Krugman: On Tracking Inequality

Paul Krugman tries to clear up some confusion over measuring inequality, and institutes a new practice - posting links to the data sources in his columns. He hopes other commentators will do the same thing and save us all a lot of time chasing down sources behind arguments:

On Tracking Inequality, by Paul Krugman, Money Talks: Now that rising income inequality has become a big political issue, people are throwing around a lot of numbers. Some of these numbers are reliable, other aren’t. But how are readers to tell the difference?

Well, one thing that might help is knowing where the standard sources are.

The first point of call is data from the Census. Census numbers are based on the Current Population Survey, a questionnaire filled out by a sample of Americans, then extrapolated to the nation as a whole. For historical comparisons, go to Historical Income Tables.

Data there are gathered under several categories: households (people living together), families (they have to be related), and individuals. (Formal definitions) As of now, only the household data have been updated to 2005, which is why I recently turned to Table H-13 – Educational Attainment of Householder – to show that most Americans with a college education have lost ground in recent years.

The Census data are the key source for assessing how most Americans are doing. However, they do a poor job of tracking incomes at the very top, for two reasons. First, because Census data are based on a limited sample, not the whole population, they’re unreliable in tracking the income of small groups – and the really rich are a small group, who just happen to bulk large in the economy. Second, the 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. In particular, what you won’t learn from Census data is the extent to which rising inequality is a story, not about the top 20 or even the top five percent of the population, but about the top one and the top 0.1 percent.

Fortunately, there’s another source of information: income tax returns, which aren’t top-coded. Tax return data are especially useful if you want to look at long-term trends going back before 1947, which is when the Current Population Survey data begin; high-income Americans have been paying income taxes since 1913. The I.R.S. does its own analyses of these data, and the Congressional Budget Office produces reports based on a merge of Census and I.R.S. data, but the most convenient and comprehensive analyses come from Thomas Piketty, at the Ecole Normal Superior in Paris, and Emmanuel Saez at UC Berkeley. Their latest data set is at Prof. Saez’s Berkeley home page (Excel file.)

There are other sources, too – which I’ll explain when I use them. You see, I’ve decided to institute a new policy. On inequality, and in fact on many matters economic, it’s all too common to have numbers – some from unknown sources – flying in all directions. The issues are hard enough without clarity about where numbers come from. So from now on I’m going to post sources for the numbers in each column on TimesSelect, with links where possible (it usually is.) Basically, this is the same thing I do when filing my columns; I always provide sources and links to my copy editors. But now I guess my explanations will have to be grammatical! Anyway, I hope that other economic commentators will follow the same practice, which is easy in this Internet age, and will save all of us a lot of confusion.

This is an example I received from Paul Krugman showing what a sources and methods email to his copy-editor looks like. (Brad DeLong also posted this):

From: Paul Krugman xxxx
Date: Sep 7, 2006 5:56 PM
Subject: Column
To: xxxx

Hi xxxx   

On cell. Fact-check:   

Economy is poor: http://pollingreport.com/consumer2.htm

Most people worse off: small gain since 1970s:

Median household income in 2005 dollars: http://www.census.gov/hhes/www/income/histinc/h06ar.html

Wages and inflation: http://www.prospect.org/deanbaker/2006/09/david_brooks_swings_and_misses.html

Household income of college graduates: http://www.census.gov/hhes/www/income/histinc/h13.html

95th, top 1, top .01: http://elsa.berkeley.edu/~saez/TabFig2004prel.xls, Table A4   

Tax rates for the wealthy:  http://elsa.berkeley.edu/~saez/piketty-saezNBER06taxprog.pdf, p. 51

I hope other commentators will respond to the challenge and post their sources and methods as well.  There's no good reason not to.

Comments

[Emmanuel]

First, because Census data are based on a limited sample, not the whole population, they’re unreliable in tracking the income of small groups – and the really rich are a small group, who just happen to bulk large in the economy.

Technical point: Krugman is right in saying that the data used in this instance is based on a sample (the Current Population Survey) and not the whole population. The Census Bureau only conducts a census--not a survey--of the entire population every ten years. I just wanted to note that to avoid possible confusion.

[anne]

http://www.cepr.net/publications/europe_2006_09_19.pdf

September, 2006

Old Europe Goes To Work: Rising Employment Rates in the European Union
BY JOHN SCHMITT AND DEAN BAKER

Introduction

During the 1990s, many economists, international organizations, and politicians promoted U.S.-style labor-market flexibility as a solution to Europe's employment problems. Through much of that decade, the United States did have a significantly higher share of its working-age population in jobs than Europe as a whole. For example, by 2000, when the 1990s business cycle peaked in the United States, for 15-to-64 year olds, the employment gap between the United States and Europe was 10.5 percentage points. During the 2000s, many European policymakers have continued to urge U.S.-style reforms, even though the employment gap between the United States and Europe has narrowed considerably. By 2005, for example, the overall employment gap had fallen to 6.1 percentage points, reflecting both declines in employment rates in the United States and increases in employment rates in Europe....

[Bruce Wilder]

Sources and methods for the National Review Online or the Wall Street Journal editorial page are likely to read like something extracted from a Harry Potter novel: "Defense of the Dark Arts" and similar titles.

[anne]

Now, what I should do is burst to song, "birds do it, bees do it, even da dee dees do it," in relation to how Europe can possibly be labor friendly and yet have any employment at all let alone employment as high as ours. So, if birds in Europe do it, why can't bird here do it? Please forgive my being so suggestive, and I can't even remember the words. Oh, the children, the children....

[anne]

"Birds do it, bees do it, even educated fleas do it. Let's do it, let's fall in love."

[This post has not yet been rated, but is suggestive as heck, at least for me, so protect the children.]

[spencer]

One thing I just learned this year is that when Census does its real income and income distribution data it uses the CPI-RS rather then the CPI-w.

The CPI-U-RS is a price index of inflation that incorporates most of the improvements in methodology made to the current CPI-U since 1978 into a single, uniform series.

It shows that inflation has been about a half-a- percentage point lower annually than the CPI-w.

So all the complains that the real income data is
overstated because the cpi overstates inflation is
not correct

[calmo]

Time for a family-rated picture of some birds people.
Do I have a method? Do I have sources?
Lemmesee, Bruce has "Defence of the Dark Arts" so I better go with "Zarathustra".
Ok, I can toughen up and reference the BLS and such. But it won't be as much fun as Owl pictures from-you-know-who.
You figure we should drop the mud slinging and adopt the rapiers? the pistols? the formal regalia? (I look good in tights people.) You would be amazed.
Ok, maybe bowled over.
It could be a cramp on my non-bibliographical style, your's?

[anne]

Spencer, please continue your thoughts on the census data time income and income distribution series.

[anne]

Calmo, you are a humored gem. There have however been a host of fine comments for several days running.

[Richard]

spencer: One of life's little ironies: the crowd that believes CPI overstates inflation in income discussions is sometimes the same crowd that believes it understates inflation when it comes to investment discussions . . . .

[Michal Lehuta]

If higher and higher ratio of population has a college education, the difference between college meadian and total median goes down by definition.

[Richard]

Michael: true enough. But one (crude) measure of skewness is to divide the mean by the median. As a within-group measure, it's overall increase should tell you whether or not disparity is increasing, and for which groups the disparity is increasing more rapidly.

Take the "Household income of college graduates" listed above and do a simple exercise:

Mean/Median
HS Bachelors Professional Graduate or more Ph.D Degree
1991 1.17 1.18 1.14 1.182005 1.29 1.30 1.30 1.54
Note that . . .

(1) Educational level has not had much effect on skewness (at least as measured by mean income / median income).

(2) The one group for which that is not true are professionals - presumably physicians and lawyers.

(3) All groups have experienced increased income skewness over a 14 year period.

[Noumenon]

It would be great if everyone did this. Lately my dad says "3/4 of their customers are Republicans" whenever anyone mentions Wal-Mart. I know I have no chance of finding out where this number came from because George Will or somebody made it up.

[anne]

Notice, then, that Paul Krugman's sources which were always readily available at

http://www.pkarchive.org/,

will now be routinely and immediately available at

http://krugman.page.nytimes.com/.

The sources, however, were always available on the archive through all the baseless criticism, so I do not imagine there will be any less criticism.

[calmo]

Those writers who care about getting the facts straight practice 'due diligence', cite their sources (and sometimes opposing views) and methods (sometimes even the limitations of those methods). These scholars are nauseatingly thorough about this path to The Truth which they usually belittle when they get there by claiming it might not be the real thing, not for long, not in all circumstances, not for some people, not on rainy days...
Most of us just want our views confirmed as The Truth against all our detractors' Rubbish.
I can take a word or 2 from Anne, yes?

[Arne]

When I look at a chart of the data, I wonder why people are not talking about what policy change happened in 1987 instead of 2000.