Slow Income Growth for Middle America

The economic challenges and strains facing middle-class Americans are likely to get a good bit of attention between now and election day, at least from the Obama campaign. They include sluggish income growth, heightened financial insecurity, rising health care and college costs, and falling home values. Each of these is important, but the most critical in my view is slow growth of incomes.

The following chart tells the story. It shows inflation-adjusted GDP per capita and median family income from 1947 (the earliest year for which the income data are available) to 2007. To facilitate comparison of the over-time trends, each is indexed to its 1973 level. Since the mid-to-late 1970s, growth of income at the median has been slow — very slow — relative to growth of the economy. The current decade, with no improvement at all in median income, is especially striking.

The dashed line in the next chart shows what median income would have looked like had it risen in sync with per capita GDP. The difference is huge: in 2007, the median family’s income would have been $91,000 instead of $61,000.

Various excuses and rationalizations have been offered: It’s okay because Americans now get more in employer benefits instead of in their paycheck. Family size has shrunk, so slow income growth isn’t a big deal. A lot of those in the bottom half are immigrants, and even with slow income growth they’re better off than they would have been in their native country. None of these is compelling (see here or here).

The disconnect between economic growth and middle-class income growth is due largely to rising inequality. In the past several decades much of the economy’s growth has gone to those at the top of the income distribution.

Faster income growth wouldn’t render other middle-class strains irrelevant. But it would help.

61 thoughts on “Slow Income Growth for Middle America

  1. It would also be interesting to see number of hours worked per year, and total compensation per hour versus per hour output, as those are the really important numbers for defining living standards and quality of life.

    Also, the same for the middle class, not just for the working class, which is what you are really talking about.

    As to that the middle class is defined not by being in the middle of the income pyramid, but by being in the middle of the functional classes, with a supervisory or professional role; that is being a trustie or a scribe of the ruling class.

    A McDonalds “manager” is a burger factory foreman, working class, not middle class. To some extent many previously middle class jobs have been sunk into the working class by commoditizing them and the educational qualifications they require. In particular the vast majority of white collar jobs are currently working class.

    Median family income is almost by definition working class (except perhaps in Luxembourg or Kuwait), even if white collar working class. It used to be that white/blue collar largely matched the functional divide between working and middle class, but no more.

    I understand that working class people are as snobbish as anybody else and want to think that they are middle class (as in “my collar is white!”), but someone might argue that gives them false class consciousness.

    Thus usually middle class earnings are in the top quintile, let’s say from the 2nd to to the 20th percentile; the ruling class does not reward with a middle class income what they see as bulk headcount. To have a traditional middle class lifestyle one needs earnings starting at least at the 20th percentile.

  2. I think you misunderstand one of those criticisms (aka “excuses”). The criticism of your chart is that you’re comparing apples and oranges, i.e. per family income to per person income. If you think per family numbers are more compelling, then use mean family incomes instead of mean per capita incomes.

    Personally, I think family is a pretty arbitrary construct that is variable in its meaning over time and space. This makes it, at least to me, a pretty bad unit of study. But whatever, to each his own.

    Anyway, I agree that the data show there is growing inequality, but your chart gratuitously distorts the magnitude of that trend. This means your claim of “slow — very slow” relative growth is unfounded.

  3. Pingback: Real Time Economics : Secondary Sources: Inflation, Income Growth, Globalization

  4. This is not a particularly good post.

    Comparing per-capita GDP and median incomes mixes apples and oranges. There are many points here.

    1. GDP is calculated using the GDP deflator. Median incomes are adjusted using the CPI. After WWII, the GDP deflator and the CPI tracked. However, starting in the 1970s they diverged and have continued to move separately. A quote from Dean Baker should help

    “The consumption deflator used to measure real wages has shown a much higher rate of inflation than the output deflator used to measure productivity growth. This is due to the fact that the price of many consumer goods and services, like health care and education, have risen considerably more rapidly than the price of investment goods like computers. If a consumption deflator is used to measure output, then the rate of annual productivity growth is reduced by 0.2 percentage points in the period from 1973 to 2006. In the period from 1947 to 1973 the consumption deflator actually increased less rapidly than the output deflator.”

    2. No one can consume GDP. NDP (GDP minus depreciation) is the potential basis for consumer outlays, investment, etc. For various reasons, NDP is a declining fraction of GDP.

    As mentioned above, Dean Baker has written on this subject. See “Falling Wage Shares” ( and “The Productivity to Paycheck Gap: What the Data Show” (

    Adjusting for these factor markedly reduces the gap between per-capita GDP growth and median incomes. It does not eliminate it however.

  5. Pingback: links for 2008-09-04 - Kevin Bondelli’s Youth Vote Blog

  6. IANAE (I an not an economist) but…

    The increasing disparity in wealth is undisputed, even by people who disagree with the numbers shown. People can disagree with the extent of the inequality may differ, but the fact that the wealthy have gotten a larger share of the newly created money isn’t.

    It seems to me that this ultimately weakens the economy and can [ultimately] hurts the wealthy. The percentage of wealth that gets spent decreases as a person’s wealth increases. The economy is roughly 2/3 consumer spending, therefore as a larger portion of the economy becomes concentrated at the top, a smaller portion of overall wealth gets spent.

    Could someone with a stronger background in economics give their viewpoint of this? Am I missing something?

  7. Peter, comparing medians to means is a convenient way to look at the change in distribution. If the mean is increasing faster than the median, then its very likely that there’s increasing inequality (as measured by the gini index, for example). Its not comparing apples and oranges. Also, there’s no reason medians have to be deflated differently than means. I’m not sure why you think this.

    Richard, I agree that massive increases in inequality can be bad. It would especially be bad if income growth was stagnant or negative for most people and only a small elite were getting richer. In fact, most incomes have been rising and the professor wants to make the case that there have been large differences in the growth rates between income groups (i.e. both rich and poor have gotten richer, but the rich have become much richer than the poor). As I explain above, his evidence, the graph, is misleading because it compares family medians to individual averages so its not clear what the magnitudes are.

    But suppose we grant the professor his conclusion, suppose everyone is getting richer but the rich much more so. Is this really a problem? Does it make sense to enact policy (e.g. higher taxes on the rich) that makes more equal growth rates of incomes across income classes but runs the risk of reducing everyone’s income growth rates? What model of human behavior justifies this policy trade off?

    Maybe an example will help. Suppose under current policy, my neighbor’s income increases by 25% every year and mine only goes up 10% per year. Now suppose there was a policy that would decrease my growth rate to 5% a year and his to 10% a year. The new policy would make growth rates more equal, but reduce everyone’s growth rates. Should I support that policy? Why?

  8. Wow, I actually did a very similar analysis a few months ago, except I used the Census Bureau’s (actually Lebergott’s) median value of a full time job. Of course, not everyone can get a full time job, but I figured I could get a sense of how our society values work. I came up with very similar results. My divergence began in the 1960s, but the ratio of a full time salary to per capita GDP has been moving downhill since. You can see my numbers at:

    I also did a related analysis on the cost of a home in hours per year. Basically I used the median wage, the median home price and the median mortgage rate. Here, the big drop was in the late 70s and early 80s when a house went from 600 to 800 hours, with some scary highs of over a 1000 hours in the golden Reagan years. A typical work year is less than 2000 hours. You can check my work at:

  9. pushmedia1,

    Comparing means to medians is quite legitimate. What is not appropriate is to compare medians based on the CPI to means based on the GDP deflator. Note that the CPI has significantly diverged from the GDP deflator in recent decades.

  10. Peter, what evidence do you have that the medians and means reported by the professor use different price deflaters?

    You’re right that this would be a mistake if the professor did it, but without having the data in front of you how can you know he made this mistake?

  11. If families are smaller, you would expect families to earn less because there are fewer workers per family. You would also expect no effect on GDP per capita.

  12. Pingback: The Ambrosini Critique » Blog Archive » I like my redistribution with a side of toast

  13. To say that the disparity is due to income disparity is stating the obvious end result but not what it is ‘due to’. In my opinion it is due to two things: A more educated populus and the USA running a Current Account deficit.

    1) From 1977 to now the percentage of Americans that have college degrees has gone from 10% to 25%. We should all be thankful for the work that these people have put in because our increasingly complex and technical world needs more educated people. Those with these degrees get more money, obviously. This also shows up in the income distribution ranked by education. In 1977 the average college person earned only 10% more than the average high school graduate. Now it is about 60%. So, the income disparity has come from a new education disparity. Its all pretty logical.

    2) Since the 1990’s our current account has gotten massively in the red. Using China as an example they want to sell us goods but do not want to buy goods from us so the money they have gotten from us gets recycled into our capital markets. (They may buy Treasuries but it means that others in America don’t have to so their money now goes into stocks, real estate, etc.) These increased capital flows has had the normal effect: it raised valuations which has benefited people that have assets in the first place, ie those college graduates. So, they are getting richer.
    But if China was buying our goods instead of hoarding capital we’d have jobs and income for ordinary workers and the income disparity would be a lot less.

  14. Pushmedia1,

    “Peter, what evidence do you have that the medians and means reported by the professor use different price deflaters?”

    I downloaded the data from the BLS, Census, EROP, and BEA and reproduced the chart above. Just as I stated above, GDP was deflated using the GDP deflator and median family incomes were deflated using the CPI-U. I can send you a spreadsheet with all (lots and lots) of the data.

    However, there are (at least) two more errors worth noting.

    1. Per-Capita GDP is a poor measure of economic performance because it fails to take into account the worker / population ration. This ratio has risen dramatically since WWII. Stated differently, per-worker GDP has risen much less than people think, but this has been masked by a strongly rising worker / population ratio.

    2. The number of workers per-family has risen strongly since WWII (female labor force entry). This should have caused a stronger rise in per-family incomes. In other words, the actual data understate how badly the median family is doing by not taking into account the rising number of workers per-family.

    As you can see comparing per-capita GDP to median family incomes is a muddled mess.

  15. Peter and pushmedia1,

    The points you make are good ones. I tried to keep this simple, but perhaps went too far in that direction. A more straightforward comparison is of the trends in productivity and median compensation. But I think family/household outcomes are of greater relevance, since earnings (and benefits) are typically pooled. For this, GDP per person (or family) is better than productivity.

    Using net national (or domestic) product, deflating with the CPI-U-RS (instead of output deflator), and measuring per family rather than per person, the rise since 1973 is less than for GDP per capita. The median family income using this counterfactual would have been $83,000 in 2007, rather than $91,000. Lower, but still far above the actual level of $61,000. And there’s the additional fact of many more families now having two earners.


  16. Lane, is it not true that the number of low income households with two earners has declined while the number of high income households with two earners has increased dramatically?

  17. Lane Kenworthy,

    First, I would like to acknowledge an error on my part. At several points, I have suggested that the median family income data was deflated using the CPI (as in CPI-U). This is not correct, the median family income data is actually deflated using CPI-U-RS. Unfortunately, this creates yet another problem. The CPI-U-RS has risen substantially less than the standard CPI (CPI-U). This makes the trend line for median family incomes look better than it would if the CPI-U was used.

    Why the Census uses the CPI-U-RS when the rest of the government utilizes the CPI-U isn’t clear. TIPS and Social Security are adjusted using the CPI-U. The BLS does appear to use the CPI-W. However, the CPI-W and CPI-U aren’t very different.

    I agree that median family income is a useful of well being. However, it has the very serious problem that the number of earners has risen over time (as you state). I find that real median family income has risen just 6.7% since 1973 if you deflate the nominal value using the CPI-U. The increase using the CPI-U-RS is 22.3%. I don’t have the numbers at hand. However, I am reasonably sure that the number of earners per household has risen more than 6.7% and possibly more than 22.3%. This means that earnings per working family member may well have declined at the median.

    To get a better understanding of the data, I checked Net National Income per-worker. NNI has risen 52.1% using the GDP deflator since 1973. Using CPI-U-RS the increase is 40.5%. Using CPI-U it is only 22.7%.

    Over the same period mean (not median) wages for production workers (as defined by the BLS) have fallen 16.7% measured using CPI-W, 18.0% using CPI-U, and 6.1% using CPI-U-RS.

    Clearly the data supports a very large divergence between output per-worker (or family) and actual wages. However, I don’t think the data is clearly understandable unless it is done on a apples to apples basis.

  18. This blog appears to to demonstrate two points :

    a) Economists are unable to communicate meaningful aggregate income measures in an understandable fashion (at least understandable to sociologists)

    b) Sociologists shouldn’t dabble in other fields

  19. “It seems to me that this [income inequality] ultimately weakens the economy”

    Absolutely right. John Kenneth Galbraith in his classic “The Great Crash, 1929″ listed high income disparity as the number one reason for the crash and resulting depression.

    What doesn’t receive nearly enough discussion, in my opinion, is that extreme income inequality can also weaken and even endanger our democracy. How many more decades can middle and lower class families be expected to put more workers and hours into the workplace for little or no real gain in income before they snap? With the extraordinary financial pressures mounting on families due to excessive debt, mounting costs for health care and education, and historical destruction of asset values, I would expect the snap to come sooner rather than later.

  20. 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 — comparing income to income) increased 100% 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 (should be somewhere inside the ballpark):
    1st quintile income grew 175.4%. (I mis-remembered above that my adjustment doubles top quintile income—rather it approaches doubling income growth.)
    Best inequality graph I’ve seen in a long time:

  21. Pingback: Sometimes a chart says more than 1,000 words: GDP growth vs. median wage stagnation | Observationalism

  22. I admire your work and your charts and wonder if you have any suggestion as to where I might find graphs or smoothable data for various countries having as the abscissa the percentage of the population and as the ordinate the percentage of the gross income of all persons earning less than a person at that percent point (including those with no income).

    Such a curve goes from the origin to the gross national annual earnings on the y-axis and the slope is proportional to the income of persons on the x-axis. If the lower half of the earners aggregated one quarter of the gross the x mid-point of the curve would pass through a point one quarter as high in the y direction as the top point, which is the gross national earnings.

    In all countries the real curve is flat at the beginning, representing children and beggars living under bridges, and the last part is an almost vertical spike. These two very important regions are not detailed in most graphs. The census, for example, manages to ignore them by dealing only with quintiles.

  23. Pingback: How the Economy Failed or “Dragons We Know” | One Eyed Guide

  24. Dear Sir:

    I like your paper very much. I want to write a paper with above pictures. Would you please give me ti permission?

    Thank you for your consideration.

    Best regards.

    Chenguang Lu

  25. Pingback: The Universal Matrix – Artigos » Blog Archive » Queda na Inovação: Dica para aplicar a Matriz na prática.

  26. Pingback: Tyler Cowen: missing the point and more « Comments on Global Political Economy

  27. Pingback: Monday Reads « Sky Dancing

  28. Pingback: The great decoupling « Consider the Evidence

  29. Pingback: How Globalization affects Countries and Markets « Philosophy of Market Economics

  30. Pingback: Chart of the day « occasional links & commentary

  31. Pingback: “The Great Decoupling” | oolaah

  32. Pingback: Is the Internet Destroying Our Standard of Living? | BNET

  33. Higher unemployment and off shoring have kept salaries low. The fed is not supportive of low unemployment rates. They want growth without inflation. Keeping wages low is the best way to keep inflation low. Its the feds version of cutting expenses to make their numbers.

  34. Pingback: We’re Richer Than We Think « Building a Better GOP

  35. Pingback: As the wealth of America went to the top 1%, the rest of us just stagnated « A Man With A Ph.D.

  36. Pingback: Things We're Supposed To Be Quiet About -

  37. Pingback: Things We’re Supposed To Be Quiet About | CATA NEWS

  38. Pingback: I hate Reagan, episode 69,468 squared | The Hackenblog

  39. Pingback: Dr. Hackenbush Gains Perspective » Blog Archive » It’s morning in America and 99% of us are working the night shift

  40. RE: the often-made comment in the last paragraph from Pushmedia1 above: And what if the reality is exactly the OPPOSITE: if the effect of marginally increasing the share of the pie to those at the top DECREASES the growth rate of the size of the pie? The two greatest increases in income disparity over the last century were in the years leading up to 1929, and to 2008.

  41. Pingback: Class Warfare? It Should Be. « Political Nutz

  42. Pingback: ITESM IQ80 » Antes era mas facil—


  44. Pingback: クルーグマン「ぼくらが黙ってなきゃいけないらしい問題:経済成長と所得」(2012年1月14日) – 道草

  45. Pingback: Things We’re Supposed To Be Quiet About – « aepxc

  46. Pingback: Romney: Income inequality is just 'envy' - Page 13 - City-Data Forum

  47. Pingback: Anonymous

  48. Pingback: Why is the middle class disappearing? Actual data. | Occupy Friday Harbor

  49. Pingback: IT job cuts — does the cloud have a silver lining? - Domicity Ltd.

  50. Pingback: The dumping and trade debate in 8 paragraphs « Antidumping Blog

  51. Pingback: Message to my Mother 4 « A Man With A Ph.D.

  52. Pingback: Who’s driving? | Stephen K Stetler's Blog

  53. Pingback: Something’s missing… « American Bulwark

  54. Pingback: The Right-to-Work, 4.1% Higher Wage Myth | Dirigo Blue | Maine's Source for Progressive Political News

  55. Pingback: The Right-to-Work, 4.1% Higher Wage Myth « J.uris D.ebtor

  56. Pingback: Amy Volk, Oklahoma, Teachers, and Right to Work Legislation . . . Oh My! | Dirigo Blue | Maine's Source for Progressive Political News

  57. Pingback: The Right-to-Work, 4.1% Higher Wage Myth | Pine Treeconomics

  58. Pingback: Amy Volk, Oklahoma, Teachers, and Right to Work Legislation . . . Oh My! | Pine Treeconomics

  59. Pingback: Mark Perry on income inequality | Pine Treeconomics

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s