Arnold Kling  

Shifty Statistics

PRINT
A Middle-Class World... The Culture of Growth...

Peter Orszag writes,


According to Yale's Jacob Hacker, the average family had a 7 percent chance in the early 1970s of seeing its income drop by half or more. By 2002, that probability rose to nearly 17 percent.

This statistic raises more questions than it answers. For example, consider the three hypothetical families below, each of which has $40,000 in income in 2000 and takes a 50 percent hit to income in 2002.
Family20002001200220032004
A$40 K$40 K$20 K$20 K$20 K
B$40 K$40 K$20 K$50 K$60 K
C$40 K$80 K$40 K$40 K$40 K

I can see feeling sorry for family A, with their permanent loss of income. However, family B recovers nicely, and family C "suffers" from a spike in income in 2001.

Jacob Hacker leads you to think in terms of family A, but I would like to see what proportion of the families who suffer 50 percent income drops in a year represent family A. What if it turns out to be small?

Hacker's book, The Great Risk Shift, also says this (p. 96):


The problem is often called the "middle-class squeeze" (a term [Elizabeth] Warren coined), and it's certainly real. According to a recent analysis of consumer spending and family income, a typical two-earner family in which both partners work full time had to work 28.7 weeks in 1979 to pay for housing, medical care, college tuition, transportation, and taxes. By 2005 a typical two-earner family had to work 32 weeks just to pay these basic expenses (despite the fact that they were working about 2 weeks fewer a year to pay their taxes). [emphasis mine--AK]

I would say that if you have to include college tuition and omit food as a "basic expense" in order to show middle-class squeeze, then middle-class squeeze is probably a myth.

I expect academics to write with a respect for readers and, above all, for the truth. When they write for the general public, they should take particular care to try to avoid the use of misleading statistics. Most academics, regardless of politics, try to live up to those standards. I am happy to engage those with whom I disagree.

When it comes to dealing with dishonesty, however, I believe in trying to police your own side. That means that it's up to someone other than me to blow the whistle on Hacker.


Comments and Sharing


CATEGORIES: Income Distribution



COMMENTS (9 to date)
Buzzcut writes:

From the negative side, part of what he's seeing is the "two income trap". Living a lifestyle that can only be afforded on two incomes makes your lifestyle MORE risky than had you only had one income. Two incomes makes it twice as likely that someone is going to get laid off. If the income that is lost is needed for "basic expenses", then you're twice as screwed.

But does he acknowledge the two income trap? I'm betting no.

Randy writes:

I would guess that the category A family is the most common, but that most category A families are really category C families when seen from a broader historical perspective. That is, the relatively wealthy middle class of the 40s, 50s, 60s, and 70s, was the anomoly. The multiple income to make ends meet requirement of today is more the norm. Can anyone point to any time in history, except the years above, when it wasn't the norm for women (and often children as well) to work?

Nathan Smith writes:

One problem with including "college tuition" as a "basic expense" is that it reads into the statistics a "revolution of rising expectations" that has taken place. But there's another problem: Don't most people borrow to pay for college? And isn't that perfectly appropriate, considering that college is supposed to be an investment in human capital? It's as if one included the down-payment on a house as a "basic expense." To assume that people can, or should be able to, pay for college tuition out of pocket, without borrowing or saving beforehand, is misguided.

dearieme writes:

What about a couple who let out a house, she's a medic, he's a professor and they both cycle to work? They should be doing well. Are there many of those in the US?

Nathan Whitehead writes:

Isn't this blog post "blowing the whistle" on Hacker?

ptm writes:

One problem is that there's some evidence suggesting that family B could be worse off than a family that made 40K the whole time. Obviously they had 10K more income, but increasing the likelihood of negative income shocks seems to have negative consequences itself. This may be through stress or something like that, although that's just guessing.

I think Norm Waitzman at Utah has done a bit of work on this.

Lord writes:

I believe you would find A to represent about a third of cases, B another third, and not C (a decline by half rather than an increase) another third, while C would be highly uncommon. Increased variance diminishes total return by half the variance even when the mean does not change.

Victor writes:

A natural result of having young women in the workplace will be an increase in the year-to-year volatility of household income, at least during the years when the women are of child-bearing age. Who is worse off: the woman who only takes 8 weeks of paid vacation to have a kid, or the woman who puts her career off for a year? Why can't we let the woman decide? Why do we have to declare that women who forego a year of income are suffering "new risks"?

Presumably Hacker dealt with that and its late-life cousin -- early retirement -- in his study. The real nub of all this is that it may be erroneous to assume that we (the American people) are voluntarily foregoing income in 2002 at the same rate we were foregoing income in the 1970s.

patrick Garrett writes:

sooo i agree and dissagree with that little graph. i agree with the fact that the amount you have to work to survive has increased but i dont understand how you an put college tuition in as an expense. most people dont go to college sooo i think adding in food expenses would make your claim a little more valid.

Comments for this entry have been closed
Return to top