A look at U.S. income growth

From Jordan Weissmann:

Hat tip from @JustinWolfers.

By the way, Robert Gordon has a new paper pessimistic about future economic growth,  using this tagline at the end of the abstract:  “A provocative “exercise in subtraction” suggests that future growth in consumption per capita for the bottom 99 percent of the income distribution could fall below 0.5 percent per year for an extended period of decades.”  Mark P. Mills is more optimistic.


I'm certainly no pessimist when it comes to technological advancement. Frankly, I think the best is yet to come - especially in robotics, AI and medicine. However, I am pessimistic with regard to income distribution. The economic rewards of technological advancement seem to be falling into fewer and fewer pockets.

It would be interesting to see the above chart but also with a breakout for the top 1% or even the top 5%. Also, would like to see the above for total wealth.

Agreed. Even the top 1% is too broad. The top 0.1% is where a great majority of the increases show up.

I was going to type the same thing. I'd guess the top 5% splits into half or so which looks much like the top 20%, in terms of trends, and the rest which look like the top 1%. So it would have been better to show results for the top 1% and let people deduce the top 5% by interpolation.

Nate and MPS, take a look at Figure 2 in the paper by Emmanel Saez: http://elsa.berkeley.edu/~saez/saez-UStopincomes-2010.pdf You are right that the action is up in the top of the income distribution, but I don't think that really changes the basic message of the above chart.

It would be interesting to see what definition of income is used (hopefully an inclusive one) and to see one for consumption.

It would also be interesting to see what is happening with actual people over their lifetime, as opposed to arbitrary and constantly shifting "buckets" of people.

But that would spoil the beauty of the message.

You can only plot stuff for which data exists. Is there a reliable dataset that tracks what you are asking for?

Rich, not much beauty in the hollowing out of the middle class. More substantively, researchers have looked at income histories for the same person over several years and income across generations of the same family. Some of the highest quality income data comes from IRS tax records or Social Security earnings files. But these don't have a lot of demographics so other studies use the Panel Study on Income Dynamics (here you get wealth and a bit of consumption too). I don't have the meta study to cite (Piketty and Saez is a good place to start) but while you're right there's year-to-year shocks and there are life-cycle concerns (like retirement), there's also a lot of persistence in wealth, income and consumption. So I'll agree that these details might steal some beauty from the chart, but I doubt it would dramatically alter its message.

I think a response to many of these claims is Alan Reynolds " Income and Wealth". For example, in response to the oft-repeated claim that the middle class is vanishing, he points out many of these measures used a fixed definition of middle class; the number is shrinking because people are getting richer and moving out of the group.

His work can be found at Cato and he was just recently on CSPAN's Washington Journal. I&W is a bit pricy, and is not available for Kindle, unfortunately.

This is what I wish the election were about.

You mean the sudden rise of the middle/upper-middle class? Or the trend from 1980-2000 that became of interest ten years after it ended and looks quite different when viewing consumption?

I mean the failure of the non-upper-classes to reap the benefits from all our productivity growth over the last 30-40 years.

Have you given any thought to the fact that we are importing non-upper class by the millions to drive down inputs per worker? Has it occurred to you that pumping a trillion simoleons in high-powered money into the economy inordinately benefits the early recipients of the new dollars?

To try to drag on a silver lining: Is the 2000-2010 decade the one with the lowest inflation of all these?

Average median income per annum rise of 1% on the graph during the 70s strongly implies these are real growth numbers, though I admit to not looking up the Census report myself

I looked it up at Pew and you are right.

I think its important to remember that the chart above comes from PEW and not from Weissman or Wolfers – PEW deserves the link backs/shout outs.

A lot of improper tagging these days.

Forgive me, but the data says that the incomes of the top 5% were falling faster in the 2000-2010 period that any other group except than the lowest 5th.

Seems expected when the end of the decade is the collapse of an asset bubble that had been growing over the first eight years.

What I want to see are state transition probabilities for people in each quintile (plus top 5%, 1% and .1%) and age cohort. So if someone is 21-25 years old and in the first quintile, what's the probability they transition into a higher quintile within 5 or 10 years. Same for someone who is 61-65 and in the top 1% transitioning down.

This paper has some info: http://www.bos.frb.org/economic/wp/wp2011/wp1110.htm), but not by age. However, the 10 year transition probabilities for the bottom and top quintiles are both over 40%!

This is Mean Family Income by household. Average household size has been falling for decades. I think the same chart on a per capita basis would be more informative.


Household income is too affected by non-economic factors (family size, divorce rates, mate choice). See for instance, this political calculations blog post The Real Story Behind "Rising" U.S. Income Inequality.

I briefly looked over your link, and their post wherein they show "that there has been absolutely no significant change in the level of inequality among U.S. individual income earners from 1994 through 2010" takes its data from census tables that cut off at $100,000 (that is, they lump everyone in that bracket together). This is, of course, where the bulk of the inequality growth has been. This is also probably why they give Gini coefficients indicating much more equality than normal. Unless I'm badly mistaken, that site is garbage.

That is a deficiency in the Census data on individual income, but my point in linking to the post was not about individual income inequality, but the problem of using household or family data to find trends in U.S. income.

This piece by the Federal Reserve Bank of Minneapolis: Where Has All the Income Gone? highlights some of the problems with using household data from the U.S. Census Bureau (as is used in the graph above). For example: "Average household size declined substantially during the past 30 years, so household income is being spread across fewer people. The mix of household types—married versus single, young versus old—also changed considerably, so the “median household” in 2006 looks quite different from the “median household” in 1976."


"The Census Bureau uses a narrow definition of income in its report on median household income that focuses on money income and excludes nonmonetary sources of income. The BEA, in contrast, defines personal income as income received from all sources. Examples of income excluded by the Census Bureau, but included by the BEA, are employer contributions to employee pension and insurance funds and in-kind transfer payments such as Medicaid, food stamps and energy assistance. These sources of income contribute to economic well-being and should be included in the definition of income."

I tend to disregard anyone trying to weave narrative on U.S. income using household data alone.

And I tend to disregard any source that includes a bald-faced lie in their second sentence, as yours does. They go on to repeat that lie several times, and your entire link is based on the same misuse of census data. Why would you try to pass them off as an authority? It is not a deficiency in the Census data, it is their dishonest use of it that is the problem.

It doesn't take much for you to think that something is a blatant lie rather than a misinterpretation of the data. Whether or not individual income inequality has risen (which it has) or not does not change their central idea that social factors (in this case mate selection) rather than economic factors have driven a rise in household income inequality. For that they only have to show that the rise in individual income inequality is less than the rise in household income inequality which it is.

Here is an NBER paper that deals with the top coding problem of the Census data on individual income. They find that it does cause an understatement of individual income inequality, and that there has been an increase since 93' but that the smaller increase since 93' remains:

"Because consistently topcoded public use and internal data do not contain information
about the very highest incomes, they understate the level of income inequality. However, once
the break between 1992 and 1993 is controlled for, income inequality based on all these data
series yield the same trends, namely a rise in inequality between 1975 and 1992 followed by a
significantly smaller increase in inequality after 1993. Our key findings are generally robust to
whether income inequality is summarized using the Gini coefficient or three other widely-used

Here is an older paper which finds that changes in earned income inequality cannot even explain half of the change in family income inequality from 1979-1996. Among the causes for the rest of the increase in inequality are "growing positive correlation of husbands’ and wives’ earnings" (i.e., mate selection as in the explanation of the political calculation blog) and "a sharp decline in the proportion of Americans who live in families where a married couple is present."

I don't know why this graph is getting so much play. The data is getting less descriptive. Why aren't the 30's and 40's included? Do we really need talk about income distribution to be further simplified? The goal should be to get more people thinking about the questions raised by Burkhauser's research, not pump out more infographics that contain little information.

Is this trend diminished, mirrored, or more extreme if you look at wealth as opposed to income?

I understand that food stamps, for instance, don't count as part of "income".

Actual consumption, including gov't benefits, is what should be discussed if "poverty" is a significant part of the issue.

I don't believe any more than 1%, if that, of US citizens are "poor" based on yearly consumption (including soup kitchen and charity aid). Good data about objective poverty levels (less than X calories per day, for instance), would be good. I think with honest definitions, most non-rich in America are middle class -- enough food, clothes, and a place to live with indoor plumbing, heating in the winter.

Is that with or without debt?

How many of the top 5% were downsized out of a job or took early retirement. That was the case with me and I can tell you that of course my salary dropped significantly as I moved to being a pensioner! In my field, pharmaceutical R&D, there were huge layoffs as the result of mergers throughout the industry. Several hundred people that I worked with during the decade were not employed by the end of it. Fortunately, most had nice buy out packages are could take early retirement. Of course this is anecdotal but given the large downsizing that took place coupled with the financial melt down the 'large' apparent decrease in the top 5% is not surprising to this former member of the club.

Your experience is precisely what such graphs do not show. They do not follow the same people because who is in each group changes significantly over time.

It has everything to do with energy. EROEI rates are falling leading to us getting less done.

It would be interesting to see this beside a graph of stock market returns for each decade.

Does this include benefits or adjust for inflation? My understanding is that there is no gap between mean income and productivity growth once you include benefits. Median income does trail downwards due to positive skewness, even including benefits, but mean income doesn't.

Even if it is fatally flawed for not including benefits, it still shows the huge cost of employer-provided, tax-deductible health insurance. It is a serious issue that most health care funding comes from people spending 10k-20k a year on insurance, and they don't even realize health care is costing them anything.

Your understanding is incorrect.

Compensation includes all benefits including health care insurance.

I seriously doubt that many people do not know that their health insurance is paid for by their employer..

The only thing excluded from compensation is stock options.That might significantly impact total compensation, but it would have a minimal impact on mean income.

It does show inflation is not bad for the poor.

No, it doesn't really show that inflation is not bad for the poor. The graph is income not expenses. High inflation tends to hit fuel and food costs disproportionately, which also happen to make up a higher portion of the Poor's budget.

I presume these are real income growth figures (adjusted for inflation).
What I don't get is - how can income growth be negative for all quintiles from 2000-2010? Given that the economy has grown, albeit bumpily, over the past 10 years?

The main story of the graph seems to be the stagnation in the bottom quintile incomes since 1970.
I think it merits five alternative explanations, all of which are partly true.

A) Post war reconstruction boosted bottom quintile incomes in the 50s/60s.
B) Wartime destruction in Europe/Japan meant that American producers had less competition from foreign imports when vying for domestic demand
C) Increase in the size of govt and welfare programs and Keynesian policies resulted in more equitable growth - an explanation embraced by the Left
D) The cultural revolution in the 60s/70s - resulting in higher divorce rates and the single mother phenomenon hurt the prospects of the bottom quintile post 1970s - an explanation embraced by conservatives.
E) The emergence of hi-tech/knowledge economy meant that the poorly educated bottom quintile was ill-equipped to reap the rewards of economic growth post 1970s.

The negative growth from 2000 to 2010 across quintiles is still hard to explain, unless household size has dropped substantially over the past 10 years.

Per capita real GDP is still well below it's 2007 peak.

That's Hope n' Change in action for you!

While median household income has slipped by $3,719 (census.gov) between 2000 and 2010, wages have essentially stagnated during the same period. The main reason behind the decrease in median household income is that one of the spouses works less hours now. Overall, this trend has not helped our consumer driven economy. That’s why we are stuck in the first gear in term of growth.

We are doomed.

Comments for this post are closed