Arnold Kling  

Squeezed Up

Hayek and the Internet... Higher Education Lobby...

The Washington Post (possibly more reliable link here) does not use phony documents, but the thrust of the story, including the headline "More U.S. families struggle to stay on track" is at variance with the facts that they present. The key chart, which is included on the jump page in an obscure way and never discussed in the article, reports on the change in the distribution of income, using 2003 dollars (I converted the data to tabular format).

Income DistributionPercent of Households
$75K and up8.226.1
$50K - $75K16.718.0
$35K - $50K22.315.0
$15K - $35K31.125.0
under $15K21.715.9

The article emphasizes that the middle has shrunk, from 22.3 percent of households to 15.0 percent. What it does not point out is that the two categories below the middle also have shrunk, from 52.8 percent of households to 40.9 percent. Adjusting for inflation, the percentage of households with incomes over $50,000 has climbed from 24.9 percent in 1967 to 44.1 percent in 2003.

The article's claim that it has become harder to stay in the income range of $35,000 to $50,000 is correct, if what you mean by "harder to stay" is that it has become difficult to avoid being squeezed up into a higher category.

UPDATE: More data and analysis, from Robert Rector.

UPDATE 2: Bruce Bartlett found similar problems with a New York Times story a few weeks ago.

For Discussion. In what ways do these data overstate or understate the degree of upward mobility in the economy?

Comments and Sharing

CATEGORIES: Income Distribution

TRACKBACKS (9 to date)
TrackBack URL:
The author at Newmark's Door in a related article titled writes:
    This is fine work: Arnold Kling trashes the conclusion of a front-page article in the Washington Post using data that appear in the article itself. (The chart he refers to is here.) [Tracked on September 24, 2004 6:31 AM]
The author at Right Mind in a related article titled The Poor Getting Poorer writes:
    TITLE: The Poor Getting Poorer URL: IP: BLOG NAME: Right Mind DATE: 12/06/2004 09:38:24 PM [Tracked on December 6, 2004 9:38 PM]
The author at Right Mind in a related article titled The Poor Getting Poorer writes:
    TITLE: The Poor Getting Poorer URL: IP: BLOG NAME: Right Mind DATE: 12/06/2004 09:38:33 PM [Tracked on December 6, 2004 9:38 PM]
COMMENTS (13 to date)
spencer writes:

The major adjustment the data needs before you can make the judgements about living standards is adjustments for the increased labor force participation. In 1967 the standard of living for a middle class family was supported by a much smaller share of the population working. It is the story that you use to really have a typical family of 5 supported by one person working and the wife staying at home to raise the kids. But that has become quite rare now because it now takes two incomes to realize the middle class income. This would make a major difference in your analysis.

A. West writes:

Besides number of workers per household, one would also want to control for age of workers in the household.

One should also treat the problem as a panel data set, preferably looking at repeated observations of the same households, and modeling the dynamics of the units over time, including autocorrelation effects. Then you could really observe the movements of household income volatility and the factors related to it.

Lawrance George Lux writes:

Such Income tables are important to Economics, but mean little without comparison scales of Price increases. I personally bought a new car in 1969 for $2000. It was a Compact, admittedly, but still brand new. I bought eight pairs of Levis in 1967 for $6 apiece. Equivalent replacement for the Above would be $12,500 and $22 apiece. The Coup de grasse must be a friend who bought a House for $18,500 in 1965, and sold it for $92,300 last year. Tell me that American Labor has won the Numbers game. lgl

Tom writes:

Spencer seems to be making two points: (1) Being in the middle takes more income than it used to, which requires more mothers to work outside the home. (2) Families with, say, two persons working outside the home lose "something" that's not captured in price indices.

On the first point, it's true that being in the middle takes more income than it used to. But that's because real incomes have risen since 1967, mainly because of greater productivity arising from capital investments of various kinds (educational, technological, informational). In fact, per capita GDP (in 2000 dollars) rose from $17,535 in 1967 to $35,790 in 2003 (source: That kind of increase didn't happen just because a higher percentage of married women were in the workforce in 2003 than in 1967. (In fact, real-income figures, such as those I've cited, include the imputed value of household work done by stay-at-home spouses.) Real incomes have risen so much that many spouses who work outside the home could quit working and their families could live as well as they did in years past. But, by definition, families with two spouses working outside the home prefer the income and the things that it buys (including child care and better educations for their children).

That leads to the second point about the "cost" of working outside the home. As I've just suggested, working outside the home is a personal choice. Those who make that choice must feel that their families are better off as a result. Now, it may be that that choice has costs the form of, say, more crime than would otherwise be the case because a growing proportion of children lack parental supervision, etc. To sort that out, someone would have to estimate the cost of crime and other externalities that can be attributed to the increase in the average number of working parents per household. I doubt that the number is large, relative to the economic gains attributable to the higher percentage of working mothers. Has anyone analyzed that issue?

Patri Friedman writes:

lgl - that's why these numbers are all using 2003 dollars. That means inflation is taken into account. The inflation factor from 1967 to 2003 is 5.2, which explains your examples. The $35K-$50K range for 1967 was actually more like $7K - $10K in 1967 dollars.

A. West writes:


Another thing missing: hours worked. Last I remember was those going significantly up for every group; at the least break out the hourly wage.

John Thacker writes:

Jason McCullough-- not really, and definitely not over a lifetime. (See here, for example.) Of course, women's hours worked outside the home have gone up significantly, so if you don't break that out the numbers can be misleading, and imply that people are working more. Counterbalancing that, the average time being retired has greatly increased. People have to be able to afford being retired, of course, so that's another sign of wealth. (Even if the retirees' income comes from Social Security, *somebody* is having to work to provide both the retirees' income and their own.)

Certainly it's true that, economically, there's value provided with a homemaker that's not counted in these statistics. (It shows up particularly when wealthy dual-income couples are willing to hire housekeepers.) But people are choosing to work. And quite obviously it takes more real income to be in the middle if everyone's real incomes are growing.

Reporters somehow never let the numbers get in the way of a good, compelling story. What's with their over-reliance on anecdotes and stories vis-a-vis numbers and data.

Two cases in point
1) The Internet bubble: anecdotes about the wondrous ecommerce experience led to the mother of all bubbles (reality: even today, after 5+ years of 30% growth, online sales are less than 2% (sic) of all retail sales)

2) The Outsourcing alarms: India's total exports of all IT/ white-collar services are less than $10 billion today. And that's going to hollow out a $11 trillion GDP economy, where services are more than $7 trillion?! I don't think so.

"Hours worked over the last century have fallen" doesn't refute "hours worked have increased in the last 30 years."

I can't find the blasted table at the moment, but I do remember seeing data that showed:

1) Increases in hours per worker for both men and women at every point on the income scale.
2) But increases in *hourly wages* only occured for men in something like the top half of the income distribution. Stagnant or declining below that.
3) Women's hourly wages did increase at every income point, however.

The increased time in retirement (and implicit extra hours working to pay for it) is an interesting theory. This paper shows significant jumps in expected retirement length since the 1970s:

That still wouldn't explain stagnant or declining hourly wages for bottom half men, of course.

I have been looking at that chart and wondering what on earth they were using it to support "the middle class is disappearing" for, though. Figured it out, you actually need to look at this graphic:

Check out the bottom left corner graph; "percentage of all individuals who live in families that earn above, at, or below the median". *Individuals* in families who earn less than half the median have increased from 18% of the population to 22.5%.

I suspect the deal with the main graph is the effect of all those families switching from one earner to two. a) Couples tend to have similar income, b) there's more dual-earners the farther up the income scale you go, so c) it appears that the poor are disappearing enmasse, when d) they're really not - it's a demographic artifact.

Dave V writes:

More members in the family working to earn a higher family income. How many families in the 60's had two or three cars? How many had 5 bedrooms and 3 bathrooms in their home? TV's are now small if they are 20 inches, and many have tvs in their kids room, kitchen, ect... I don't have any numbers to back things up, but looking at my parents vs my own lifestyle there is a striking difference. To me the argument that it now takes two income earners to get by doesn't hold a lot of weight. People want more and better "things" so they work more. I don't have any kids, and my wife works. But if we wanted one nice car instead of two, had a smaller home, and didn't take as many "big" vacations we could easily get by on one income.

If you do assume you're accurate about your observations, do note that "people have rationally decided to work more so they can get more things" is not really equivalent to "people have gotten richer" or "incomes have gone up."

Comments for this entry have been closed
Return to top