Felix Salmon

Chart of the day, median income edition

By Felix Salmon
October 10, 2011

Why has no one thought to do this before? Every month, the Current Population Survey goes out to a nationally representative sample of more than 50,000 interviewed households and their members. And in one of the questions, those households — or at least the households who didn’t answer the same question the previous month — are asked how much money they made, in total, over the past 12 months. That question has now been asked in 138 successive months, since January 2000. Which means that with a bit of clever analysis, it’s possible to put together an apples-to-apples comparison of what has happened to household income every month.

And when you do that, the results are very scary indeed.


The red line, here, is median real household income, as gleaned from the CPS, indexed to January 2000=100. It’s now at 89.4, which means that real incomes are more than 10% lower today than they were over a decade ago.

More striking still is the huge erosion in incomes over the course of the supposed “recovery” — the most recent two years, since the Great Recession ended. From January 2000 through the end of the recession, household incomes fluctuated, but basically stayed in a band within 2 percentage points either side of the 98 level. Once it had fallen to 96 when the recession ended, it would have been reasonable to assume some mean reversion at that point — that with the recovery it would fight its way back up towards 98 or even 100.

Instead, it fell off a cliff, and is now below 90.

In dollar terms, median household income is now $49,909, down $3,609 — or 6.7% — in the two years since the recession ended. It was as high as $55,309 in December 2007, when the recession began.

Some of this decline has been hard to see because nominal incomes have been holding very steady: before taking inflation into account, median household income was $51,465 in December 2007, and $51,140 in June 2009. But even then, over the past two years, nominal incomes have shrunk significantly to the current level of $49,909.

All of these numbers come from Gordon Green and John Coder, economists who both worked at the Census Bureau for more than 25 years. They’ve now set up a private company, Sentier Research, to collate these household income figures every month; the full report costs a reasonable $20.

Why is this work being outsourced to private-sector economists, rather than being done by the Bureau of Labor Statistics and published officially? I’m having dinner with a government statistics wonk on Wednesday, and will be sure to ask him.

But in the absence of any good reason to discount the reliability of these numbers, it’s definitely worth taking them seriously, and asking why incomes have eroded so quickly and dramatically over the past two years. We’ve known for years that America has a huge unemployment problem. But I had no idea that the plight of the employed was this bad.

21 comments so far | RSS Comments RSS

All productivity gains acruing to capital and not broad labor.

More beneficial taxation of capital as opposed to labor. Compounded with that force of nature which is compound interest. A policy which is focussed on making rent seekers whole, an erosion of collective bargaining power in the labor force. A real asymetry of power in society and representation.

On top probably a depreciation of capital employed in the U.S. in real terms, reducing real productivity per hour worked (GM -> Walmart).

A central bank policy which can be reduced to the provocative line: We can create as much money as we want, as long as employees don’t get a hold of it to cause inflation in goods and services.

Posted by Finster | Report as abusive

Interesting observation, thanks for bringing it to our attention.

There is an obvious explanation. Suppose that the wages for the employed stay steady, but that we have a 5% increase in unemployment. Thus, 5% of households will have their income cut in half (assuming most households are two income). This will cause a drop in the reported numbers.

Is that effect enough to explain the drop in your graph or is there additional downward pressure on wages? Locally, I know that state workers in FL took a 3% cut, which is an affect both of the effect of the Lesser Depression on tax revenues and lack of a competitive labor market. I hope somebody takes up this challenge and looks into this.

Posted by ralmond | Report as abusive

In a situation where income inequality is growing, I think a chart of the median (not mean) income over time is crucial. The only argument in favor of letting the superrich become superricher (beyond “people have a right to what they’ve earned”) is that, even though CEOs make very high multiples of what line workers or middle managers make, the line workers or middle managers are still better off in absolute terms than they would be if (say) high income were taxed at higher rates. If wealth is very concentrated at the top, you can come up with stats showing that society is wealthier in the aggregate; but median income info can tell a different story.

Posted by jfruh | Report as abusive

Actually, the fall in the mean household income was not accompanied by any significant fall in the mean persoanl income as estimated from the same CPS data
http://mechonomic.blogspot.com/2011/09/o n-some-methodical-mistakes-in.html

Posted by ikitov | Report as abusive

What Finster said.

Posted by FCBonanno | Report as abusive

It would be interesting to see U-6 plotted against this graph as opposed to the headline unemployment.

Posted by djiddish98 | Report as abusive

Nevermind, U6 largely mimicking headline unemployment

http://research.stlouisfed.org/fredgraph .png?g=2FL

Posted by djiddish98 | Report as abusive

A big piece missed here is the role of inflation. By this chart, income went up in the recession (because of disinflation) and then dropped from 2009 to now because prices moved higher (even though prices are still lower than pre-recession). Thus, income, which have moved higher since the recession (but not by much) shows a real big drop because of the yo-yoing in prices. As prices come down now we’ll see the opposite, but unemployment will still be high.

Posted by finn0123 | Report as abusive

Median income goes up during the recession as the lower-paid (note I do not say “skilled”) workers lose their jobs while the Senior Execs keep theirs (and/or receive higher wages).

Notably, the recovery from the Bush recession (which was worse for employment from a private-sector-recovery perspective than the current one) shows median income rising for a while until the private sector recovers. (The number of people employed in the Private Sector at the beginning of the 2001 recession is only reached again–absolute number, not as percent of workers–in May of 2005.)

Posted by klhoughton | Report as abusive

klhoughton, if a member of a lower-income household loses their job, it does not affect the MEDIAN at all. The median only drops if some of those who were previously ABOVE the median earn less.

Neither the median nor the mean is a complete representation of a distribution. If the distribution is normal, then the two will be the same. If it is heavily skew, as income tends to be, then the median will be significantly lower than the mean. But neither observation is sufficient to explain what is happening here.

Anybody have the data to plot and compare the distribution before and after?

I’ll note, also, that the median income was more or less constant through ~2009. The 10% drop has all been recognized in the past two years — and shows no sign of slowing.

Posted by TFF | Report as abusive

well its not apples to apples because households change over time–if the response of a 25 year old to losing his job is to move into his parent’s place, then that eliminates a low-income household from the data and makes household income rise… it’s strange that the income of the median man fell most in 2008 and 2009 and income of the median woman fell most in 2008 and 2009, and income of the median married-couple family with kids fell most in 2008 and 2009, but then household income fell most in 2010 and 2011…suggests composition of households is changing.

Posted by adamoneill933 | Report as abusive

“suggests composition of households is changing”

Supported by anecdotal data as well… Young adults aren’t moving out (or are moving back with parents). May be changing in other ways as well.

Posted by TFF | Report as abusive

Oops — I got that backwards.

Maybe kids were moving in with parents for 2008-2009 but are now moving out again?

Posted by TFF | Report as abusive

I have three brilliant ideas about how to resolve this problem. My ideas are so good, they are actually the policies of the United States.

1. Let’s run a $600 billion a year trade deficit and throw away 5 million jobs. We don’t need them anyway. What’s 5 million jobs among friends? Of course, trade deficits also strongly predict subsequent economic crises. I guess admitting the “free” trade is a poisoned chalice is unthinkable among sensible people.

2. Let’s open our markets to China while China systematically hollows out the U.S. economy. China’s policies have brought vast miserly to the U.S. in recent years. A careful study “The China Syndrome: Local Labor Market Effects of Import Competition in the United States.⇤” found that China accounts for at least 1/4th of U.S. manufacturing job losses. Not a big problem living in D.C. presumably. Let Ohio rot.

3. Let’s import millions and millions of unskilled workers (and welfare recipients) even though we are creating zero new jobs over a ten year period and public finances are in disarray. Immigrants (legal and illegal) are literally replacing Americans on a staggering scale. I guess unemployed Americans should get a Ph.D. in molecular biology and stop complaining.

You see, solving America’s economic problems is so easy we are already doing it.

Posted by peterschaeffer | Report as abusive

50 years ago:
http://www.youtube.com/watch?v=NPlcE3Gco Fc

not so much

(though I am not to optimistic that it looks better in much of europe. in particular, german inflation adjusted income went down quite a bit the last years, in an attempt to increase competitiveness of the german industry).

Posted by GPEB | Report as abusive

It may be worth bearing in mind that total compensation statistics, which include the employer’s contribution to health insurance, have been rising even while take-home has been falling.

Posted by TomMaguire | Report as abusive

This chart will continue on for the next 10 years in the same fashion. Rich western workers who earn 5 times the global mean wage will need to continually prove their worth to the multinational companies which employ them. These companies (Wal-mart, GE, and every other member of the Fortune 500) account for a larger and larger % of global GDP each year.

What can a 25 year old American do for GE at $30,000/year that a better educated 25 year old South Korean won’t do better for $20,000.

I’ve said it a dozen times and I’ll say it a dozen more. Keep building up your skills and save and invest like crazy because this is as good as it gets for the rest of your life if you live in the rich west.

Posted by y2kurtus | Report as abusive

It is interesting to note that where these lines cross represents roughly the March ’09 market low. The correlation escapes me, if there is one. Any perspectives?

Posted by netvet | Report as abusive

The cross is purely dependant on the scaling choice on the left and right axis. No information to be gained from that.

Posted by Finster | Report as abusive

As has been mentioned, the crossing of the lines is a completely false and misleading fabrication based on the axis scaling.

For this data to be represented meaningfully would require either 1) two charts on which each data set is plotted with no interaction between the two or 2) one chart with a common axis, displaying something like the % deviation from the start point.

This chart is a rather unnecessary distortion of data. I am certain that the message in the data is interesting, but we can’t see it clearly with this chart.

Posted by jlbriggs | Report as abusive

Regardless of the unemployment rate, when you look at the trend starting in Jan. 2000, the slope is only interrupted by the housing bubble (2006-2008). This generated an increase in income; all the builders, real estate agents, bankers, and house flippers. When that burst, we went right back to where we would have been. I don’t see a definite correlation between the two pieces of data. It’s like saying, “when ice cream sales go up, there are more shark attacks”. Maybe there is another piece of data that ties to two together – like it’s a hot summer so people go swimming and eat ice cream. I would like to see a chart of the household income index for 1982 to present to get an idea on how well trickle down economics has worked.

Posted by djstreck | Report as abusive

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/