« Father Time | Main | Claims about Africa »
Where has all the income gone?
-
The U.S. Census Bureau reports that median household income stagnated from 1976 to 2006, growing by only 18 percent. In contrast, data from the Bureau of Economic Analysis indicate that income per person was up 80 percent.
Three data issues adversely impact reported median household income gains: the choice of price index, a change in the mix of household types and the measure of income used.
After adjusting the Census data for these three issues, inflation-adjusted median household income for most household types is seen to have increased by 44 percent to 62 percent from 1976 to 2006.
That's from Terry Fitzgerald at the Minneapolis Fed. I am not sure if he is asserting that his alternatives measures are just that -- alternative measures -- or if they are the true and correct measures in the sense of being better than the alternatives. In any case this is the latest look at a long-contentious issue. I thank Don Boudreaux for the pointer.
Posted by Tyler Cowen on November 16, 2008 at 06:24 AM in Data Source | Permalink
Comments
Can't median household income have increased 18% while per capita income increased 80%? The answer to the question "Where has all the income gone?" would have to be "disproportionately to the already well-off." I don't know if that's the case or not, but those two data points are not necessarily at odds.
Posted by: bigmike at Nov 16, 2008 9:37:44 AM
The problem of defining incomes and other variables, such as business receipts, has been with us a very long time. Relatively recently, there was a paper by Day and Henry (IIRC) comparing the definitions from IRS's Statistics of Income program (where they work) with those of BLS and Census.
On top of that, the highest incomes tend to not reply to surveys, even from Census. Some of BEA's data, however, arise from the tax records through the Statistics of Income data (hence the reference above).
Posted by: Paul McMahon at Nov 16, 2008 10:40:14 AM
It appears that someone out there does not assume ceteris paribus for things that we know have not held constant over the past 30 years. A household of 2.69 people that makes $X (adjusted for inflation) is better off than a household of 3.14 people that makes $X (adjusted for inflation).
The general public loves y = mx + b models, because it is something even a caveman can grasp. It also plays well for people who like to use the narrative fallacy to dupe people into conforming to their beliefs.
I find it ironic that the people that understand that assuming God is an irrational assumption, so strongly cling to the assumption of ceteris paribus, which in many cases, including this one is quite irrational.
Posted by: Jay at Nov 16, 2008 10:52:15 AM
I'm always looking for good info on inflation-adjusted wage growth, so thanks for that.
I thought I would ask in this thread if anyone knows of a good breakdown of costs of goods historically.
For instance, many friends of mine would counter that data with "yeah, well sure, but everything costs more now -- education, health care, etc."
Beyond the specific problems with those examples (e.g., what we call health care in 2008 is a fundamentally different good than health care in '76), I would like to be able to show how cheap groceries have become, for instance.
Any helpers?
Posted by: Curious at Nov 16, 2008 11:20:01 AM
As some folks here have said, the first things that came to mind were "household size has changed and income distribution has become more skewed". The Minnesota Fed result leaves some room for these two effects, but also finds different uses of the term "income" between the statistics.
I kind of object to the separation of the fact that they use a different measure of inflation from the fact that they use a different definition of income as a separate contributor to the difference. Each of them approximately deflates by the change in an index of prices of items that are spent from its definition of income; it would be at best unfortunate if they used the same index to deflate fairly different things. Part way through, as they apply the changes one at a time, they're doing that (deflating one basket of goods with the prices of a different basket of goods), which results in a comparatively meaningless number.
Posted by: dWj at Nov 16, 2008 11:35:38 AM
CBO data seems to be the most consistent time series on income since 1976, as near as I can tell. Perhaps I simply haven't heard the critiques of their data set, but it is the one I generally use. They even adjust for household size.
Posted by: Coyote at Nov 16, 2008 11:41:22 AM
.."because it is something even a caveman can grasp"
eh, picking on 'cave men' again? Isn't that 'ad hominem spelunkum' or something?
I can't get why it's not so obvious to academics and elite policy makers that the lower middle class has been in economic despair all through the 'boom' years of BushCo. Parallels to the growing economic inequality of the Roaring Twenties are striking. Bigmike is on the money.
Also, those minnesota fed reports seem to be outliers in contrast with the policy wonks that keep saying there is a credit crisis. Tyler or someone here had a post about how credit (in general) is still being delivered at a pace as before according to the minnesota fed so what gives? According to my locality, small banks are just doing dandy these days....so why are we bailing out Wall Street Banks?
on another thread why are commentators on the media begrudging Detroit a miserly 25 billion while no one's complaining (much) about AIG's continuing greed for the public troth? (ah, thread jacking again... apologies ) '
the link to the minn fed didn't open for me btw...
Posted by: datadave at Nov 16, 2008 11:47:11 AM
"Bigmike is on the money. "
For every complex problem, there is a solution that is simple, neat, and wrong. Bigmike provides us this solution.
"Also, those minnesota fed reports..."
Guilt by association?
Posted by: Jay at Nov 16, 2008 1:53:54 PM
Some nice tidbits from the article:
"Finally, there is the odd statistical fact that almost 60 percent of people live in households with above-median income"
"Each household type has considerably higher median income growth than the overall household median growth of 26 percent....married-couple households—the largest type—had a median income gain of 42 percent, while female householders with no spouse present—the second largest type—had a striking 56 percent gain in household incomes.....How can all subgroups grow faster than the entire group? But there is no contradiction. The explanation lies in the changing household mix"
Posted by: assman at Nov 16, 2008 2:07:15 PM
Household size has been declining. No great surprise, this means that the increase in income per household will be less than the increase in income per capita. There are also all kinds of issues with changing household mix etc.. Quoting Thomas Sowell, "When I write and see somebody quoting household income, he's trying to make things look bad." (http://media.hoover.org/documents/uk_sowell_transcript.pdf)
Posted by: mgunn at Nov 16, 2008 2:58:24 PM
My favorite unit of measurement is the median family of four: two parents, two children. Birthrates have been pretty similar for the last 35 years, so the family of four is pretty standard.
Posted by: Steve Sailer at Nov 16, 2008 4:09:02 PM
"Birthrates have been pretty similar for the last 35 years"
Even better is the implicit assumption that birthrates were "normal" from 1946 - 1960. You hear little talk about how the baby boomer generation reached their peak real earning years during the 1990s.
Posted by: Jay at Nov 16, 2008 4:37:08 PM
Well, according to the CBO, which does adjust for household size, the quintile change in income is as follows (since 1979):
Q1 10%
Q2 20%
Q3 25%
Q4 34%
Q5 85%
I'd say that paints a fairly middle of the road picture of income gains in the last few decades. You could probably read that data to support either side of the income disparity issue.
Posted by: Coyote at Nov 16, 2008 4:56:04 PM
Who am I kiddin? There ain't a bunch of ways to read that data. The income is mostly going up to the top.
Posted by: Coyote at Nov 16, 2008 5:12:32 PM
Coyote:
What is the following probability???
P(X is in top quintile | X was in top quintile last year)
Posted by: Jay at Nov 16, 2008 5:19:16 PM
What is the following probability???
P(X is in top quintile | X was in top quintile last year)
p= about 83%, from what I've read. (should I use three periods, like you used three question marks? I am not aware of all internet traditions.)
Posted by: Coyote at Nov 16, 2008 5:36:09 PM
Coyote
Where is your data from? This is the closest I found at the CBO (http://www.cbo.gov/ftpdocs/76xx/doc7693/12-04-LaborForce.pdf) but it is not even close to your figures.
Posted by: dale at Nov 16, 2008 6:29:29 PM
Ya got me Dale. Shoulda' check my sources, cause I remembered wrong.
It's about 58%, from this study.
Posted by: Coyote at Nov 16, 2008 7:03:00 PM
Actually, it was the other data I was wondering about - the quintile change in income.
Posted by: dale at Nov 16, 2008 7:20:11 PM
Oh, that.
Historical Effective Federal Tax Rates: 1979 to 2005.
Table 1C.
Posted by: Coyote at Nov 16, 2008 7:38:18 PM
"P(X is in top quintile | X was in top quintile last year)"
What is this? My family is has been in the top quintile for over a decade (currently about $100k). There must be a pretty strong carryover from year to year?
Posted by: liberalarts at Nov 16, 2008 7:54:03 PM
"Finally, there is the odd statistical fact that almost 60 percent of people live in households with above-median income"
Nothing odd here. The bigger the household the more earners, on average, hence bigger households tend to have above median household income.
Posted by: Bernard Yomtov at Nov 16, 2008 7:54:22 PM
"There must be a pretty strong carryover from year to year?"
Not so much. The top quintile is significantly determined by capital gains returns. The people that end up at the top in any given year have likely taken on significant risk in one way or another.
Take for example the variability of income. If you take the standard deviation of average income by quintile, divided by the long run average one should not be surprised by the results...
From Bottom to top quintile (higher % implies more risk):
8.31%
6.29%
7.74%
9.61%
21.34%
Posted by: Jay at Nov 16, 2008 11:25:54 PM
Deflating wages by the PCE deflator is a valid approach.
But deflating fringe benefits part of compensation by the PCE deflator significantly overstates the growth in real fringe benefits. This is because the main reason fringe benefits have grown faster than wages is the more rapid rise of health insurance cost. So if you deflate this large portion of compensation by the overall PCE deflator you are using a biases estimator that significantly understates the inflation rate in fringe benefits.
Posted by: spencer at Nov 17, 2008 2:05:49 PM
People are delaying marriage until later in life, but as personal income increases, fewer young adults are forced to live with their parents. This leads to a huge number of households with one person in them at the lower end of the income distribution, dragging down the median household income.
Posted by: Stormy Dragon at Nov 19, 2008 4:23:56 PM
