no big deal, because everyone has been making big gains. Why, incomes for the bottom fifth of the U.S. population have risen 186 percent since 1979!

If this sounds wrong to you, it should: that’s a nominal number, not corrected for inflation. You can find the inflation-corrected number in the same Census Bureau table; it shows incomes for the bottom fifth actually falling. Oh, and for the record, at the time of writing this elementary error had not been corrected on The Journal’s website.

On his blog, Krugman added a post titled "Department of Corrections, and Not" which recounts that Bret Stephens has contacted the Times to protest his contention that Stephens' error in using nominal incomes has not been corrected on the WSJ’s website. Krugman goes on to explain that Stephens is apparently referring to this correction but that that is not what he calls a "correction". He continues:

What, after all, is the purpose of a correction? If you’ve misinformed your readers, the first order of business is to stop misinforming them; the second, so far as possible, to let those who already got the misinformation know that they were misinformed. So you fix the error in the online version of the article, including an acknowledgement of the error; and you put another acknowledgement of the error in a prominent place, so that those who read it the first time are alerted. In the case of Times columnists, this means an embarrassing but necessary statement at the end of your next column.

Nothing like that happened in Stephens’s case. Someone reading his original column on line will see exactly the same piece that was originally published, bogus statistics and all, with no hint of a problem and no link to his mea culpa.(Or at least that was true a few minutes ago — maybe they’ll scramble to fix it now.) Maybe one in a hundred will hear somewhere that there was a problem — but for everyone else the misinformation is continuing to spread.

Krugman need not have worried as there appears to be nobody at the Journal scrambling to fix the original editorial. As of this moment, it has still has not been corrected. By the way, if you attempt to view the original column directly, you will likely just see the first few lines with a prompt to subscribe. However, there is an alternative method that usually seems to view the entire article. Go to google.com, search for "Obama's Envy Problem", and then click on the first item which should be titled "Bret Stephens: Obama's Envy Problem - WSJ.com". This should bring you to the same URL but show the entire article. It may be best to use Internet Explorer for this as Chrome does not always seem to show the whole article. As can be seen, the increase of 186% for the bottom 20%, based on nominal numbers, is still there.

Rather than fix the original editorial the Journal has posted a response by Stephens titled "Stephens: Krugman and the Ayatollahs". In it, Stephens states that "a formal correction was posted on Jan. 5 and I addressed the subject at length on Jan. 3". Can't anyone at the Journal read? Do they truly not understand that Krugman is talking about correcting the original editorial still posted online? Why hasn't the Journal corrected it? A cynical person might suggest the following:

The initial editorial was read by tens or hundreds of thousands of readers and the use of nominal numbers helped to make the case that Stephens and the Journal editorial staff supported. But, as Krugman suggested, the corrections might be seen by one in a hundred of the original readers. I don't subscribe to the Journal so I don't know where those corrections were printed in the paper. But Krugman seems to indicate that a correction was not included at the end of Stephens's next column, a place where his regular readers, at least, would be likely to see it. And by leaving the original editorial uncorrected, the same editorial will be read of additional readers, likely via search engines like Google, and the same flawed numbers will continue to help make the case which Stephens and the Journal support.

Stephens also took his arguments to Twitter, tweeting the following:

Paul Krugman nastily accuses me of "crude obfuscation," in the course of which me makes a dumb factual mistake. See if you can spot it.

As someone tweets back, I think that Stephens meant "he" instead of "me". In any event, following are the final three responding tweets (at this time):

PK with an excellent point. Any "correction" should be done as an addendum to the original article.

Man up, make a legit correction, and get over it.

Why not link to your original column and Krugman's in your acknowledgement of the error? Why not note error in original column?

It will be interesting to see if Stephens responds. What will be more interesting will be if the Journal will finally correct the original editorial or include a correction at the beginning and/or end of it. If they do not, I hope that people with large readerships like Krugman continue to hold them to account. Of course, Stephens and the Journal are free to hold Krugman to account for any uncorrected columns that the Times continues to post online. However, they have not yet shown that they recognize the principle, much less are prepared to follow it.

Do Publications Have Any Responsibility to Screen Their Editorials? (Part 2)

At the end of my last post, I wrote:

It will be interesting to see if Stephens or the Wall Street Journal publish a correction to this egregious mistake. That might give some indication as to whether they are concerned about their reputations or are simply pushing a point of view by whatever means possible.

As it happens, a correction written by Stephens titled "About Those Income Inequality Statistics" was posted on January 3rd. In it, he admits his error in using data that had not been corrected for inflation, stating:

In my Dec. 31 column on income inequality, I used a data set from the U.S. Census Bureau to make the case that incomes in the U.S. have been growing across the board, even if the incomes of the wealthy have grown faster than those of others further down the income scale. But I wrote those lines looking at a set of numbers that had not been adjusted for inflation.

Professor Krugman, in a post on his New York Times blog, takes me to task for this. Had I done so looking at the inflation-adjusted table, it would have shown the incomes of the bottom 20% essentially stagnating since 1979 (and long before then, too), though it also would have shown incomes for the top 20% rising far less dramatically.

He goes on to mention that Krugman accuses him, not of making an honest mistake, but of "pulling a fast one". He then continues:

My mistake is all the more unfortunate because the basic point I was making is right: Americans are getting richer across the entire income spectrum, even if they are getting richer at very different rates. That much is confirmed by data from the Congressional Budget Office. The CBO finds that between 1979 and 2007 income for poor households grew by 18%, for the middle classes by nearly 40%, and for the top 81-99% by 65%. It's the top 1% who have made out very handsomely, with a jump of 275% over nearly three decades.

Hence, Stephens feels that his basic point is still right despite the fact that he was forced to switch from using Census data to CBO data to maintain it. He feels this despite the fact that he goes on to admit that the differences between Census and CBO data is complicated and ultimately subjective. In any event, the CBO numbers calculate that the income of the bottom 20 percent of households increased by 18% over the 28 years from 1979 to 2007. That works out to about 0.6 percent per year compounded.

The idea that this shows that "Americans are getting richer across the entire income spectrum" is highly debatable. As mentioned, Stephens had to go with CBO data instead of Census data. In addition, he is basing this statement on a mere four categories. The CBO data gives no indication as to the variation of income increases in the bottom 20 percent. This is especially the case when you consider the variation in inflation. Those who have been hit by high inflation in health care and higher education likely do not feel "richer". In addition, you would expect incomes to rise faster than inflation due to productivity and the real growth in the GDP. The following graph shows the growth in output and wages since 1950:

As can be seen, wages generally kept up with productivity gains until the 70s but have generally lagged since. The following graph shows wages and corporate profits as a percent of GDP since 1950:

As can be seen, wages as a percent of GDP stayed fairly level to the mid-70s but have generally declined since. Meanwhile, corporate profits as a percent of GDP have risen sharply since about 2001. The actual data and sources for both of these graphs can be found at this link.

In addition to Stephen's correction, the Wall Street Journal has posted a correction that reads as follows:

Inflation-adjusted income data from the U.S. Census Bureau show that incomes declined by 2.6% for the bottom quintile between 1979 and 2012, increased by 5.7% for the middle class (third quintile), and by 42% for the top quintile. Bret Stephens's column of Dec. 31 (Global View, "Obama's Envy Problem") used non-inflation adjusted income figures.

However, there is one major problem remaining. If you look at the original editorial, you'll see that nothing appears to have changed! The Wall Street Journal website is still displaying the same editorial with the same non-inflation-corrected numbers that they have posted a correction for! What is the point of the correction if you're not going to correct the original posting?

It was a positive step for Stephens and the Wall Street Journal to post corrections to this story. However, it will now be interesting to see if the Wall Street Journal corrects their original editorial. That might give some indication as to whether they are truly concerned with correcting their mistakes and limiting their effects or are simply going through the motions.

Do Publications Have Any Responsibility to Screen Their Editorials? (Part 3)

Do Publications Have Any Responsibility to Screen Their Editorials?

On Dec. 30th, the Wall Street Journal ran an editorial titled "Stephens: Obama's Envy Problem". The editorial was written by the WSJ's Bret Stephens and begins as follows:

As he came to the end of his awful year Barack Obama gave an awful speech. The president thinks America has inequality issues. What it has—what he has—is an envy problem.

Further on, Stephens states:

Mr. Obama tried to prove that in his speech, comparing present-day income with that halcyon year of 1979: "The top 10 percent no longer takes in one-third of our income—it now takes half," he said, suggesting that the rich are eating a larger share of the national pie. "Whereas in the past, the average CEO made about 20 to 30 times the income of the average worker, today's CEO now makes 273 times more. And meanwhile, a family in the top one percent has a net worth 288 times higher than the typical family, which is a record for this country."

Here is a factual error, marred by an analytical error, compounded by a moral error. It's the top 20% that take in just over half of aggregate income, according to the Census Bureau, not the top 10%. That figure is essentially unchanged since the mid-1990s, when Bill Clinton was president. And it isn't dramatically different from 1979, when the top fifth took in 44% of aggregate income.

Stephens would have done well to check Obama's sources before trumpeting the superiority of his own numbers. Paul Krugman gives the following explanation on his blog:

What’s going on here? Stephens is citing the Census data, which everyone who knows anything about inequality knows has a problem with very high incomes thanks to “top-coding”. The Piketty-Saez data, which use tax returns to estimate income shares, do indeed show the top 10 percent receiving half the income, up from 42 percent in 1995. Maybe you don’t like those estimates, but Obama made no mistake – while Stephens did.

An article titled "Recent Census Data Significantly Understate The Increase in Income Disparities" explains this in more detail:

In contrast, the Census data are based upon a voluntary population-based survey. They cover virtually the entire population. However, the standard data on income that the Census Bureau publishes do not include capital gains income, which is a large source of income for high-income individuals. The Census Bureau also places an upper limit on the amount of certain types of income it counts for any individual, disregarding income above these amounts; this is done for confidentiality and other reasons. For example, the highest salary that can be recorded is $999,999. Anyone with a salary above this amount is recorded as having a salary of $999,999. As a result of these and other factors, the Census data significantly understate the income of those at the top of the income spectrum as well as recent increases in income among this group.

In fact, Table A3 of a spreadsheet on the Piketty-Saez site shows the income share of the top 10 percent to have risen from 34.21% in 1979 to 50.42% in 2012. Hence, Obama's statement that "The top 10 percent no longer takes in one-third of our income—it now takes half" was correct.

Despite Krugman's contention that "everyone who knows anything about inequality knows [the Census data] has a problem with very high incomes", this error might be understandable. However, Stephens then goes on to make an error that is just astonishing. He continues:

Besides which, so what? In 1979 the mean household income of the bottom 20% was $4,006. By 2012, it was $11,490. That's an increase of 186%. For the middle class, the increase was 211%. For the top fifth it's 320%. The richer have outpaced the poorer in growing their incomes, just as runners will outpace joggers who will, in turn, outpace walkers. But, as James Taylor might say, the walking man walks.

Stephens appears to be using the Census data from this spreadsheet of Table H-3 on the Census website as it exactly matches his numbers. The following numbers are from that spreadsheet:

Table H-3.  Mean Household Income Received by Each Fifth and Top 5 Percent
            (in current and 2012 CPI-U-RS adjusted dollars)

           Lowest   Second    Third   Fourth  Highest    Top 5
    Year    fifth    fifth    fifth    fifth    fifth  percent
-------- -------- -------- -------- -------- -------- --------
Current Dollars
    1979    4,006    9,964   16,428   24,108   43,265   65,984
    2012   11,490   29,696   51,179   82,098  181,905  318,052

% Change    186.8    198.0    211.5    240.5    320.4    382.0

           Lowest   Second    Third   Fourth  Highest    Top 5    Top 5
    Year    fifth    fifth    fifth    fifth    fifth  percent  percent*
-------- -------- -------- -------- -------- -------- -------- --------
2012 Dollars
   1979    11,808   29,369   48,422   71,060  127,526  194,491  214,767
   2012    11,490   29,696   51,179   82,098  181,905  318,052  433,937

% Change     -2.7      1.1      5.7     15.5     42.6     63.5    102.0

Source: U.S Census Bureau, Table H-3 (except last column)

* last column from Piketty-Saez, Table A6

As can be seen, the spreadsheet gives the incomes in "current dollars" and in "2012 dollars". The latter measure adjusts the incomes for inflation, converting all of the numbers to "2012 dollars". The former measure (current or nominal dollars) is NOT corrected for inflation. This is the measure that Stephens saw fit to use. If one uses the figures that are corrected for inflation, however, the incomes of the lowest fifth actually went down 2.7 percent from 1979 to 2012, not up by 186% as Stephens contends.

Regarding Stephens' choice of non-inflation-corrected numbers, Miles Kimball, Professor of Economics and Survey Research at the University of Michigan, writes the following:

It is hard to read the 186% figure in this passage in any way that is not egregiously misleading. The gist of the argument is that a rising tide is lifting all boats, so that the passage seems to suggest that the bottom 20% have been lifted by 186%. Even on the editorial page, the Wall Street Journal’s journalistic standards should be higher.

Correcting dollar figures for inflation is a concept so basic that it is likely known by every living economist. In fact, it is likely known by the great majority of living adults. How could Stephens and the editorial board at the Wall Street Journal have missed it? As Kimball says, it's hard not to read this as "egregiously misleading". It will be interesting to see if Stephens or the Wall Street Journal publish a correction to this egregious mistake. That might give some indication as to whether they are concerned about their reputations or are simply pushing a point of view by whatever means possible.

Do Publications Have Any Responsibility to Screen Their Editorials? (Part 2)