"The Chewbacca Defense"

Once again, Alan Reynolds has asserted that Paul Krugman's claims about the top-coding of Census data are in error. Paul Krugman sends an email with his response:

Here's what I'd say about Reynolds on all this:

Here's what the CBO says: "CBO's adjustments have the biggest impact on high-income households, substantially increasing the income of that group above the levels reported by the Census Bureau." That is, the CBO explicitly states that the income share numbers reported by the Census are lower than those they estimate using their method in part because the Census data are top-coded. End of story.

Reynolds has now made two stabs at this. The first time he started yelling CBO! CBO! without, apparently, noting that the reason I pulled the quote was not to defend the CBO procedures but to show that the people at CBO - who do, presumably, know how the Census numbers are constructed - say that top-coding does reduce the reported high-income share. So, by the way, do the people at EPI (see here). EPI actually does a lot of work trying to correct for the top-coding problem that Reynolds says doesn't exist. Does Reynolds really want to claim that these people don't know what the Census data contain?

On the second stab, Reynolds still does not, as far as I can tell, address the issue of whether I was right to say that top-coding reduces estimates of top incomes. Instead, he tells us that some researchers have access to the full data. That doesn't change the fact that he made a false accusation.

I think the best way to understand what's going on is that Reynolds, rather than admit that he was wrong, is engaging in the Chewbacca defense.

On the broader issue: Reynolds says that various statistical issues have created a false impression of rising inequality. Now, serious researchers, from CBO to the IRS to Piketty and Saez, have looked at those issues, acknowledged them, but concluded that they don't make enough difference to change the picture in any fundamental way. How are those who aren't experts on these data to judge these competing claims?

Well, here's where Reynolds's personal history becomes relevant. Over the years he has repeatedly made demonstrably false accusations about the unreliability of data indicating growing inequality. In each of the cases documented by Brad and others, he did exactly what he did in his first response to my note on top-coding: yelled about how the thing was all wrong without even reading the material he was criticizing. At a certain point you just have to dismiss him as not worth paying attention to.

One last point: we have a number of indicators other than government data on what's happening to very top incomes and wealth - things like estimates of executive compensation. All these indicators point to a continuing rapid rise at the top compared with the middle. So any claim that the rising inequality we see in both Census data and in tax returns is some kind of statistical illusion faces an additional credibility problem.

Update: This is related: Ignorance and inequality. This too: The Rich, the Right, the Facts. Both are from 1992, address very similar issues, and put this into a broader perspective.

Update: Paul Krugman has a  brief follow-up here, and then here.

Comments

[howard]

i ignored alan reynolds second response to the previous posting, but since brother paul has stirred himself, let's note, yet again: reynolds is a propagandist, not an analyst.

he claims that those who notice the increasing income disparities are motivated by psychological affliction, not study of reality.

he is, in short, the perfect wsj oped writer, which is why we find him there so often. it's a shame to see his tripe make its way onto a reputable site like this one.

[pgl]

How did one of my favorite Star Wars characters (Chewbacca) get assosciated with argument by confusion. Then again - most of the claims from the right on these issues (including who pays the tax burden) amount to abuse of the data.

[Sanpete]

From the other thread (as it's currently edited):

Krugman has spoken for himself in the latest post. I've tried to let both Reynolds and Krugman speak for themselves. And they have - and I don't think it supports your conclusion.

My conclusion was that Krugman and (by implication) Mark were wrong about Alan's original comments on this, and that Alan has shown more professionalism in this instance that those he's arguing with. And I still see nothing to change that.

This is the Krugman quote that started this:

The [Census] questionnaire is “top-coded”: if the individual interviewed has earnings higher than $999,999, those earnings are recorded simply as $999,999. Since a lot of income growth in the last few decades has taken place among people with multimillion-dollar incomes, the Census data miss an important part of the story.

Alan correctly pointed out that the questionnaire isn't top-coded but that the public-use data are, and that the census data only miss an important part of the story if the are used in their privacy-censored form, or without correction for that.

As far as I can tell, Krugman was the one to turn to the CBO. Krugman's latest comments relating to that don't contradict in any way I can see anything Alan actually said. I can't find the false accusation Krugman refers to. The reference to the Chewbacca defense is completely gratuitous, from what I can see. And the rest is a personal attack that, as far as I'm concerned, has no place here. I'm very disappointed in Krugman, whom I've previously had great respect for.

[Mark Thoma]

Sanpete:

You will find you are able to say what you want to say here - so feel free to make your points.

[Sanpete]

Mark, if I seemed to imply otherwise I apologize--the editing I was referring to was that of your own post, which I think changed between the time I started my response to it and when I went back to quote it. I don't object to that; just found it disconcerting.

[Mark Thoma]

I don't do that very often at all - though I do edit my comments for clarity. In this case, as I said in the previous post, I decided I didn't want to open that particular avenue.

[wcw]

Sp, while I agree with you that Mr. Reynolds has exhibited a professional demeanor, I find the Chewbacca-defense critique to be anything but gratuitous. If you click through to the linked comment, you will find Mr. Reynolds citing a paper, misunderstanding its main finding, cherrypicking pullquotes and misunderstanding a regression. He subsequently suggests that his critics rely on faith in credentials, all the while failing to answer the very simple questions I've been asking him about the data.

The only thing that would make this more like South Park's depiction of Johnny Cochrane would be if Mr. Reynolds were black. Is he? I assume not, given his employ, but you never know.

[anne]

The problem is definitely not with Paul Krugman. There is no question but the wealth and income inequality is increasing, and the increase becomes starkly clear when looking to the gains wealthiest. While IRS data allows for readily looking at the gains of the wealthiest, census data does not.

Paul Krugman may a slight mistake as far as I can tell in that census income data allows for recording incomes above $1 million, but the mistake is not important because census analysis does not expressly account for the wealthiest in analysis.

Alan Reynolds is wrong and arguing unfairly. Paul Krugman is right. Saez and Piketty and David Cay Johnston are right.

Again, the idea the wealth and income inequality is increasing is clear as can be and only for deception is the argument otherwise.

Thank you, Paul Krugman and Mark Thoma and Brad DeLong.... I never questioned the conclusion, for I am not a fool.

[maria]

Reynolds is clearly wrong and probably incompetent. So how soon we can expect MG to come to his rescue and defend him with a 5000+ word post? LOL.

[Sanpete]

WCW, the first paragraphs of Alan's response dealt with the issues raised in the topic post, directly, clearly, succinctly. That's not evasion or obfuscation. You ignore that in the opening of your earlier response, where you agree with those charges against him, as well as here.

Alan went further and included some argument related to the issue of the census data that he apparently believes supports his overall thesis, which is where you find problems. But those points aren't about the topic of this thread or the previous one. Even if your technical points are correct, it isn't clear to me from what you say that Alan misunderstood the paper; possibly he was cherry-picking, something I find often enough on all sides of most issues. I think some of your questions were a little vague or too broad, so I'm not surprised they weren't all answered. As for his dig about faith, it may not apply to you, but I think we can see it well enough in some of the comments, which are long on trust in some people, or long on personal attacks, and short on particulars about the issues at hand.

[anne]

http://www.nytimes.com/2006/01/29/national/29rich.html?ex=1296190800&en=784822e4b0735ee5&ei=5090&partner=rssuserland&emc=rss

January 29, 2006

Corporate Wealth Share Rises for Top-Income Americans
By DAVID CAY JOHNSTON

New government data indicate that the concentration of corporate wealth among the highest-income Americans grew significantly in 2003, as a trend that began in 1991 accelerated in the first year that President Bush and Congress cut taxes on capital.

In 2003 the top 1 percent of households owned 57.5 percent of corporate wealth, up from 53.4 percent the year before, according to a Congressional Budget Office analysis of the latest income tax data. The top group's share of corporate wealth has grown by half since 1991, when it was 38.7 percent.

In 2003, incomes in the top 1 percent of households ranged from $237,000 to several billion dollars.

For every group below the top 1 percent, shares of corporate wealth have declined since 1991. These declines ranged from 12.7 percent for those on the 96th to 99th rungs on the income ladder to 57 percent for the poorest fifth of Americans, who made less than $16,300 and together owned 0.6 percent of corporate wealth in 2003, down from 1.4 percent in 1991....

[anne]

http://www.nytimes.com/2005/06/05/national/class/HYPER-FINAL.html?ex=1275624000&en=f1af44c9cec8c79e&ei=5090&partner=rssuserland&emc=rss

June 5, 2005

Richest Are Leaving Even the Rich Far Behind
By DAVID CAY JOHNSTON

When F. Scott Fitzgerald pronounced that the very rich "are different from you and me," Ernest Hemingway's famously dismissive response was: "Yes, they have more money." Today he might well add: much, much, much more money.

The people at the top of America's money pyramid have so prospered in recent years that they have pulled far ahead of the rest of the population, an analysis of tax records and other government data by The New York Times shows. They have even left behind people making hundreds of thousands of dollars a year.

Call them the hyper-rich.

They are not just a few Croesus-like rarities. Draw a line under the top 0.1 percent of income earners - the top one-thousandth. Above that line are about 145,000 taxpayers, each with at least $1.6 million in income and often much more.

The average income for the top 0.1 percent was $3 million in 2002, the latest year for which averages are available. That number is two and a half times the $1.2 million, adjusted for inflation, that group reported in 1980. No other income group rose nearly as fast.

The share of the nation's income earned by those in this uppermost category has more than doubled since 1980, to 7.4 percent in 2002. The share of income earned by the rest of the top 10 percent rose far less, and the share earned by the bottom 90 percent fell.

Next, examine the net worth of American households. The group with homes, investments and other assets worth more than $10 million comprised 338,400 households in 2001, the last year for which data are available. The number has grown more than 400 percent since 1980, after adjusting for inflation, while the total number of households has grown only 27 percent.

The Bush administration tax cuts stand to widen the gap between the hyper-rich and the rest of America. The merely rich, making hundreds of thousands of dollars a year, will shoulder a disproportionate share of the tax burden.

President Bush said during the third election debate last October that most of the tax cuts went to low- and middle-income Americans. In fact, most - 53 percent - will go to people with incomes in the top 10 percent over the first 15 years of the cuts, which began in 2001 and would have to be reauthorized in 2010. And more than 15 percent will go just to the top 0.1 percent, those 145,000 taxpayers.

The Times set out to create a financial portrait of the very richest Americans, how their incomes have changed over the decades and how the tax cuts will affect them. It is no secret that the gap between the rich and the poor has grown, but the extent to which the richest are leaving everyone else behind is not widely known....

The analysis also found the following:

¶Under the Bush tax cuts, the 400 taxpayers with the highest incomes - a minimum of $87 million in 2000, the last year for which the government will release such data - now pay income, Medicare and Social Security taxes amounting to virtually the same percentage of their incomes as people making $50,000 to $75,000.

¶Those earning more than $10 million a year now pay a lesser share of their income in these taxes than those making $100,000 to $200,000....

[anne]

http://www.nytimes.com/2005/06/05/national/class/HYPER-FINAL.html?ex=1275624000&en=f1af44c9cec8c79e&ei=5090&partner=rssuserland&emc=rss

One way to understand the growing gap is to compare earnings increases over time by the vast majority of taxpayers - say, everyone in the lower 90 percent - with those at the top, say, in the uppermost 0.01 percent (now about 14,000 households, each with $5.5 million or more in income last year).

From 1950 to 1970, for example, for every additional dollar earned by the bottom 90 percent, those in the top 0.01 percent earned an additional $162, according to the Times analysis. From 1990 to 2002, for every extra dollar earned by those in the bottom 90 percent, each taxpayer at the top brought in an extra $18,000....

[anne]

http://www.nytimes.com/2007/01/01/us/01deaths.html?ex=1325307600&en=21d5790bcc52e5fd&ei=5090&partner=rssuserland&emc=rss

January 1, 2007

3,000 Deaths in Iraq, Countless Tears at Home
By LIZETTE ALVAREZ and ANDREW LEHREN

Jordan W. Hess was the unlikeliest of soldiers.

He could bench-press 300 pounds and then go home and write poetry. He learned the art of glass blowing because it seemed interesting and built a computer with only a magazine as his guide. Most recently, he fell in love with a woman from Brazil and took up digital photography, letting both sweep his heart away.

Specialist Hess, the seventh of eight children, was never keen on premonitions, but on Christmas of 2005, as his tight-knit family gathered on a beach for the weekend, he told each sibling and parent privately that he did not expect to come home from Iraq.

On Nov. 11, Specialist Hess, 26, freshly arrived in Iraq, was conducting a mission as the driver of an Abrams tank when an improvised explosive device, or I.E.D., blew up with brain-rattling force. The blast was so potent it penetrated the 67-ton tank, flinging him against the top and critically injuring his spine. His four crewmates survived. For three weeks, he hung on at Brooke Army Medical Center in San Antonio, long enough to utter a few words to his loved ones and absorb their kindness.

On Dec. 4, Specialist Hess slipped onto the ever-expanding list of American military fatalities in Iraq, one that has increased by an average of more than three a day since Oct. 1, the highest three-month toll in two years. On Sunday, with the announcement of the death in Baghdad of Specialist Dustin R. Donica, 22, of Spring, Tex., the list reached the somber milestone of at least 3,000 deaths since the March 2003 invasion.

The landmark reflects how much more dangerous and muddled a soldier's job in Iraq has become in the face of a growing and increasingly sophisticated insurgency. Violence in the country is at an all-time high, according to a Pentagon report released last month. December was the third deadliest month for American troops since the start of the war, with insurgents claiming 111 soldiers' lives. October and November also witnessed a high number of casualties, 106 and 68 respectively, as American forces stepped up combat operations to try to stabilize Baghdad.

"It escalated while I was there," said Capt. Scott Stanford, a National Guard officer who was a commander of a headquarters company in Ramadi for a year, arriving in June 2005. "When we left this June, it was completely unhinged. There was a huge increase in the suicide car bombs we had. The I.E.D.'s were bigger and more complex."

"And it was very tense before we left in terms of snipers," said Captain Stanford, a member of the Iraq and Afghanistan Veterans of America. "I don't know if there were more of them, or if they were getting better."

This spike in violence, which has been felt most profoundly by Iraqi civilians, who are dying by the thousands, has stoked feverish debate about the nation's presence in Iraq. Many Democrats in Congress are urging a phased withdrawal from the country, and the Bush administration is leaning toward deploying additional troops in 2007. If the conflict continues into March, the Iraq war will be the third longest in American history, ranked behind the Vietnam War and the American Revolution.

President Bush did not specifically acknowledge reaching the milestone of 3,000 American deaths, but a White House spokesman, Scott Stanzel, said the president "grieves for each one that is lost" and would ensure that their sacrifices were not made in vain. The campaign against terrorism, Mr. Stanzel said, will be a long struggle....

[anne]

Forgive me, I posted the above article on the wrong thread. The proper memorializing thread was: The "7-Year-Old's Soccer Syndrome."

[wcw]

Anne, could you pretty-pretty-pretty please cease posting entire NYT articles? You could simply link to their content, copied to free geocities pages. You would make the same point, without clutter. Note how well links work for me below.

Please do so. I for one would be extremely grateful.

Sp, that first paragraph is not responsive in the least. Mr. Reynolds claim was that Krugman was "entirely wrong" about censoring in the CPS data, which claim he defends in pars. 2-3. I accede that in that latter, he does "directly, clearly, [and ]succinctly" admit that, "the Census strips out such information (as well as addresses) before releasing public use files," hardly a defense of the original claim. I find that admitting error only implicitly and only in your third paragraph shows that your "preference is for obfuscation over illumination." What convinces you otherwise?

On the issue of cherrypicking, scan this PDF with charts from that paper, or read the whole thing (MS-Word DOC). The abstract should convince you he misunderstands the paper. The Gini trend should do the same for income inequality. The regression specifications should explain my critique. The discussion should limn his cherrypicking.

If he's a personal friend ("Alan"), may I gently suggest you encourage him to tilt at some other ideologically satisfying windmills? He's embarrassing himself here.

[anne]

No, how is that? The article portions posted are in direct response to the complaint that the issue of whether inequality is growing was not being expressly addressed. the issue has been repeatedly addressed and will continue to be for imbalances in economic influence and tax structure will insure increased inequality in wealth and income. Also, there is a constituency that is influential, that supports denial of increasing inequality, and denial will continue and will need to be countered.

But, I did and do apologize for mistakingly posting a portion of the difficult article on casualties on the wrong thread.

[wcw]

The articles are someone else's writing, not yours. By all means, cite them, use a telling quotation or two, then to link to the full content elsewhere.

Please.

Mr. Reynolds may be a cherrypicking statistical lightweight, but at least he knows how to cite a paper so that those with interest can find and read it themselves.

[Sanpete]

Anne, the point your articles address isn't the point at issue in this thread, which is about the ridiculous top-coding kerfuffle.

WCW, I'm having trouble following your meaning. Alan's first paragraph explains why he objects to Krugman's comment that the Census data is limited, which is a confusing claim, at the least. Can't see why you think the paragraph wasn't responsive.

This next point has already been made several times, so it's hard to believe that it's still necessary to make it again, but I can't understand what you say otherwise. Apologies in advance if I'm just missing your point. Krugman's remark was about the census *questionnaire*, which he claimed was top-coded, so that the highest incomes aren't recorded. It isn't; they are. The income data is collected, and is used for some purposes. This is clearly what Alan was referring to from the start and, again, is what Alan explains in what you apparently take to be some kind of admission of error on his part, which it isn't at all. What he said is obviously an explanation and defense of his original claim that Krugman was just flat wrong on this. Which he was. It's Krugman who has yet to acknowledge his own error, and who has misread the whole issue rather outrageously.

As Alan pointed out, the mistake by Krugman had already been pointed out here by others, with no objection, but when Alan repeated the point, it was suddenly treated as moronic. I think it's very likely that some strong non-intellectual factor has clouded people's reading and reasoning. I don't think he should be the red-faced one in this case.

I still don't see the evidence that he misunderstands the paper, though I still see the case for your charge of cherry picking. It appears to me his comments were limited to the aspects of the findings that he finds congenial to his goals.

I don't know the man at all. I refer to him as "Alan" because he had posted in the same thread I did, and blogs are informal that way. If Krugman had done the same, he'd be Paul to me.

[anne]

Well, tomorrow I will ask reference to try to run down a census "questionnaire" because suddenly I realize I am not even sure what is being referred to. The census is handled by interview and questionnaire, and I seem to remember there is a difference. But, if there was a mistake, as I thought, the mistake should not effect Paul Krugman's conclusions on data analysis by census and IRS.

[Richard]

"Really though, as far is went, the stategy of the initial conflict of the Iraq War worked."

But the mistake is to not see the reconstruction as part of the strategy; indeed, as the essential element. It is not to capture territory, but hearts and minds that is crucial.

And think of how bloodlessly the Eastern European countries left communism; think of how the Philipines averted the continuing dictatorship of Marcos. We are at our best (and at our strongest) when we use non-military strategies.

For democracies, the use of force is almost always an admission of weakness and failure. And it is better to attend to that, to find the right way to struggle with policy, than it is to turn a blind eye to the use of the military and to then focus on the "techne" of force.

[anne]

Oh, by the way, though I think Alan Reynolds entirely wrong about the tendency to inequality we are experiencing and the problem such a tendency will bring, I do not question his credentials and would never think to or his ability to handle the analysis. I think the argument useful, even though deceptive.

[anne]

Richard, like me you may have posted on the wrong thread, but your comment on resolution of conflict short of using force is quite important and needs to be expanded on.

[Mark Thoma]

anne:

If it's at all helpful, the earliest reference I know of where Paul Krugman talks about this is in 1992 in Ignorance and inequality. At that time, the top was $125,000.

But anyway, if you read this, you'll see it's nothing new. Remember, this is 1992. Here's a quote:Poor reading. The Fed study has also been attacked. Conservative polemicist Alan Reynolds argued in a Wall Street Journal Op-Ed article that since the top 1 percent of a sample of 3,000 families contains only 30 families, the results were unreliable. In fact, the Fed used a sophisticated, two-stage sampling procedure that allowed it to ''oversample" the wealthy; the estimate of average wealth for the top 1 percent was actually based on more than 400 families. The important point is that the Wall Street Journal printed a column attacking the Fed study written by an ''expert" who apparently didn't understand the report he was criticizing.I just noticed this. It explains more about Paul Krugman's reaction. This too: The Rich, the Right, the Facts

[anne]

Darn, do I ever understand.

Thank you so much, Mark.

[wcw]

Sp, the problem is that Mr. Reynolds (whose own formalism I echo) describes the rather minimal error of ascribing to the questionnaire itself what Reynolds well knows is quite true of the public-use data as "entirely wrong." I do not know who else pointed out the error, but my suspicion is that the correction lacked that polemical and inaccurate flair. Of same, agitprop critiques like "Chewbacca defense" are perfectly well-specified.

On the minutiae, the census footnotes are confusing. Cf #23, which relates Data collection method changed from paper and pencil to computer-assisted interviewing. In addition, the March 1994 income supplement was revised to allow for the coding of different income amounts on selected questionnaire items. Child support and alimony limits decreased to $49,999. Limits increased in the following categories: earnings to $999,999; social security to $49,999; supplemental security income and public assistance income to $24,999; and veterans' benefits to $99,999.

I had as a result always believed that the data were themselves censored, especially given the early-'90s discontinuity in the quintile tables. Were they constructed from the public data? I do not know. Perhaps our intrepid interlocutor could illuminate the issue. I am happy to believe he can do more than obfuscate. Let's have it, sir.

[Sanpete]

what Reynolds well knows is quite true of the public-use data

But what's true of the public use data isn't the point either.

WCW, we have different ideas about what constitutes polemic and inaccurate flair here, and apparently also about who would be more guilty of that in this case. Whether it's worse by that measure to call a false statement "entirely wrong" or to muddle the response and resort to personal attacks on other topics as Krugman did I'll leave to you. We also disagree about the supposed aptness of the Chewbacca remark, which seems to me to apply more readily to Krugman's comments than to Alan's.

I will say this much. The truth of the details of the data recording and processing isn't as plain as I thought it was. Krugman probably wasn't as far off as I thought in what he probably was trying to say, and Alan probably wasn't as correct as I thought either. "Top-coding," properly speaking, is something that is done to the data after it's collected and before it's made public, and isn't a matter of what's on the questionnaire. However, the CPS questionnaires, and the data processed from them, do have cut-off top values as well. The top amounts since 1993 or 1994 are variously reported as $9,999,999 recording/1,099,999 processing, in a piece by Welniak on how to use the CPS data (page 10 table 3, cp page 4); or $1,099,999 and $999,999 (for different types of income, presumably both figures are for processing limits), in Burkhauser et al (Table 4, the title of which is in error as to years covered); or possibly just $999,999, if the footnote you cite means what it seems to (I think it may only refer to 1993). According to Welniak, who is the most explicit on this issue (and whose paper is also relevant to the larger issue of income inequality comparisons over time), the number of cases that exceeded the processing limit in 2000 was 26 (out of about 78,000); the number that exceeded the recording limit was zero.

Krugman would be wrong to think the data needed just isn't there, as he says. Alan would be right that incomes well above $1,000,000 are recorded, and ought to be sufficiently accurate for most years. But apparently the limit used for in-house processing isn't far higher than that $1,000,000, so I'd guess he's mistaken if he thinks the most correct data is actually used in the figures he mentions, as he seems to believe, and Krugman would be close to right if he were referring to the processed data rather than the collected data. In any case I was wrong earlier when I wrote as though there were no upper limit in the recorded values. That's all I know about it.

[Mark Thoma]

Appreciate the information - thanks.

[Bruce Webb]

All Left critics of Right Economics are malevolent misrepresenters of obvious ecomonic reality who may on occassion in their blind pig ramblings around the numeric woods occassionally find a $9,999,999 acorn. But they are WRONG, DEAD WRONG! AAAAH! OOOOHHHH!!! Well that is the Chewbacca Defense.

And that in a nutshell is the problem. I am sure that in his hundreds of columns over the years Prof. K mispoke or misstated somehow and somewhere. But there has been a tendency in the last couple of weeks for Right 'Economists' like Mankiw, Sowell and Reynolds to suggest to suggest that the Economic Left are simple doofuses who don't understand the most obvious economic realities.

Well no. We may be wrong, there might be another explanation for the economic numbers, but the days of rolling boulders down on us from the Austrian Alps are over. Uncle Miltie is in the grave and the Economic Right needs to bring and explain numbers. I know that the Economic Right's model assumes 1.7% ultimate productivity. I know what 1.7% ultimate implies for 7.5% returns from equities. That Don Marek or his intellectual betters are dimly becoming aware that they are responsible for the numbers underlying their polemics is their problem.

I don't know whether I actually came up with this catch phrase back in the day. I think so but I don't want to crown myself another Emperor Misha II. But for what it is worth:

"This is dKos. You don't get to simply make shit up."

These are the econoblogs. You can make a case that higher marginal tax rates somehow crowds out investment. Or try. But for God's sake bring numbers and get them right. And don't hurl monkey feces at the dinner party before you do.

Reynolds and PRS and Marek have continually resorted to cheap name calling and lazy factoids. That those factoids occasionally veer close to something we call "Reality" doesn't give them a Get Out of Jail Free Card on something/everything. The Right invented a little term called "Fisking", which in origin made running everything Robert Fisk ever reported on Iraq through their own Reality Lens and laughing their asses off. Well that Reality Lens has proven to be seriously distorted, and most of Fisk's reporting has held up remarkably well.

So to return to Reynold's and economic reporting: Simply asserting that all markets are universally and ultimately efficient because Uncle Miltie met Herr Mises in the Austrian Alps and agreed that it was so doesn't cut it anymore. Top level generalizations need to be consistent with ground level data. Yes it is hard work, and yes it would be easier to eat surrender monkey cheese and whine at the Cato/Club for Growth Annual Picnic. But this lazy ass "Prof K is an idiot. So There." line of attack is not going to work. Razor thin logic chopping to show that while our general point is perfectly correct, in certain circumstances we missed a nuance in the data and therefore you get a free pass to float implausible economic numbers, well those days are over.

You don't have to like 1.7% ultimate. You don't even have to know what that means or what it implies for returns on equities. But you better know that we know.

And BTW where DID Reynolds go to school? Mark may be reticent to fling the goods around, and I would be the last person to tell you you are defined by your CV. Which doesn't mean its not relevant.

[maria]

"And BTW where DID Reynolds go to school?" As far as I can tell he first went to UCLA and then to SacState. In neither place, as far as I can determine, did he get a degree.

Now, more importantly, perhaps:

Take note of Pearlstein's insight, elaborated by Brad Setser, that one of the driving forces behind increasing inequality is the trade deficit (and there is no doubt about that, right? LOL). It serves to depress wages of labor that compete with cheap foreign labor (think China, but also others) while at the same time it increases demand for US financial assets that are held domestically overwhelmingly by the rich. And by depressing interest rates, it serves to keep those financial assets levitated.

[anne]

The fight seems to have begun in March 1992 and continued just as now to June 1992:

http://www.pkarchive.org/economy/DisparityDespair.html

March 23, 1992

Disparity and Despair
By Paul Krugman - U.S. News & World Report

http://www.pkarchive.org/economy/IgnoranceInequality.html

June 1, 1992

Ignorance and Inequality
By Paul Krugman - U.S. News & World Report

Paul Krugman has been noting what is easily evident, that wealth and income inequality have been increasing. While Alan Reynolds has been countering that what looks to be increasing inequality is only a poor reading of and use of data.

Paul Krugman has been and is right, wealth and income inequality has been and is increasing. The increase becomes expecially evident when the wealth and incomes of the richest are looked at and there is a minor problem. Much census data and analysis involving income does not allow for examination of the richest. When analysts use data that does allow for examination of the richest, the complaint is census data should be used and the rest is faulty. Nonsense.

Emmanuel Saez and Thomas Piketty and David Cay Johnston have given us data and analysis that clearly shows precisely what Paul Krugman has long been arguing, shows what I know simply for paying attention to portfolio analysis and tax structure, wealth and income inequality is increasing.

[anne]

Paying attention to portfolio analysis and tax structure, tells me with no doubt at all, that data analysis for 2005 and 2006 will show pronounced increases in wealth and income inequality. The long term trend that Paul Krugman has so readily identified from 1992 at least, that is so well supported by analysis from different data streams, continues.

"The Chewbacca defense" is to no avail.

[Bruce Webb]

Without necessarily agreeing with you on the trade question I will say that for anyone familiar with the structure of California Public Universities going from UCLA to Sac State raises a lot more questions than it answers. Going from Chico State to any UC campus to a Professorship at U of O is a trajectory up. Going from a UC campus that offers PhD programs to a Cal State campus that doesn't is to plummet. For anyone whose last port of call was Cal State Sacto to be passing judgement on a Princeton Professor just begs the question of what 'Alan's' (we are not first name buddies) qualifications are.

As I said all too many commentators coming from the Right are all too willing to roll down boulders from the Austrian Alps rather than presenting actual argumentation.

And for that matter a little less "(and there is no doubt about that, right?" with or without the "LOL)" would be welcome. Because no I don't know for sure that the explanation for income inequality is due to the trade deficit. In my opinion economists and economic commentators should be striving for a little less dogmatism and a little more impiricism. Most of my problem is with the "It's so obvious contingent"

"Property is theft" is a childish slogan and not a program for action, and I have no intention of putting up with slightly more nuanced statements of that just because the brickbats are coming from the Left and pouring down on Marek's head,

[maria]

Bruce: I am not an economist and not competent to do intricate analysis of complex economic questions so like most non-experts I have to rely on the apparent qualifications of the "expert" in question. Now as to the trade deficit factor: this is not my idea. It was stated by Pearlstein in, as I recall, the Washington Post, and taken up by Brad Setser who endorsed it. I have confidence in the expertise of Brad Setser, as I do in that of Paul Krugman, and as I do not in Alan Reynolds.

[Sanpete]

This is a case of people reading what they want to read into things and ignoring the facts in the process. Krugman accuses Alan of the Chewbacca defense in connection with the top-coding issue, where Krugman was clearly wrong and has yet to recognize it. This thread and the previous one illustrate the ways in which complex economic and political issues are affected by nonrational factors. Fairly plain statements are misread, much misdirection and personal attack is substituted for relevant analysis and facts. That Alan may often be the one guilty of such things, if I'm to believe the attacks here, hardly justifies the fact that it's his opponents here who have been in the wrong in this case, and who have been guilty of the very things they accuse him of, including the Chewbacca defense.

Maria, I'm sure I have no more economics education than you do (one class about 30 years ago), but I still find it necessary to deal with the issues rather than the degrees if I want to have the best chance of knowing who's right and who's wrong. You'll never beat Milton Friedman's credentials, but I won't accept his word on that basis. Neither is it a good idea to take sides on a specific issue like top-coding on the basis of who has the best formal education. If you don't understand the issues well enough to examine the arguments, I think it's better to maintain some skeptical distance than to just line up with the most advanced degree or reputation. Usually there is no cost to that approach.

[anne]

The answer to the top-coding matter, is that the Census Bureau both uses paper questionnaires that limit the range of incomes that can be entered and personal interview questionnaires that allow for any income to be entered. Published census data and analysis does not allow for income trends among the wealthiest of households to be isolated. IRS data is need to isolate income and wealth trends among the wealthiest.

There really is no problem with Paul Krugman's synopsis, nor is there reason to question what Saez and Piketty or Johnston find as growing wealth and income inequality from separate IRS data analyses.

[anne]

When I mentioned David Cay Johnston's analysis of IRS data reflecting wealth, the analysis was immediately attacked as based on incomplete data, since activity in tax sheltered accounts does not reflect asset ownership. No matter, the control of wealth reflected through IRS accounts is acute and astonishing enough under any circumstances. When 1% of households control 57.%% of corporate shares, as they did by 2004, we have inequality that is acute.

[anne]

As with evolution or global warming, wealth and income inequality however will be argued away by those who wish to so argure no matter the evidence. Charles Darwin has only been confirmed these last 150 years, so who can tell what tomorrow will bring although every evolution denying school board member in Kansas running for election was just defeated in this remarkable election. Oh, and California really really is taking global warming seriously.

[Sanpete]

Anne, we understand top-coding as it applies specifically to this little Krugman/Reynolds feud differently. I won't repeat all the details; they're in my post from last night.

says...

speaking of "blind pig ramblings", we have this stupidity:

BWebb -- And BTW where DID Reynolds go to school? Mark may be reticent to fling the goods around, and I would be the last person to tell you you are defined by your CV. Which doesn't mean its not relevant.

maria -- "And BTW where DID Reynolds go to school?" As far as I can tell he first went to UCLA and then to SacState. In neither place, as far as I can determine, did he get a degree.

BW, did you not read the last thread?? MG explained that A. Reynolds already cleared that up and he provided the previous three threads of the discussion. Go read the post.

maria is playing the dumb and innocent here. Dumb, yes. Innocent, of course not. She was told where to look for AR's info on his graduate studies posted on this blog. She was already advised on where to find AR's response.

Talk about blind pig ramblings. Too dumb to read what is not only available but pointed out to the sidetrack harrassers.

MG also cleared up some of Webb's previous misinformation about Sacramento State graduate and PhD program offerings.

And PK doesn't show any appreciable knowledge about Census questionnaires. Otherwise, he wouldn't have made those dumb statements. To make matters worse, he has failed to own up to his error and state the correction.

People can accuse AR and others of quibbling over PK's first statement, but PK looked the part of the fool in skirting around his stupid mistake never correcting it. PK doesn't earn any credibility points for playing the game he played on this one. Nor do some of his defenders.

What is maria's college and degree, by the way? Cat got your tongue, maria?

[maria]

"Maria, I'm sure I have no more economics education than you do (one class about 30 years ago), but I still find it necessary to deal with the issues rather than the degrees if I want to have the best chance of knowing who's right and who's wrong."

So you think you have all the expertise needed to deal with the issues yourself? I don't give myself that much credit as an economist or statistician. And when two people hold diametrically opposed positions, I go with the one who has the better intellectual credentials and who is not employed by a think tank with an "attitude" and an agenda. Why should I chose the other person? You tell me.

[anne]

Sanpete, I read your comments carefully and think the only difference is on emphasis. I do think I understand and appreciate your complaint, and I learned just why and how Saez and Piketty have used the IRS data for analysis. Thank you for being so precise and clear. I tried to be, but may not have been so.

[maria]

I might further add that as far as I can determine an overwhelming number of competent economists (and statisticians) believe that inequality is increasing and has been doing so for some time. Only a very few on the fringe think the opposite like Reynolds. I view this as a common sense matter just as with global warming. Since the vast majority of competent geographers and climatologists believe in global warming, I would believe them rather than the handful with dubious credentials who claim the opposite.

[Paul Krugman]

Hi guys. I haven't been checking this forum, but I just looked, and there seems to be some confusion, even an impression that I was somehow evasive or wrong on some point. As far as I can tell, what I said in Times Select was completely right.

I have no idea why there have been some confident assertions here that the questionnaire itself isn't top-coded, but the Census Bureau's own publications say it is. The key term is "reporting limits." Look at footnote 4 from this paper:

http://www.census.gov/hhes/income/incineq/p60204/p60204txt.html

"The Census Bureau introduced computer-assisted personal interviewing (CAPI) in January 1994 to the Current Population Survey. The March 1994 supplement permitted households to report up to $1 million in earnings, up from $300,000, and we made parallel increases in the reporting limits for selected other income sources. Both of these changes affected the data."

This doesn't say that the Census has the full data, but truncates it; it says that there's a limit on the income you can report. And the paper, if you read it, also says that revising this limit affects inequality measures like the Gini coefficient.

Paul Krugman

[maria]

Let me approach the matter from a different direction. Do you know of any eminent economists (or perhaps better, any ones employed by a first rate university)who agree with Reynolds? Not all eminent economists are leftist by any means. Take Harvard. Does any economist on the faculty support Reynolds? Barro? Feldstein? Freeman? Mankiw? Summers? Or Chicago? Berkeley? Yale? Princeton? Columbia? If you can name an eminent economist who supports Reynolds I would like to know of him.

[Sanpete]

Maria, I don't have the expertise to settle all points (to put it mildly), but I can read English fairly well. In this case I had little trouble following the arguments and finding the information relevant to settling things to my satisfaction. No economics degree was needed. Again, the topic here was the spat between Alan and Krugman over top-coding, not their larger dispute over inequality. My English skills got me that far.

As I said before, you're generally under no obligation to take a stand on issues you can't follow for yourself enough to get a good idea how solid the views are. If you just recognize you don't know something and withhold judgment, that's seldom fatal. This is top-coding minutia, not global warming.

As for attitude, I think Krugman has his share. I suspect it's just more to your liking.

[maria]

I am no particular expert on the Gini Index but nonetheless have used it in some historical work I have done. From what I do know, if top incomes are underreported, or say people with incomes over one million are recorded as having incomes of one million and no more, then if their actual income were to be recorded, the Gini Index would rise. In other words, capping or limiting the reporting of top incomes has the tendency to understate the degree of inequality.

[maria]

"As for attitude, I think Krugman has his share. I suspect it's just more to your liking."

No, his reputation is what is more to my liking. And the fact that his employer does not have an "agenda" or party line on social questions as does Cato. Do you know of any eminent economist who agrees with Reynolds on the matter of whether inequality is rising or not? After all that is what the real issue is.

[Mark Thoma]

Thanks Paul K. - appreciate having this cleared up.

In case anyone misses it, Paul Krugman left a comment not too far above this one explaining why what he said in Times Select is correct.

Mark

[anne]

Thank you, Paul Krugman. Forgive my confusing your original New York Times explanation. I will send along the Census explanation to the kind reference specialist who earlier sent me a sample questionnaire. I had thought the Census Bureau included all sampled incomes in analysis, but I did not understand the all samples are limited in incomes reported, and I was simply wrong. You are a gem.

[anne]

Mark Thoma, thank you for persisting since I am evidently still learning to read. Never did it occur to me however that Paul Krugman was being evasive. I simply did not understand for too long what appeared inclusive Census analysis. There is a reason my sister is the attorney.

[wcw]

After digging way more into this than I had had any intention of doing, it appears to me that Welniak is the authority here.

The paper that Reynolds cites, Burkhauser and Feng (2006), summarizes,even the internal CPS data, which are not subject to top coding, have been censored to various degrees over time (Welniak, 2003). Welniak (2003) expounds,In addition to an increasing number of income sources collected in the CPS ASEC, the values recorded for these sources also increased. In 1967, the format of the CPS questionnaire allowed for the recording of amounts up to $9,999 for each of the eight income sources. In 1970, the format of the questionnaire changed allowing the recording limits to increase to $99,999 for six of the eight income sources (wages and salaries; farm self-employment; nonfarm self-employment; interest, dividends, estate, trust, or rent; unemployment compensation, worker’s compensation, government employee pensions, or veterans payments; and private pensions, annuities, alimony, royalties, or regular contributions from people not living in the household). In 1979, the questionnaire allowed the recording of up to $99,999 for 23 income sources.4 In 1985, the limit for recording earnings from longest job increased to $299,999. The final recording limit increase occurred in 1993 when each of the four earned income sources allowed the recording of amounts to $9,999,999.

So, when Krugman wrote, "[t]he [Census] questionnaire is “top-coded”" he should have said censored.

When Krugman wrote, "if the individual interviewed has earnings higher than $999,999, those earnings are recorded simply as $999,999" he should have inserted the words "from any of the four earned income sources" between earnings and higher.

When Reynolds wrote "entirely wrong" he should have written slightly misinformed.

And when he wrote, "incomes well above $1 million are definitely included," he should have written as high as $4 million.

Have I missed something? I think we have it now.

[maria]

I went to see what some other "eminent" economists have to say. I found that Mankiw thinks that inequality has not changed much since 2000, but that is a far cry from Reynolds' claim that it has not increased since 198? (I forget his exact date). Mankiw mainly talks about the why of the increase, and does not dispute that there has been one. Barro accepts that it has increased. Feldstein thinks it has increased. Milton Friedman thinks it has increased, as his letter to Reynolds said. And here is another Harvard economist, Freeman:

Richard B. Freeman

"Over the past two decades, income inequality in the United States has massively increased. This jump owes to the unprecedentedly abysmal earnings experience of low-paid Americans, income stagnation covering about 80 percent of all families, and an increase in upper-end incomes. The rise in inequality-greater than in most other developed countries-has reversed the equalization in income and wealth we experienced between 1945 and 1970. The United States has now cemented its traditional position as the leader in inequality among advanced countries.1

These facts are not in dispute. From the Milliken Institute on the right to the Economic Policy Institute on the left, virtually all analysts agree that something has gone seriously awry with our income distribution. And absent some major national effort to change things, the new inequality is likely to continue (the recently reported 1995 decline in poverty and uptick in wages notwithstanding). The next recession will surely exacerbate it, and the forces contributing to wage losses at the bottom-foreign competition, immigration of low-skilled labor, technological changes, shifts from manufacturing to service industries, declining union density, subcontracting, and so on-are unlikely to reverse themselves anytime soon."

DeLong also thinks it has increased. I don't have the time to review any more, but would love to hear of an eminent economist who agrees with Reynolds.

[Sanpete]

The key term is "reporting limits."

Paul, the more relevant term is "processing limits." As explained above, according to Welniak (page 10 table 3, cp page 4), the reporting limit is $9,999,999, but the limit used for processing the data is $1,099,999. From what he says, it appears it's the latter limit that "modestly" affects estimates of income inequality, between 1 and 5%. In the year he uses as an example, 2000, there wasn't a single report that hit the reporting limit (26 went over the processing limit). It appears that if the raw data could be used, there would usually be no effect. $9,999,999 annual earned income must be fairly rare.

As far as I can tell, what I said in Times Select was completely right.

As far as I can tell, you were wrong about the amount of the reporting limit (which no source I could find calls "top-coding," a term the nerds who use this data appear to reserve for privacy coding after data collection), and you were wrong about whether the reporting limit affects the data processing. There's a different, lower limit that does that, which apparently could usually be avoided by using the raw data. You were right that there is a limit that affects the reported figures.

WCW, you can see from the above where there might be some further quibbles, but I think we're basically on the same page now.

Do you know of any eminent economist who agrees with Reynolds on the matter of whether inequality is rising or not? After all that is what the real issue is.

Maria, I agree that it's the far more important issue, but it isn't the issue in this thread. I'm trying to focus on what is the real issue in this thread, trivial as it may be.

[maria]

I was curious to know about the number of households with incomes above a million $; and also the number of those with incomes above 5 million. I could not find the data on the Census Bureau website. So I sent an email asking for it and was told that the Census Bureau does not have such information. In other words, their data does not go into incomes over one million (or whatever the upper limit is for their tables now). In short there are limits on reporting top income. And that surely results in a Gini calculation that is lower than the real one.

[maria]

Sanpete: Often discussions into minutiae are simply excuses for the display of intellectual ingenuity. How many angels can dance on the head of a pin sort of thing. I think this is what we have here and it is producing diminishing returns. The whole top coding question has importance only with regard to Reynolds' claim that income inequality has not risen since the 1980s. This is clearly WRONG, and that is the essential question. I'd prefer to concentrate on the important issue and leave the quibbles about angels and pins to others.

[wcw]

Actually, after rereading everything (deep breath) one more time, I am still not sure about those processing limits. Do they really make for an effective censoring of income data over $4m (4 income sources, each censored), or are sources of earnings summed before censoring?

[maria]

wcw:

Are you sure you are not wasting your time? Have you nothing more important to do? LOL.

[Paul Krugman]

Sanpete

I didn't know about the distinction between recording and processing; as you can see from the cite I gave, Census publications about trends in income inequality describe a "reporting limit" which doesn't make that distinction. So I was wrong to say that it was a limit on the questionnaire - though I think it's an excusable error, given what the Census wrote in its own discussion of inequality trends.

But the fact is that the income share data from the Census are missing the upper tail. Remember, that's what this dispute is about: whether the Census data fully reflect top incomes. Well, they don't.

Now, you might say that it's only a few people who are censored. But the fact that it's only a small number of observations doesn't mean that it's not significant: there's a lot of money up there.

The Piketty and Saez data say that $1 million a year roughly defines the bottom of the top 0.1%, which has almost 7% of total income, so lopping off that top would seem to be very significant.

And I was definitely right about another point: the number of CPS observations in that top category is very small, too small to provide reliable estimates. That doesn't mean that it's only a handful of real people in that category. There's yet another source of data on earnings, from the Social Security Administration, which is cited by Piketty and Saez; it's at
http://www.ssa.gov/OACT/COLA/awidevelop.html, in the upper left hand corner. If you look at the data for 2005, you'll see that there were something more than 70,000 people earning more than $1 million.

So to get back to the original point: I made a small error on an obscure point about how the data are collected. The questionnaire actually goes higher than $1 million, although the Census doesn't use the information.

But Reynolds is wrong about the economic points: despite his strident assertions to the contrary, Census income distribution data do not take very high incomes into account, and the CPS data really aren't a large enough sample to follow those incomes accurately. Which is why we need alternative approaches - either relying on income tax data, as CBO, IRS, and Piketty-Saez do, or the clever adjustments based on Pareto distributions that the people at EPI do. Both approaches give a higher top-income share than the Census data.

And yes, inequality is rising.

[Sanpete]

Paul, what you say makes sense to me. I think Alan would have a better point if the raw CPS data were available, though, so we could cover up to $10 million. Welniak's analysis implies that in most years no data would be lost that way. Some adjustment might still need to made for outliers, which are further out on the rich side than the poor, but that's into stuff I don't know much about.

WCW, Welniak's table 3 (page 10) shows a "questionnaire limit" of $9,999,999 for each of the four main income categories, with a processing limit of $1,099,999 for earnings, $99,999 for interest, 100,000 for dididends, and $99,999 for rent. It appears from this that the upper processing limit from those four is $1,399,997, but some incomes below that amount would also be limited (as when you make $1 million in earnings and $200,000 in interest). At least that's how it reads to me.

I'd prefer to concentrate on the important issue and leave the quibbles about angels and pins to others.

That's fine, Maria. We seem to have managed.

[calmo]

So simple.
How many angels?
Well 42 byjesus!
The official documents show 70,000 making at least $1M in 2005 (and likewise Greenspan's income was only 39 times less than the ave CEO in the housing sector), but do we need to exercise our scholars to arrive at this evidence of an increasing income disparity? [What are we going to do with our scholars if we don't people?] We do.
Otherwise the scholars might win or lose some debate about it...which could spread and infect the non-scholars like maria who care not for the number of angels on pin-heads.
That means you people...who are not on the bridge of that 460 ft yacht. Nope, that's Mr Ellison and probably many dancing angels:
http://www.powerandmotoryacht.com/megayachts/1104top100yachts/index1.aspx
You don't see maria on the bridge, do you?
But do you care? No, you are waiting for Alan to tell you that it is a mirage and that ever since 1988 the chances of you on the bridge look better.
The numbers speak for themselves.

[anne]

Well, turns out I was right in my reading om top-coding from the beginning. I can so read. But, as I pointed out the mistake Paul Krugman made was slight and easily understandable given the need to pay attention to every possible word on how the Census Bureau records and uses income data, and the slight mistake in no way changes the conclusion on wealth and income inequality.

Notice, I am expressly attaching wealth inequality to income inequality as does David Cay Johnston, though Alan Reynolds argues we should not, but we right and Reynolds is wrong.

Yes; wealth and income inequality are increasing as I have argued from the beginning. Portfolio and tax structure, and bargaining imbalances between corporate labor and management assure inequality will be found to have increased since 2004.

Thank you, Paul Krugman.

[anne]

http://www.calvorn.com/gallery/photo.php?photo=6991&exhibition=7&ee_lang=eng&u=21730,10

Black-bellied Whistling Duck
Camman's Pond-Merrick, NY.

Nice :)

[Lafayette]

anne: "Portfolio and tax structure, and bargaining imbalances between corporate labor and management assure inequality will be found to have increased since 2004."

Tax structure is the major tool for rectifying wealth inequality.

The Pickety-Saez method was employed (by Pickety) in France. The inequality in wealth distribution is SIGNIFICANTLY different from that estimated, by the same methodology, in the US.

Why? Tax structures are significantly different, particularly at the upper echelons - being far more imposing. This is also true of the Scandinavian countries.

A plutocracy just doesn't happen. It is crafted.

[maria]

Of course Krugman is right. If you don't give full weight to the top incomes for whatever reason (and you can quibble about the reasons endlessly) the Gini Index comes out lower than it really is. Fundamental observation and how much controversy it has produced, much of it misguided. So many trees and twigs to fixate upon that people haven't been able to see the forest that looms so clearly.

[anne]

Lafayette:

"Tax structure is the major tool for rectifying wealth inequality.

"The Pickety-Saez method was employed (by Pickety) in France. The inequality in wealth distribution is SIGNIFICANTLY different from that estimated, by the same methodology, in the US.

"Why? Tax structures are significantly different, particularly at the upper echelons - being far more imposing. This is also true of the Scandinavian countries.

"A plutocracy just doesn't happen. It is crafted."

Important argument.

[Lafayette]

HomeDepot Exec Resigns

From the NYT (today) : “Robert Nardelli, the chief executive of Home Depot, … has abruptly resigned, the company said today.

He will receive about $210 million in compensation from the company, including the current value of retirement and other benefits he was already entitled to, the statement said.

Since he was hired as chief executive six years ago, the board has awarded him a total of more than $240 million in salary, bonuses and stock grants.”

At the turn of the 20th century, America invented the “Robber Barons”. At the turn of the 21st century, it is inventing the “Robber Baby-Boomers”.

Greatly significant advance in the culture of mankind, that. A giant rip-off of shareholder value, approved by the BoD and perfectly legal.

Got a problem with it? File a class-action suit to recuperate wildly exaggerated severance payments ... and spend the next ten years suing the BoD.

[Alan Reynolds]

Anne wrote: "I am expressly attaching wealth inequality to income inequality as does David Cay Johnston, though Alan Reynolds argues we should not, but we right and Reynolds is wrong."

My point about Johnston's piece is that we cannot infer from tax returns how much corporate stock is held by the top 1%. Why not? Because a very large amount of corporate stock is held in nontaxable form. That always included pension funds, foundations and university endowments, but in recent years also adds trillions accumuating in IRA, 401(k) and 529 plans, among others.

Couples with incomes above $160,000 are not alowed to participate in some of the best tax-deferred savings plans, like the tax-free Roth plans, and their contributions to other such accounts are strictly limited.

The rest of us, however, have been rapidly shifting from taxable to nontaxable savings. The 2005 Proceedings from the National Tax Association contains some revealing IRS data about that. This is yet another of many reasons why income tax data is a very poor way of estimating the distribution of income, wealth, ownership of stock or tax-free bonds, etc.

Tax data is a little better for labor income, as Ian Dew-Becker suggested, but only after 1987 -- as in the April working paper from Schwabish at CBO. There was a lot of shifting from execs and middle managers being paid in capital gains to W-2 compensation around 1987, because the tax on salaries went way down and the tax on gains went way up.

[Alan Reynolds]

Mr. Krugman wrote that SOI tax data were better than Census data because the latter are from a "limited sample." They both use a limited sample. He wrote that Census distribution estimates count all income above $1 million as if it was $1 million. After changing the subject to something I wrote in 1992, he finally got around to saying "I was wrong to say that it was a limit on the questinoanaire." End of story.

I never said the Census data were flawless, nor did I mean to suggest that my conclusions about distribution since 1988 were based on Census data for pretax, pretransfer income. They are not. On wealth inequality, for example, I could just as well use Piketty and Saez, or Ed Wolff, or the updated 1997 revisiion of that 1992 Fed/IRS working paper Krugman recalled.

I only mentioned the Census figures for top 5% in the WSJ because they are much flatter and 10 percentage points smaller than those from Piketty and Saez, even though both exclude transfers and capital gains. The reason for that huge difference has nothing to do with top coding in the numerator. The reason is that Census captures much more income in the denominator than Piketty and Saez do, though not as much as the BEA does with personal income. In the latter case, 60% of the difference between personal income and tax-return income is transfer payments. But the other 40% is important too. It includes such things as the AGI gap (tax avoidance), nonfilers (including nearly half of the elderly), and the tax-favored dividends and interest income of personal trusts and pension funds.

Ratios have a numerator and a denominator and we have to look at both. I hesitate to make such a pedantic point except that some poster here could not even figure out that I must have graduated in economics, in 1965, in order to attend graduate school (at night) in that subject.

[Mark Thoma]

Alan Reynolds:
Thanks for your comments. Let me suggest some rebuttal. The key claim you make, that the Census data do take high incomes into account, is suspect, while Paul Krugman's claim that they don't is supported as far as I can tell.If you do a Google search on soi oversampling returns, you will find: the SOI, aside from being a larger sample than the CPS, oversamples high-income returns, so that it can give an accurate picture of income distribution at the top.

Here:
http://www.nber.org/~taxsim/marginal-tax-rates/plusstate.html.

So this appears to be the same misconception that was made back when you questioned the Federal Reserve wealth data in 1992.What am I missing? Maybe you can explain how it's different this time?

[evagrius]

The following is data that a shmoe like me can pull from the U.S. Census on its site- notice the cut off for household incomes.
Sigh....if it's this limited for an ordinary person to obtain, how is the information on the top 1% or higher available to someone like me? What machinations do I have to do to get it? How do I, as an ordinary citizen, obtain this information.
Does Mr. Reynolds have a "secret", esoteric method?

INCOME AND BENEFITS (IN 2005 INFLATION-ADJUSTED DOLLARS)

Total households
322,399
+/-3,073

Less than $10,000
33,435
+/-2,951

$10,000 to $14,999
19,306
+/-2,433

$15,000 to $24,999
29,867
+/-2,640

$25,000 to $34,999
23,579
+/-2,411

$35,000 to $49,999
35,635
+/-3,171

$50,000 to $74,999
53,420
+/-3,419

$75,000 to $99,999
37,626
+/-3,129

$100,000 to $149,999
46,496
+/-3,304

$150,000 to $199,999
19,639
+/-2,409

$200,000 or more
23,396
+/-2,471

Median household income (dollars)
57,496
+/-1,917

Mean household income (dollars)
83,417
+/-2,868

[Lafayette]

If ALL top-level income were reported to the IRS and obtained by the Census Bureau, then the Caymans would be a meaningless group of islands in the Caribbean.

Which is NOT the case.

[Sanpete]

If ALL top-level income were reported to the IRS and obtained by the Census Bureau, then the Caymans would be a meaningless group of islands in the Caribbean.

Good point. There's lot of underreporting of income on the low end too, in the cash economy. Complicated stuff, these comparisons.

[Lafayette]

"There's lot of underreporting of income on the low end too, in the cash economy."

Analyses of the Italian economy, based on money velocity, indicated that as much as a third of the economy was once turning over off the books (in cash) - in the seventies. That has come down considerably since the fiscal authorities tightened the screws.

I suspect that it is much lower in the US, except at the low-end where people are willing to work for cash unreported and untaxed. Who can blame them ...

[Trying to follow the Topic]

And I thought I had understood the maths:
Krugmann said it was limited to 999,999$, and we have seen it is sort of limited to between 1,000,000 and 1,300,000 depending on the structure of your earnings.

So, if we are at nitpicking, just don't forget that one.
Someone renting directly the Empire State Building would be counted for 99,999$ income, or Bill gates a year when he doesn´t sell any shares and just cah dividends would be counted for 99,999$ income ).
I suppose nobody knows if that limiting can be in any way significant, and if any of the 26 cases mentioned earlier who were over the processing limit felt in that category?

More generally, the limit was not on the questionnaire, but afterwards by processing.
Conclusion: AR can say "PK was wrong, End of story". ergo no need for a Progressive Tax?
Or did I misunderstood the logic of a Cato senior fellow here?

So,help me to understand the plain english: do the people at Cato think income inequality has increased in the last 15 years, yes or no?
Apparently no, but based on which datas? AR write "I never said ... my conclusions ...were based on Census data for pretax, pretransfer income. They are not"
OK. On which datas are they based again? The thread is too long for me now, I can´t find it again.

[Lafayette]

"Bill gates (in) a year when he doesn´t sell any shares and just (his) cash dividends would be counted for 99,999$ income "

That'll be the year ... when were all dead.

[Alan Reynolds]

I am talking about disposable income, after taxes are subtracted and transfers added. That is, after all, what we have to spend. And I have always been talking about what happened since 1988, not since 1979-81. What happened between 1979 and 1988 is ancient history, not a trend.

It's not that hard to find the figures at census.gov, though the data for 2005 are not yet available. My contention, easily verified, is that aside from the 1986 spike when folks sold stocks to beat the incrased capital gains tax and the 1993 change in Census surveys (when they used computers and allowed income of up to $10 million from one job to be counted), there has been no increase in the disposable income share of the top 5-20% or the Gini index.

That just happens to be a fact. Sorry. Read my next column at townhall.com, where I suggest that President Bush, his CEA head and the Chairman of the Fed are using numbers they obviously do not understand. If one prefers to support the President, that's nice. But difficult.

[Mark Thoma]

1979 is ancient history, 1988 is not? That's funny. Keep on cherry-picking.

You can "suggest" whatever you want. Doesn't make you right. Krugman has addressed this many times (why this measure of inequality is meaningless) but you keep your ears covered pretty well.

To use a phrase of yours, ever consider that you might be wrong? Obviously not...

[anne]

Suddenly, I am lost because the definitions appear to be changing. Concentration of wealth and income has been increasing these 25 years; especially so, especially clearly when the wealthiest are set off. Heck, any portfolio manager understand this. I am lost. What are we arguing?

[Movie Guy]

Alan Reynolds - "What happened between 1979 and 1988 is ancient history, not a trend."

Mark Thoma - "1979 is ancient history, 1988 is not? That's funny. Keep on cherry-picking."

Maark, did you mean to say 1989?