Monday, April 26, 2010

In which I (essentially) win my "quick bet against Brooks"

Three weeks ago, I put up a post based on my inference that David Brooks had misrepresented a statistic cited from Stephen J. Rose's not-yet-published Rebound: Why America Will Emerge Stronger from the Financial Crisis.  Brooks reported that according to Rose, "Over the last 10 years, 60 percent of Americans made more than $100,000 in at least one of those years, and 40 percent had incomes that high for at least three."  My inference, that Rose was referring to household rather than individual income and that "Americans" should be "American households," was based on past Rose writings and a near-identical error that George Will made in citing Rose's income data claims. 

Andrew Sullivan linked to my post. Brooks complained that I was wrong, that Rebound did refer to "adults" as opposed to "households." The Dish posted a direct quote from Rebound: ""Another way to look at incomes over many years is to see how often people experienced high and low incomes. Indeed, fully 60 percent of adults had at least one year in which their incomes were at least $100,000" (p. 119).  I posted a retraction and apology -- conceding that whatever the truth behind the numbers, I was wrong to jump to the conclusion that Brooks had misquoted Rose (though I did originally acknowledge the "highly unlikely" possibility that Rose's numbers or terminology may have changed from 2007 to 2010).

Today my copy of Rebound finally arrived (Update: my full review here). I am happy to report that while the book does indeed (of course) contain the sentence above, Rose's definition of terms earlier in the chapter makes it clear "their incomes" signifies the household income of each "adult" -- adjusted, moreover, for age and household size (see Update 2 and Stephen Rose comment below).

The chapter in question, "The Myth of the Declining Middle Class," begins with a bit of terminological slippage that's clarified later but turns up in Brooks' representation. My emphasis in all cited passages below:
These longitudinal data show that many people have long term incomes of over $100,000 a year--a level that permits a minimum level of discretionary expenditures in most parts of the country. This large base of families with reasonably high incomes provides the underpinning of our mass-consumer market as evidenced by crowded suburban malls (103).
That conflation is more or less explained later in the chapter, when Rose discusses the various factors that need to be folded into apples-to-apples comparisons of household incomes across different eras and income levels:
Analysts agree that one's standard of living is best reflected by total family income--a figure that includes all sources of income (e.g., earnings, business income, rental payment, interest and other financial income, and transfer payments such as Social Security, unemployment insurance, and welfare and disability payments) from all members of the family....

Few dispute that the same amount of money does not support the same level of consumption in households of different size, and researchers have developed methodologies to align income and size to get equivalent standards of living.  For example, in the approach that is used here, a single person with $50,000 of income has the "equivalent income" of a family of three with $86,600 or a family of four with 100,000(106)
I am not sure what family size Rose uses for his baseline in estimating the incomes of "Americans."  He does note that average family size dropped from 3.3 persons per household in 1967 to 2.6 in 1998, stressing that family income needs to be adjusted to reflect that change.  At a very rough estimate, it would seem that a solo individual with an income of $70-75k in the most recent ten-year period would in Rose's formulation have an income over the $100k threshold. Rose further adjusts his data set for age, limiting his longitudinal analysis to "prime-age adults who are never younger than twenty-six (they've finished their education) or older than fifty-nine (they've not cut back work in preparation for retirement)."

I should not have cast my inference that Brooks was misquoting Rose as a near-certainty without being able to verify it. Literally, there was no misquote -- or rather a minor one, converting Rose's "fully 60 percent of adults had at least one year in which their incomes were at least $100,000" to a more active verb formulation: "Over the last 10 years, 60 percent of Americans made more than $100,000." Brooks' re-cast also edits out a ghost of pronoun slippage in Rose's studiedly vague formulation: "adults" had years in which "their" incomes were over $100k. While "their" grammatically agrees with "adults," keeping both in the plural somehow highlights the elision by which household income (the term Rose uses in earlier writings citing similar statistics) becomes the income enjoyed by the individuals in the household.

If Brooks read the full chapter, -- and it's not clear whether he had an advance copy or simply interviewed Rose -- he should have noted the multiple lenses through which that vision of "60 percent of adults" enjoying the magic six-figure income level was filtered.

UPDATE: Brad DeLong translates Rose's "60% earned $100K at least once in ten years" claim:

My reading of Stephen Rose is that the best way to report his finding is: "fully 60% of adults aged 34-59 in 2005, over the previous ten years, had at least one year in which their household's total income excluding capital gains would have supported material living standards for its members that, when adjusted for family size and composition, would have equalled or exceeded the material living standards of a family of four with household income greater than $100,000 in 2009."
In the comments, I asked, "why a family of four?" DeLong's response: "That may be wrong--but it seemed to me the natural way of reading, since figures are usually given either in per capital terms or for a household of four. But I am trying to check up on it right now..."

UPDATE 2: Thanks, Stephen Rose, for weighing in directly in a comment, below. Rose confirms that the $100k in question here is household income but clarifies that it is not adjusted for family size. That means that the passage about methodology cited above, explaining "the approach that is used here," does not extend (fully) to info provided under a later subhead that appears later in the chapter. The passage that concerns us is in a section analyzing data from  The Panel Study on Income Dynamics (PSID) (under the subhead "Long-Run Incomes). PSID does provide data adjusted for family size; in discussion of this data, Rose cites some stats that are size-adjusted and some that aren't. Specifically, Rose stresses in paragraphs immediately preceding the sentence paraphrased by Brooks that median U.S. family income growth since 1967 is much higher when adjusted for size.  In retrospect, I can see that figures cited in the text in this section are size-adjusted when Rose spells that out, and not size-adjusted when he doesn't (also, I should have noted that PSID data cited in the nearest chart is size-adjusted for a family of 3, not a family of 2.6 as I speculated, based on Rose's citation of average family size in 1998).

Related posts
David Brooks gins up another faux consensus
David Brooks has a point...
The David Brooks consensus: bring back W.
David Brooks' free market fantasy
David Brooks lambastes Democrats for agreeing with him
David Brooks shadows forth a conservative Obama


  1. Also, how much of that "Over $100,000" year came from selling an asset, most likely a house? And thus was a "One Time" source of income - or one that went right back into a newer, larger home.

  2. Rose's original phrasing is so tortuous and ambiguous as to be almost useless. It's still unclear as to whether he's looking at household or individual income. Is this net income? Earned or unearned income? What's the distribution over different age groups? How many of these cases are the result of one-time events, such a sale of a home, or an inheritance? Rose's initial assertion is absurd, because it lacks specificity and context. My mother sold her home four years ago, because she could no longer live on her fixed income and need the cash. She had lived in her home for 35 years, so her net income on the sale was over $100,000. Is this a positive economic indicator?

  3. I still have no idea of what Rose means when he says that "many people have long term incomes of over $100,000 a year." What is a "long-term income"? I don't believe that 60% of American "prime age" adults have reported a gross income of $100,000 or more to the IRS. I don't know why he includes Social Security when he's talking about people under 60.

    My "best guess" is that what he means is that $100,000 a year for a family of four "permits a minimum level of discretionary expenditures in most parts of the country." For an individual living alone, $50,000 is an equivalent income. Therefore, an indvidual living alone with a $50,000 income really has an income of $100,000 a year. He's defining "an income that permits a minimum level of discretionary expenditures" as "$100,000 a year." So if you have an income that permits a minimum level of discretionary expenditures, then you are making $100,000 a year, even if you're only reporting $50,000 to the IRS. Did somebody say "no matter how you slice it, it's still baloney?"

  4. If I were single and made 50K a year, that will barely pay for my 1300 sq ft of mortgage and payments on a new car and utilities and insurance, etc. here in OKLAHOMA CITY, let alone most other places in this country. 100K a year sounds like a lot but in places like, say, Chicago or the Bay Area, you have to make that much just to live somewhat comfortably and not be busted all the time.

    The real worth of a dollar, and hence of a salary, can only be measured compared to the cost of living in a particular place.

  5. Let's see..... "over half of Americans have made over $100,000 in one year! Gee, we're all rich, aren't we?. Of course, obviously when I say $100,000 you understand I really mean $50,000 if you're single. And I'm not counting old people or young people. But I am counting one time asset sales and there were those stock market and housing bubbles where big gains were quickly followed by big losses and crashes and bankruptcies and stuff....." What a deceptive load of garbage. This "research" amounts to defining figures however you wish all for a juicy soundbite. There's just as much justification to adjusting family income up from $100,000 for larger families and making a single person the baseline as the other way round but then you wouldn't get that impressive sounding 60%. Sad.

  6. If you add in asset sales (a house) and probate (mom died), then, tah-dah, I made over $100,000 a year twice in the 90's, even though my salary never exceeded $33,000 per year.

    Brooks needs to issue a clarification - he strongly implied this was earned, ie: salary, perhaps + investment, income.

  7. Is there no way to simply have Rose clarify his dataset and calculations? If a stat is published in a book does that automatically qualify it to be quoted as fact by a news organization or opinion writer?

  8. Thanks, commenters. While I'm no economist, I should say that while Rose's numbers sit oddly when clipped out, and while I think his explanations for them in this general-audience book leave some blank spots, and while I'm sure that many economists like Jacob Hacker would find plenty to dispute, I think Rose adds valuable perspective, and that his general point that Democrats and economists most influential with them paint too bleak a picture of the long-term *average* economic health of the broad middle class has some validity. The book is worth a read. I'd like to hear/read Rose go head-to-head with Hacker or, say, Elizabeth Warren -- I will look for their print encounters.

  9. Hi. It's me the author of the book and data.

    I have debated many people on the state of the middle class and discuss Hacker's and Warren's works in my book. Any time someone wants to arrange a debate with them or anyone else, I am available.

    The number that has caused great confusion in a number of places is not intuitively clear. The Panel Study on Income Dynamics (PSID) follows the same people over many years. Because people change their living situations over 10 years, it is necessary to follow the same person and use the total household/family income as the reference point for that year. Household income is the standard measure of standard of living and the only adjustment I make to income is to adjust each year into 2007 dollars.

    So these data are not adjusted for family size even though in other points of the chapter I do adjust for family size. This is common procedure as reflected in the definition of poverty which varies with household size.

    Hence, it is a simple process to look at the household income for each person in each of the 10 years and determine the number of times they were in households with incomes over $100,000. I can provide the output tables or the programming code to anyone who wants to validate this finding.

    David Brooks quoted this factoid without much context. For those who are interested, they can read the whole chapter.

    Steve Rose

  10. It's worth noting that at a gross family income over 100,000 in NYC, rent of a 1-bed room still constitutes 35% of net income! In other words, geography matters too, both because of expenses and state and local taxes.

  11. Glad to be of help. I try to be very careful with data which often requires great precision with language to explain what is being measured.

    A couple of words in defense of Brooks:
    1. He had an advance copy of the book and picked this provocative factoid even though I did not highlight this point. Because I only had one sentence on this finding, it is easy for the reader to be a bit confused.
    2. It is often very difficult to describe data, especially when it concerns following individuals but reporting their household incomes as they move from household to household. David used a few words to describe a complex numbers and was bound to be less than precise.
    3. He uses data to provoke rather than tell the whole story. He certainly was successful because I've been involved in another long interchange on this point on another listserv.

    They are other provocative data points in my book Rebound and hope that people will read the it and come to their own conclusions.