Where has US household income gone ?

by John Q on September 13, 2008

I was at a seminar the other week on inequality in US household income, and I asked the speaker about something that’s puzzled me for a while. I didn’t really get an answer, so rather than do a lot of work myself, I thought I’d try this crowdsourcing all the cool kids are talking about. Here’s the puzzle.

Over the past 40 years or so, real median US household income has risen by about 30 per cent.

US Household income 1965-2005

US Household income 1965-2005

but real US GDP per person has more than doubled. How can this be ?

I’ve done enough work to rule out a couple of easy answers. Average household size has fallen from around 3 to 2.6, but that’s not enough to account for more than a small part of the gap. And inequality as measured by the ratio of mean to median household income has gone up, but again, not enough to account for the gap. In fact, even top quintile income hasn’t quite kept up with GDP per person

Income by quintile

Income by quintile

So it seems as if the ratio of total household income to GDP must have fallen. Where has the extra income gone.

My candidate answers:
(1) A big chunk of income goes to the top 1 per cent of households and isn’t captured by the survey. The seminar gave some support to this idea, at least insofar as this group seems to have a big enough share of the total that the choice of the point where the Census Bureau stops measuring income makes a big difference
(2) A lot of income is flowing to the corporate sector and never being recorded as household income, perhaps because it is distributed in the form of capital gains, which aren’t counted. Again, a very large chunk of these would go to the top 1 per cent.



Cryptic Ned 09.13.08 at 6:56 am

The price of health insurance premiums in our bizarre system has gone up by about 100% in the last decade. Could that factor in? Since it’s probably included in statistics for worker compensation, and yet the increase in the market price doesn’t actually make any worker any better compensated; even though it can entail employers spending a lot more on the “wages/salaries/benefits” categories, all the increase goes directly to insurance companies and hospitals.


notsneaky 09.13.08 at 7:20 am

How much income, exactly (or roughly), is missing?


Sebastian 09.13.08 at 8:13 am

Could it be related to the tax-guided reluctance of corporations to make dividend payments?


abb1 09.13.08 at 8:29 am

You can convert company income into capital gains? Sounds like a great swindle.


John Quiggin 09.13.08 at 9:41 am

notsneaky, I think around 25 per cent of income, but this is v rough

Sebastian and #abb1, two sides of the same coin.


pushmedia1 09.13.08 at 9:48 am

HH size hasn’t shrunk much, but it matters how its shrinking. Mom’s still home and the kids, but Dad’s not sticking around. The number of children (age 0-16) living in households without a father (or stepfather) went from about 11% in 1960 to 29% in 2005. (source: IPUMS)


SA 09.13.08 at 10:39 am


I think this should answer your question…


peggy 09.13.08 at 11:09 am

Don’t forget that household income should have expandedas vast numbers of women entered the workforce.


Kenny Easwaran 09.13.08 at 11:27 am

I’m very much not an expert and don’t even really know what GDP means, but is it possible that each dollar that a household spends now goes through several more businesses before getting turned into someone’s salary? Although this explanation would predict that GDP per capita should be somewhat higher than actual income, because in addition to all the income for individuals, there should be all the extra times that money gets given to corporations before it’s given back to an individual, while your data suggest that median income is several thousand times higher than GDP per capita, unless I’m reading the charts wrong.

But again, I don’t really know what GDP is, so I don’t know if each transaction would get counted the way I’m suggesting.


Stuart 09.13.08 at 11:48 am

Is net increase in debt (federal, local, personal, business) a factor here? I am not 100% on how GDP is calculated, but my understanding is that government spending and personal consumption are included, so for example the $680b increase in the national debt in the last year at federal level alone could account for over $2k per capita difference.


Dave 09.13.08 at 12:01 pm

Corporate executives’ Cayman Islands bank accounts?


Tim Worstall 09.13.08 at 12:04 pm

This is something of a guess….it may or may not fill the whole gap but it will certainly fill some of it.

From the Census guys. pg 3


The Census Bureau recognizes that
measuring money income may not
completely capture the economic wellbeing
of individuals and families.4
Families and individuals also derive
economic well-being from noncash
benefits, such as food stamps and
housing subsidies, and they have
reductions in disposable income due
to taxes. While the income and
poverty estimates shown in this report
are based solely on money income
before taxes and do not include the
value of noncash benefits, the Census
Bureau computes a number of alternative
measures of income and poverty
that do attempt to account for those

What I take from that is that the numbers being used for real household income are not in fact real household income. They are real household money income. So employer provided health care (a substantial portion of compensation I think?) is not included in the given figures. Very roughly, if 13% of GDP is health care and 50% of that comes from the Govt (Medi-care and -aid) and the other 6.5% of GDP is employer provided then we’ve got some of the gap.

The other part is that well known problem with the way poverty statistics are calculated in the US. It is cash market income, leaving out anything provided in kind (Medicaid or care, so that’s another chunk of that 13% of GDP which is health care, housing vouchers, food stamps) or through the tax system (the EITC).

I’m sure I’ve garbled bits of the above but I would say that at least some of the gap JQ is puzzling over is that the Census is not measuring “total” household income but only cash household income, leaving out all benefits in kind, like health care, and such things as the EITC.

I’ve no idea whether that is enough to explain the whole gap.


abb1 09.13.08 at 12:23 pm

Hmm, Tim, but when a Medicaid recipient receives medical care – does it count as a part of the GDP?
Obviously gov. paying the doctor is included into both GDP and doctor’s income, but the step where the Medicaid customer is getting this non-monetary benefit is probably not included into either statistic.

If that’s true, then while it’s correct to say that monetary income doesn’t exactly capture economic wellbeing, it still doesn’t help to explain the discrepancy.


jim 09.13.08 at 12:23 pm

Saudi Arabia?


Tim Worstall 09.13.08 at 1:37 pm

“Obviously gov. paying the doctor is included into both GDP and doctor’s income, but the step where the Medicaid customer is getting this non-monetary benefit is probably not included into either statistic.”


That’s my point. That Medicare is definitely included into GDP figures, but isn’t included as part of household income. (This also works for regular health care.) Thus, as medical expenses become an ever larger part of the economy we’ll see that household income doesn’t rise as fast as GDP…because we’re counting such medical care as part of GDP but not of household income.

Not I don’t claim by any means that this is the whole story. Only that it might be part of it.


Jim Payne 09.13.08 at 2:01 pm

GDP is computed by adding up all the expenditures by people, businesses and government. Accountant’s would consider these to be debits. Household income is the other side of the accountant’s model and is classed as a credit. The difference between the debits and credits is our missing number and it also must be a credit. The biggest missing credit that I can see is profits that were retained by corporations and not distributed to the shareholders.

There is probably an economist somewhere that has written a paper on this issue and it certainly would be an interesting read.


jj2 09.13.08 at 2:06 pm

Medical insurance fraud. Treatment and testing centers charging insurance companies for services neither extended nor received. Someone’s got to pay for all the MRI machines to which everyone wants immediate access.


Kaveh Hemmat 09.13.08 at 2:09 pm

Peggy @7 makes a very interesting point.

Apparently 59% of women participated in the labor force in 2004 (73% of men). If the figure in ’65 was a lot lower, say 19% (I have no idea how close that is). If all men and women live in 1-man-1-woman-households (which they don’t, of course) then, if income per job stayed the same, household income should have increased by 50% just from women entering the workforce.


Kaveh Hemmat 09.13.08 at 2:10 pm

Peggy @7 makes a very interesting point.

Apparently 59% of women participated in the labor force in 2004 (73% of men). If the figure in ‘65 was a lot lower, say 19% (I have no idea how close that is). If all men and women live in 1-man-1-woman-households (which they don’t, of course) then, if income per job stayed the same, household income should have increased by around 40% just from women entering the workforce.


abb1 09.13.08 at 2:12 pm

That Medicare is definitely included into GDP figures, but isn’t included as part of household income.

Let’s say the government is A, the Medicare consumer is B and the doctor is C. The visit is $100.

You’re saying that it should be recorded as A transferring $100 to B and then B transferring $100 to C. You believe that this thing will be counted as $200 in the GDP book, but as only $100 in the Income book.

What I’m saying is that, perhaps, for the purposes of both GDP and Income accounting this is recorded as simply A transferring $100 to C. In which case this shouldn’t create any discrepancy – it’s counted as $100 in both books. I could be wrong, of course, I’m just guessing.


Denis Drew 09.13.08 at 2:20 pm

I once worked out that the Census practice of top coding income over $1,000,000 per family (not households — families have a more even number of members so I assume a more even inequality observation) may hide as much as half of all top quintile income.

Per capita income doubled since 1968. If you add up all mean family income quintiles in 1968 (when inequality was not severe and the top code served a useful purpose of not skewering the overall picture with a few billionaires’ incomes) and double that sum the result should square with adding up all mean family incomes in 2005. Any discrepancy should tell us (very roughly; per capita is not family — but still with somewhere inside the ballpark) what is missing from today’s top quintile report. Turns out that the discrepancy is about equal to the to report, suggesting that half of top family quintile report is missing from the report.

I think this resulted in top quintile growing from 40% in 1968 to almost 60% in 2005 — instead of “only” 50%.


Denis Drew 09.13.08 at 2:26 pm

I caught this fragment from Dean Bakers blogspot once — it is my main informant of where the income went:

Dean Baker * (in 18th reply on his blog post) reproduced what he called “a slightly altered table from Gordon’s ** paper, showing income shares in 1972 and 2001” — my percentage changes on the right.

0-20_______2.6%, _ 2.0%________- .6%__ -12.3%
20-50____ 16.0%, _ 11.7%_______ -4.3%__ -11.7%
50-80____ 33.7%, _ 27.2%_______-6.5%____ -7.4%
80-90____ 17.0%,_ 16.1%________ – .9%___ –
90-95____ 10.8%,_ 11.3%______ +_ .5% __+
95-99.0___12.2%,_ 14.8%______ +2.6% ___+ 3.1%
99.0-99.9__ 5.7%,__ 9.6%_______+3.9% ___+ 7.0%
99.9 -100__ 1.9%,__ 7.3%_______ +5.4%__ +12.4%

(see p. 84 of Gordon** for similar breakdown of wage income)

4.9% loss of overall share meant 26.3% cut of 0-50 percentile share.
6.4% loss of overall share meant 14.5% cut of 50-90 percentile share.


rea 09.13.08 at 2:44 pm

jj2, you think the difference between 30% of the GDP and 100% of the GDP is accounted for by health insurance fraud? My goodness!


Denis Drew 09.13.08 at 2:47 pm

Why did it happen — as opposed to where did it go?

Lower 90 percentile earners allowed their bargaining power slip through complacency. We had a much happier nineteenth labor history on this side of the Atlantic — recent unhappiness blamed on the “great compression” — more recent happiness taken for granted as productivity boomed for decades (as long as everyone is making money, there was no big pressure on the bottom) and the post compression winners and losers had not yet sorted themselves out (fewer winners to beat up on losers).

Meanwhile the American labor majority let their bargaining power slip…
…today’s underpowered remedial proposals, inadequate even in earlier times and poorer places — Obama’s $1 short of LBJ’s minimum wage proposal and the 1930s card check retread (instead of modern sector-wide labor agreements as practiced in third, second as well as first world economies) — illustrate the depth of the trench into which today’s labor has let itself fall…

…while the top utilized their bargaining power like never before.

All of which resulted in a very finite shift of 12.5% (as of 2001 — not a good year for top few percentile incomes according to Dean Baker) from the bottom 90 percentile to the top 3%. If you squeeze a toothpaste tube from the bottom up, it all comes out the top.

Since the top 3 percentile have always been highly educated we can assume it is not the bottom 90 percentile falling behind educationally that is at fault here.


Denis Drew 09.13.08 at 2:50 pm


Chris Ashley 09.13.08 at 3:08 pm

Elizabeth Warren explains it all. Technically, I suppose she’s answering a different question than the one John Quiggin poses; but the first ten minutes or so of the linked lecture are really dynamite.


OneEyedMan 09.13.08 at 3:35 pm

I’m with pushmedia1 . The composition of the population is different, and since there are far less children who don’t work but increase household size this. Healthier old people may also be able to work longer. That probably still isn’t enough.

What about government transfers? GDP per person is an average number which should essentially ignore transfers. Median household income is an order statistic which could be moved a lot by transfers. Although, given what I know about income inequality I would think that would move it in the other direction. Are these incomes net of government transfers?

I think the real answer is that the super rich, the group in the top 1% and top .1% have in fact got much richer and that pulls up of the average more than the median.


Charlie 09.13.08 at 4:48 pm

referencing comment 7 and 16:
I have often wondered if the huge number of women entering the workforce have actually had almost no effect on total household incomes. If labour was priced at a certain level, then an increase in available labour (all those women entering the workforce) would lower the price. In short, more work is being done by more people, but at a lower price. so where you had one wage earner in 1950, now you have two in 2008. But instead of now having two full incomes, you have maybe 1.2, because not only do women earn less in the first place (as has been frequently documented) but all that extra labour in the market has caused the price of labour to drop in real terms.

If anyone could point me at actual work done by actual economists in this area, I would be grateful.


ejh 09.13.08 at 5:28 pm

…the company store?


Alan 09.13.08 at 10:15 pm

Also note the recent discussion at Big Picture which describes the large difference in Q2 between real GDP and income caused by the absurd action of the GDP deflator, which actually grows smaller with the rise in imported oil.


OneEyedMan 09.13.08 at 10:16 pm

Charlie, the women entering the workforce ended up being a net positive for the size of the labor force. Since someone paid to employ these new people, it stands to reason that they were engaged in productive labor. That means that the total amount of labor supplied increased. Under standard assumptions of economic production this implies that wages (from labor) went down. That’s because if the supply of labor increased faster than the stock of capital then the marginal product of labor and marginal product of capital increased.

However, since that labor was productive, total output increased. It just happens that more of the fruits of production went to the owners of capital. But that’s going to show up in household income and GDP, because my understanding is that dividends, interest income and capital gains all show up in income measures and certainly are a part of GDP.


Joel Turnipseed 09.13.08 at 10:42 pm

Wouldn’t an important data set here be the rate of foreign investment in U.S. securities during this period (perhaps with associated data like dividend payments and stock buy-backs)?

It wouldn’t cover all your gap, but if there was a lot of increased foreign investment–especially in blue-chip, divident-paying companies or in companies that participated in significant stock buy-backs–that could account for a rise in GDP that was not matched by a rise in household wealth (especially since the productivity gains have come nowhere close to trickling down to wages, proportionally).


Joel Turnipseed 09.13.08 at 10:45 pm

Two other thoughts:

1) What does the “mean” income gap look like?

2) Given the time period involved and the math, the difference between 30% and 100% total growth over forty years implies a death by some very small cuts.


I, Kahn O'Clast 09.13.08 at 10:49 pm

You are merely describing the rise in per worker productivity.

We produce ever more with technology (machines…. from robots to computers to giant farm equipment)

That’s really all it is.

eg If one man can produce twice the widgets today than he could so many years ago but his pay rises only 30% we calculate that as a productivity gain…..


ehj2 09.14.08 at 2:39 am

An excellent article I’ve seen on the subject by Elizabeth Warren and Amelia Tyagi that appeared in Boston Review.

Elizabeth Warren is the Leo Gottlieb Professor of Law at Harvard Law School. Warren and Tyagi’s article is adapted from their book The Two-Income Trap: Why Middle-Class Parents are Going Broke with the permission of Basic Books, a member of the Perseus Books Group.

Amelia Warren Tyagi worked as a consultant with McKinsey and Company and co-founded the health benefits firm HealthAllies. She lives in Los Angeles.



Peter Schaeffer 09.14.08 at 5:17 am


This is no mystery, just basic national income math. It turns out that there are four factors that entirely account for the schism between median household income and per-capita GDP. However, let’s start with the basics. From 1967 to 2007 per-capita GDP rose by a factor of 2.1757 (117.57%). Over the same period median household income rose by a factor of 1.2956 (29.56%). The four factors are:

1. The ratio of Net National Income to GDP has fallen by 1.59% or an adjustment factor 0.984096. The main reason is faster depreciation of capital goods (PCs and the like).

2. The CPI-U-RS (used to deflate household incomes) has risen considerably faster than the GDP deflator. The primary reason is that the GDP deflator includes many goods that consumers never buy (capital goods). These have fallen in price relative to consumer goods over the years (housing, medical care, etc.). The shift is 7.62% of an adjustment factor of 0.92376.

3. Median household size has fallen from 3.268 in 1967 to 2.586 in 2007. This is an adjustment factor of 0.7915.

4. The ratio of the median household income to the mean household income has fallen by 16.9%. This is an adjustment factor of 0.83099.

If you multiply the four adjust factors, you get a total adjustment factor of 0.5979. Multiplying the total adjust factor times the growth in per-capita GDP gives almost the exact change in median family income over the period in question.

I think most folks know about the fall in median household size and the increase in the mean to median household income ratio. However, the faster growth in the CPI-U-RS versus GDP deflator is probably less known. The shift in the ratio of NNI to GDP is obscure.


John Quiggin 09.14.08 at 5:28 am

Thanks for this. I was unaware of the CPI/GDP factor, but it seems to fill the gap.


Dan Kervick 09.14.08 at 5:52 am


First, since we are talking about median household income as opposed to average household income, that would remove some of the mystery. According to the same Wikipedia article, total household income in the top quintile looks to have almost doubled, in the second it has gone up about 60% while in the middle quintile it has gone up about 35% and in the bottom two quintiles it has flatlined. Since the top two quintiles represent the lion’s share of the income, and since median income probably tracks pretty close to the middle quintile, it looks like total household income has gone up at a much higher rate than median household income. So perhaps this means businesses are using that increased income, as measured by GDP growth, to pay much higher salaries to the their top wage-earners, and not so much more to the middle and bottom wage-earners.

But more importantly, even if we were talking about average household income, not median, it would be a simple fallacy to expect that income number to rise at the same rate as GDP. It is easier to make the point if we look at individual income as opposed to household income. As you say, GDP has almost doubled since 1965. If the population had remained constant during that time, then we would expect individual income to double as well. But if the population had doubled during that time, we would expect household income to have remained constant: the overall growth in the economy would then be due entirely to more workers, not more productivity per worker.

But, in fact, it looks like the US population is about now 1.55 times the size of the population in 1965. Thus dividing the doubled GDP into the smaller growth in population, we get 2/1.55 or 1.33. So, we would expect individual incomes to have gone up by about 33% – not far from the 30% you found.


Richard H. Serlin 09.14.08 at 5:58 am

I went over this, and I think I have some good answers for you. Here’s my analysis:

The bottom line is that:

“real median US household income has risen by about 30 per cent.” and “US GDP per person has more than doubled”

mathematically, would have to be explained 100% by:

“Average household size has fallen from around 3 to 2.6”


“inequality as measured by the ratio of mean to median household income”

if the following are true:

Assumption 1) You get rid of the “around”s and “more than”s and plug in the precise numbers. For example, you plug in the exact amount real median US household income has risen, rather than “about 30 per cent”.

Assumption 2) You assume the income used in “GDP per Person” is the same as the income used in “Average Household Income”. In other words:

a) GDP per Person = Total National Income/Number of People

b) Average Household Income = Total National Income/Number of Households

As long as Total National Income is measured the same for both of these, as long as it’s the same number, then you’re fine. If not, then this is a source of your problem, and a great lead. You should look at how these are measured differently.

Assumption 3) There’s no measurement errors, math errors, or other mistakes.

Now, if all of this is true, then the following is true:

First, let’s write what you want to explain in mathematical notation:
“real median US household income has risen by about 30 per cent.” and ” US GDP per person has more than doubled”

Can be written mathematically as:

(M1/M0) / [(I1/P1)/(I0/P0)] ; Let’s call this Term 1,


Mi = The median household income at time i

Ii = Total National Income at time i

Pi = The number of people in the country at time i

Now, what are you trying to explain this Term 1 with?

a) “Average household size has fallen from around 3 to 2.6”, this is the gross change in average household size. Mathematically, it can be written as:


where Ni = The average number of people in a household at time i.

b) “inequality as measured by the ratio of mean to median household income”. Mathematically that can be written as:

[(I1/N1)/M1] / [(I0/N0)/M0] ; Let’s call this Term 2,


Ii = Total National Income at time i.

Ni = The average number of people in a household at time i.

Mi = The median household income at time i

Next, if you do some algebra, you will see that:

Term 1 = (N1/N0) / (Term 2) ; let’s call this Equation 1.

which shows that what you are trying to explain, Term 1, can, in fact, be completely explained, 100%, by the things you said couldn’t completely explain it, and were wondering why, namely (N1/N0) and Term 2. These can, in fact, explain it 100%, but only as long as the assumptions 1-3 that I’ve made above are correct. So you want to look for problems in these assumptions.

Now, let’s put our relationship equation, Equation 1, into words, to make things clearer:

Equation 1 means:

(The Gross Percent Change in Median Household Income) / (The Gross Percent Change in GDP Per Person)


(The Gross Percent Change in Household Size) / (The Gross Percent Change in the Ratio of Average Household Income to Median Household Income).

So, what’s the effect if the Household Size stays the same — meaning The Gross Percent Change in Household Size = 1, but the denominator doubles, that is The Gross Percent Change in the Ratio of Average Household Income to Median Household Income = 2?

If that happens, then the left hand side of equation 3 is cut in half; GDP Per Person grows twice as much as Median Household Income.

So, bottom line, what you have should explain everything, but only as long as assumptions 1-3 are true. The key assumption I’d look at is 2, the assumption that the income used in “GDP per Person” is the same as the income used in “Average Household Income”. For example, they may not be including a lot of transfer payments in household income, or they may not be including all benefits, bonuses, and dividends and other capital income. And then there’s income that people hide for reasons like tax evasion and other illegal activities.

Consider, UCLA economist Emmanuel Saez’s March, 2008 paper, “Striking it Richer: The Evolution of Top Incomes in the United States (Update using 2006 preliminary estimates)” at: http://elsa.berkeley.edu/~saez/saez-UStopincomes-2006prel.pdf . He writes:

We define income as the sum of all income components reported on tax returns (wages and salaries, pensions received, profits from businesses, capital income such as dividends, interest, or rents, and realized capital gains) before individual income taxes. We exclude government transfers such as Social Security retirement benefits or unemployment compensation benefits from our income definition. Therefore, our income measure is defined as market income before individual income taxes. (page 1)

End Quote

I don’t know exactly what data sources you’re using, but the answer to your puzzle certainly may lie in the details of how they are compiled.

Another point I’d like to add is that for income inequality, looking only at how quintiles have done is really misleading, because there has been so much increased inequality within the top quintile. It’s in the top 1% and the top 1/10th of 1% that you really see an explosion in income inequality.

Saez’s data series’ (available at http://elsa.berkeley.edu/~saez/) show that in 1978 the top 1% received 8 times the average income; by 2006, this soared to 23 times. For the top .01% the increase was from 86 times the average income to 546 times! At about the same time, Republican tax cuts starting in the Reagan administration and continuing through Bush II have made taxes far less progressive. The top federal income tax rate was cut by 35 points between 1979 and 2006. Average tax rates on the richest 0.01% were cut in half between 1970 and 2006, while taxes on the middle class were increased (From Princeton economist Paul Kurgan’s most recent book, “The Conscience of a Liberal” (pages 145) and his 2007 column, “Gilded Once More”, at: http://select.nytimes.com/2007/04/27/opinion/27krugman.html?_r=1&oref=slogin ). Another great source is the Economic Policy Institute’s annual book, “The State of Working America”.


Dan Kervick 09.14.08 at 6:00 am

Oops. Screw my above comment. I didn’t see you were looking at GDP per person, John, not GDP. So my last two paragraphs don’t apply.

As others have said, what about health care and other benefits? Household income only measures cash payments, right? If businesses have shifted a lot of their compensation from salaries to benefits, that would mean that increases in monetary income might not match increases in overall compensation.


Katherine 09.14.08 at 7:54 am

Could someone translate #36 for non-economists? Thanks.


John Quiggin 09.14.08 at 9:52 am

Update with explanation coming up, I hope.


Amol Shelat 09.14.08 at 12:01 pm

Hi John,

I agree with Peter Schaeffer. It looks like the GDP per person link that you put up for this post refers to nominal GDP per person, not real GDP per person. Also, you need to watch out for the inflation index that’s used on a particular data series (GDP deflator, CPI Index, PPI Index, etc.).

First, real GDP uses a price deflator (I believe this is true but I might be wrong). So, instead of using the price deflator, I calculated real GDP using the CPI Price Index. With this measure, real GDP per person only grew 76% from 1967 to 2007, not 200%. I believe that CPI is also the same measure used to add inflation effects back into household income data. Thus, using the CPI index puts the two measures on equal footing.

Secondly, if you factor in the decrease in household size from 3.3 to 2.57 over the last 40 years, voila, you get the 38% change in household income you’re looking at.



John Emerson 09.14.08 at 1:45 pm

Coming in late, but it really astonishes and infuriates me that a question as important as this one, the big political story of the last decade or two as far as I am concerned, is still only in the early stages of analysis.

Brad DeLong and Paul Krugman have been talking now and then for several years about the paradox of a jobless recovery, or the paradox of flat wages and rising productivity, but they speak of it as a baffling mystery and go on to other things that they’re more interested in and better able to handle. And they’re the liberals.

Instead we get debates about whether offshoring and immigration have had a very slight negative effect on American labor, or no negative effects at all. from the liberals.

It seems to me that for the last five years ago this problem should have been at the top of the agenda, in screaming red letters, for all non-Republican economists. But it’s been on the back burner, by and large.

I’m regarded as a troll around here regarding economics, and an ignorant one because I don’t understand economics from the inside, and for that reason I don’t post here much any more on that topic, but looking at economics from the outside, judging it by its fruits, it still seems to me to be a crippled science with a nasty rightward bias.


Denis Drew 09.14.08 at 1:49 pm

Quiggin and Shaeffer,
Does Shaeffer’s figuring (way over my pay grade) account for a quarter of the US workforce earning less than the minimum wage of 1968 as of June last year (before the first meager incremental increase) — double the average income later?

Does it account for the same gap between median income and per capita (average) output not opening up in Europe (where labor is not asleep at the bargaining wheel: sector-wide labor agreements and all that kind of thing)?

PS. For whatever it might mean to the figures: the Census inflation (CPI-U-RS — used exclusively it seems by the Census) yields lower inflation and higher growth than the BLS inflation (CPI-U — which almost everyone else follows). And again, since households can have only one person, a better look a distribution spread might be the family survey.


MQ 09.14.08 at 5:08 pm

The CPI/GDP deflator issue has been popping up as an issue for over a decade now, since the productivity growth/wage growth differential first showed up in the 90s. Economists have paid some attention to it, but not enough. Here’s Menzie Chinn on recent divergence:



Denis Drew 09.14.08 at 6:36 pm

Shaeffer’s numbers suggest that median income did not really go anywhere (I think that is what he means). I think that may be safely disputed by showing that median income share very clearly did go somewhere: downhill with the rest of 90 percentile earners.

First, Census table shows per capita income (comparing apples to apples now) doubled from 1967 through 2007).

Census mean family income quintiles table show:
5th quintile mean grew 22.4% over that span;
4th grew 31.4%;
3rd mean (effectively median) grew 47.3%;
2nd grew 64.6%;
1st (w/o adjusting for top coding) grew 95.8%.

If you play my little top coding adjustment game using per capita income for your overall growth gauge — as described above:
1st quintile income grew 175.4%. (I mis-remembered above that my adjustment doubles top quintile income — rather it approaches doubling income growth.)


Peter Schaeffer 09.15.08 at 2:36 am

Denis Drew,

Yes, it does. Income inequality in 1967 or 2007 shows up as a higher mean household income versus median household income. Note that this ratio has risen by 20.34% over the period in question.

Since incomes were not evenly distributed in 1967, the 20.34% is an increase from the 1967 level of inequality to the 2007 level of inequality in household incomes.


The changing mean to mean household income ratio is not the only, or even the best measure of income inequality. Far from it.

I recommend the household and family Gini’s if only because they are accepted standard. Note that the family and household Gini’s have risen considerably as well.

Note that median household income did rise by 29.56% (at least if you deflate nominal household income using the CPI-U-RS).


PenGun 09.15.08 at 3:27 am

Short answer: The books are cooked.

Corporations run your country for reasons that benefit them. That is really enough of a clue for smart people. The result now unfolding will be ugly … very ugly. Hi ho.


Curmudgeon 09.15.08 at 4:18 am

To reiterate posts 26 and 35, Elizabeth Warren’s work lays this out very clearly. Household income has disappeared because unavoidable expenses (e.g. health care, food and education) have risen faster then incomes.


gandhi 09.15.08 at 5:12 am

The BIG question is not where the money went, but how on earth do you get it back?


bad Jim 09.15.08 at 8:58 am

Fiftieth comment and no one has mentioned real estate?

Back in the late 70’s real estate started getting more expensive and women had to work in order for a family to be able to afford a house. Household income went up because more people were working. Eventually productivity rose, due at least in part to increasing computer use, GDP rose, and working people didn’t get to share because, among other things, we had Republicans in power to make sure that the wealthiest got to skim the cream.


John Emerson 09.15.08 at 11:09 am

Both Krugman and DeLong have speculated that “power” might be the reason. As I understand, “power” is a political concept, not an economic concept, and I think that pretty much anyone except an economist would conclude that after five strongly anti-labor Republican administrations and two neoliberal Democratic administration (weakly pro-labor at best, but then there’s Free Trade) that you’d expect that politics had a lot to do with the stagnation of wages.

It was the trained incapacity of economic professionals, who have been indoctrinated to take a deliberately incomplete model of human society as complete, which caused Krugman and DeLong to stand around sucking their fingers and talking about a paradox. And they’re the good guys.


Denis Drew 09.15.08 at 2:14 pm

I’M SO GLAD YOU ASKED. Here is my sidelined cabdriver megalomaniac plan. Thing is, if an old high school educated cabbie can hatch a reasonable sounding bounce back program out for pretty much boiler plate concepts (no mission from Mars or Marx stuff) — every economic professional out there ought to have one (or more) comprehensive plans of their own.
How to legislatively (re)impose a fair and balanced American labor market:
First, double the minimum wage to $13/hr over three years (a dollar every six months?) – and – legally guarantee inflation adjustments for incomes under $100,000.

Doubling the minimum wage could potentially add an average 50% more pay to below 50 percentile earnings ($13/hr being today’s 35 percentile wage) – accompanied by only (an easily computed) 3%* direct price increases plus perhaps (?) 3% more after other wages are pushed up – a minimum wage-force multiplier.

Next, legislatively introduce French-Canadian style (lite) sector-wide labor agreements to the US labor market (airline and supermarket employees would kill for sector-wide contracts) – and – legally mandate union certification and re-certification elections (every four years?) at every work place (periodic re-certification could clean up the most common objections to unions: entrenched, complacent or even corrupt leaderships).

Top 10 percentile incomes enjoy 40 percent of the take these days (up from 27.5% in 1973) – plenty of headroom there for the mid 50-90 percentile to rake back more missing share points through higher labor prices – a collective bargaining-force multiplier.

Finally, (at least temporarily?) hike marginal tax rates (75% over $500,000, $1,000,000?). Folks earning 2500% more than folks doing the same work 25-35 years ago will not return all the way to earth through 12.5-25% price increases – erode a force multiplier.

America’s lower 90 percentile earners never think to impose legislative hegemony to recoup the 12.5%** income share they have lost to top 3 percentile since 1973 – their unemployed force multiplier.

* http://ontodayspagelinks.blogspot.com/2008/08/3-cost-of-gdp-output-and-inflation.html
** http://ontodayspagelinks.blogspot.com/2008/08/income-share.html


abb1 09.15.08 at 2:56 pm

I don’t think doubling the minimum wage is enough, and anyway it has to be inflation indexed or linked to some index (like the poverty line for the family of four or something).

Also, the top tax rate has to go much higher than 75% and permanently – it can be argued that this whole thing started when the Kennedy/Johnson admins dropped the top rate from 91% to 70%.

Also, a wealth tax would be a very good idea at this point.


aynrand 09.15.08 at 8:13 pm

Oh, sure. Let’s raise taxes on the most productive (in terms of wealth generating) citizens. Even though income has risen faster on the top, the reason it has risen at all on the bottom is because of the actions taken by our most productive members.

It’s unfortunate that people think that income equality would raise the poor up when in reality, it often just drags the rich down.

A marginal income tax of 75% on the top would just mean that the top earners would devote their time to tax avoidance rather than engage in wealth building activities.

The cons of wealth equality are so much greater than the anticipated benefits.


gandhi 09.15.08 at 10:56 pm

The problem with giving money to the poor and taking it from the rich nowadays is that the rich will just take their money somewhere else (e.g. Liechtenstein). It’s not like they are all stuck in Sherwood Forest any more.

We really need some kind of global “people power” movement here.


abb1 09.16.08 at 6:19 am

It may be a problem in Sherwood, but it’s hardly a problem for the US of A, which just weeks ago forced the UBS to completely terminate their off-shore services for the US clients. Just like that, and they didn’t even have to bomb Zurich once.


Matt E Ryan 09.16.08 at 10:53 am

Yeah, let us have a world were we have everyone being equal, all in misery. Let the government of the US decide what everyone around the GLOBE should have. That is real development.


abb1 09.16.08 at 3:09 pm

A marginal income tax of 75% on the top would just mean that the top earners would devote their time to tax avoidance rather than engage in wealth building activities.

These extremely greedy people who really-really need to be making tens of millions of dollars every year, these freaks – I’d strongly prefer that they become criminals, bank robbers – rather than captains of industry.


LFC 09.16.08 at 8:16 pm

@56: well, you certainly chose an apt moniker.


abb1 09.16.08 at 8:41 pm

apt moniker

I don’t think so. This commenter said:

A marginal income tax of 75% on the top would just mean that the top earners would devote their time to tax avoidance rather than engage in wealth building activities.

Now, think about it, take The Fountainhead, for example. The moral of this (very naive) story is exactly the opposite: the protagonist guy wants, above all, to build his buildings – while not caring at all about the money. It’s precisely the bad, contemptible guy who’s preoccupied with making money and would probably devote his time to tax avoidance.

See? They don’t even understand their own prophet and desecrate her name. Sad, very sad.


Asher 09.16.08 at 11:09 pm

The crux of this discussion comes down to the follow: what exactly are we measuring. One problem is that we are not actually comparing apples to apples; i.e. the same people who we are measuring now are significantly different from the people being measured, historically. A good example is the very large percentage of the American population from Latin America that makes far more in the US than they would make in their countries of recent origin.

Let’s say that you have a 20 year old woman who arrived from mexico in 1988 and immediately had a son. Now, if that son currently worked in Mexico his income might be a couple thousand dollars, but it is now ten or more thousand. Additionally, if you are operating under the premise that this new worker is replacing a retiring white worker you’re forced to look into whether or not you are replacing like with like. And you’re not. The average IQ of ethnic mestizos, who account for nearly all immigration from Mexico, is around 88, while the average IQ of currently retiring workers is roughly 100. You can’t replace a 100 IQ with an 88 IQ and expect similar results.

This just demonstrates that if you’re looking at aggregate data over an extended period of time you really need to be careful that you’re measuring the same thing. An analogy might be measuring the height distribution of the Roman Senate during the late empire, a period of time when the Empire completely lost it’s Roman flavor and turned completely multi-ethnic; the Romans were shorter than the Germans.


Richard H. Serlin 09.17.08 at 12:35 am


We had much greater income and GDP growth when there was much greater equality. For an explanation why, please see my posts:




To whet your appetite, here’s part of the reason why:

Finally, what about Republican claims that tax cuts will make people work more hours because it increases their pay per hour? First, people today, by and large, work so hard, and spend so little time with their families by historical standards and compared to people in other countries, that it’s not at all clear that this is desirable, and it’s not even physically possible to work many more hours at this point. In addition, there is a point where production is decreased from more hours, because eventually it really hurts the quality of work. People become tired and burnt out. Competence and creativity is hurt. But aside from any of this, it has been shown in economics that in fact there is little long term relationship between tax rates and work hours. For most of the 20th century real wages per hour went up greatly at the same time that hours worked dropped. There’s little long term effect, and what effect there is can easily go in the other direction. Cutting taxes can decrease work hours. A key reason is the long ago established and accepted in economics, income and substitution effects, which you can find in any university microeconomics text.

The idea is this: if you raise someone’s wage from $8/hour to $12, they may go from 40 hours/week to 45, because they get $4 more for giving up an hour; that’s the substitution effect. But if you raise their wage from $8/hour to $1 million/hour, they will probably work like 40 hours per year! And then spend the rest of the year enjoying all of that money. That’s the income effect. When you raise someone’s wage per hour, they don’t have to work as many hours to live in a way that they consider well. Empirically, it appears that with taxes at about their current level, the income effect becomes greater at about the middle class level, and then we get what’s called a backward bend to the labor supply curve. Cornell economist Robert Frank has a nice brief New York Times Economics Scene article explaining all of this, “In the Real World of Work and Wages, Trickle-Down Theories Don’t Hold Up” (at: http://www.nytimes.com/2007/04/12/business/12scene.html?ex=1334030400&en=a58a714793b92179&ei=5124&partner=permalink&exprod=permalink).


abb1 09.17.08 at 7:23 am

First, people today, by and large, work so hard, and spend so little time with their families by historical standards and compared to people in other countries, that it’s not at all clear that this is desirable, and it’s not even physically possible to work many more hours at this point.

No only that, but I have to question the whole premise: desirable to whom, why would anyone want me to work longer hours? I certainly don’t want to work more hours, nor do I care whether you work longer or shorter hours. What is this all about?

Comments on this entry are closed.