The Most Important Economic Chart

March 18, 2014

If you must know only one fact about the U.S. economy, it should be this chart:


The chart shows that productivity, or output per hour of work, has quadrupled since 1947 in the United States. This is a spectacular achievement by an advanced economy.

The gains in productivity were quite widely shared from 1947 to 1980. Real income for the median U.S. family doubled during this time just as output per hour of work performed doubled. The rising tide was lifting all boats.

However, what we want to focus on today is the remarkable separation in productivity and median real income since 1980. While the United States is producing twice as much per hour of work today compared to 1980, a small part of the gain in real income has gone to the bottom half of the income distribution. The gap between productivity and median real income is at an historic all-time high today.

So where are all of the gains in productivity going? Two places:

First, owners of capital are getting a bigger share of GDP than before. In other words, the share of profits has risen faster than wages. Second, the highest paid workers are getting a bigger share of the wages that go to labor.

The net result is that families at the higher end of the income distribution have received more of the income produced by the economy since the 1980s. The latter fact has been documented meticulously by the brilliant research of Thomas Piketty and Emmanuel Saez.

The widening gap between productivity and median income is a defining issue of our time. It is not just about inequality – important as that issue is. The widening gap between productivity and median income has serious implications for macroeconomic stability and financial crises. Our forthcoming book takes up these issues in more detail.

We will also discuss some of these issues in coming posts.

Tags: ,

Bookmark and Share

83 Responses to The Most Important Economic Chart

  1. Ryan on March 18, 2014 at 9:21 ami

    How much of the leveling off of median income is a composition effect? Are there any actual families that have experienced it?

    Someone once told me to pay close attention to micro data when talking about macro aggregates.

    • neal on March 18, 2014 at 10:40 ami

      “Are there any actual families that have experienced it?”

      Looking at that chart, I would say that the answer is “Yes, about 95% of the population have this experience, if not more. There must be some families affected …”.

      less snark answer is if you don’t know anyone affected, you are living in a comfortable bubble. Enjoy it while you can!

      • Ryan on March 18, 2014 at 10:56 ami

        Looking at medians over time is always very risky, as the good professors well know. The median does not necessarily tell anyone’s story. It’s entirely possible for a median to go flat or fall even if every person is making gains, if the composition changes.

        So “looking at that chart” is not a sufficient analysis to conclude that “95% of the population have this experience.” Your own anecdotal observations, complete with the silly judgment that I’m living in a bubble, is also not sufficient to make a claim about 95% of the population.

        I asked an honest empirical question. You answered it with snark and a truly terrible assumptions about how a median is constructed.

        • JakeS on March 18, 2014 at 4:22 pmi

          For the median to flatline while every individual household experiences increase, you would need either a “demographic transition” style age pyramid (which the US does not have), or a continuous massive change in household size distribution by age (there has been such a change, but not sustained throughout the period), or a massive rearrangement of remuneration away from Main Street (which there has been) or (incl.) from young to old demographics (which there has also been).

          But the question of whether the median tracks any individual trajectory is actually beside the point: That the median income does not track the average income means that there is increasing skew in the income distribution, and that is, irrespective of demographic developments, liable to cause a number of social pathologies.

          - Jake

        • Tyson on March 19, 2014 at 6:41 ami

          Your a Troll Ryan. Go silo yourself on ZeroHedge.

        • Marie on March 19, 2014 at 6:48 ami

          I know what you are driving at, and yes, here is a story for you. My spouse redirected from academia into business at the tail end of the 90s. Jobs were plentiful and he got a good one with generous bonuses and stock options (think consulting). He ascended and shifted into banking. Things went well until the crash. Then our income, which had gone from 60 to around 300 in less than a decade (not inc stock) crashed. No jobs. In free fall he grabbed a junior post at around 90, tiny bonus, no stock. Since then, raises are scarce and promotions too. He is career road kill, tainted by a shrinking pool and his willingness to work for less. The company’s once generous pension and retirement packages have been plundered by the top brass whose bonuses are obscene. So, yes, you can see it all in action, but who will speak of it publicly? To lose is to be a loser. And shreds of hope remain, right? The joys of middle age on the scrap heap.

          • FluffytheObeseCat on March 19, 2014 at 10:03 ami

            Thanks for your succinct and excellent description of the personal reality behind this graph. Particularly the last bit in re shame-silencing of the “losers”.

          • Barry "The Economy" Soetoro on March 20, 2014 at 1:59 pmi

            So let me get this straight. Your spouse happened to live the good life in a bubble, and when the bubble burst, you think he really deserved 300K and that the economy was the problem? Have you considered the fact he never produced 300K worth of value to begin with, and that income was not sustainable?

        • Michael on March 19, 2014 at 7:13 ami

          “It’s entirely possible for a median to go flat or fall even if every person is making gains” No that is mathematically impossible, the median can’t go down if all single elements rise. “silly” please…

          • Eric377 on March 19, 2014 at 9:52 ami

            If the population being measured remains identical it is impossible for every person to go up and the median to go down. But I am clearly seeing the phenomenon he describes at work on a population including retirements, deaths and new hires. My firm’s engineering department went through a cycle in the late 90s through most of the 00s without hiring much. Now there is increasing relative new hiring and retirements as a result. The skew of the salary structure is such that the median has come down relative to 2007 but everyone still working is pretty much is made more money in 2013 than in 2012. Doubt that it could work that way across the whole country and not sure it would be a good thing anyway if that were the explanation as probably a lot of formerly decently paid folks are now out of the labor market unwillingly.

          • tom on March 19, 2014 at 10:01 ami

            Yes it can, because yo uhave a shifting composition.

            10 workers aged 40-65 making 100,000 a year, 10 workers 18-44 making 20,000 a year. Median 60,000. Demographics shift and now you have 4 workers 40-65 making 110,000 and 16 18-44 making 25,000 and both groups are making more $ but the median has lowered.

        • Debt serf on March 19, 2014 at 7:30 ami

          I grew up in a lower middle class household. Not poor but just lower middle class. Over the years however things have deteriorated. Today my parents are barely at poverty level with social security income and live in a trailer. My sister is unemployed and her husband is on workers comp and they squat in a foreclosure. my brother is on SSDI. My other brother is an unemployed electrician. Me? I’m part of the professional class. I don’t make ‘a lot’ of income but I made more money last month than my brother receives from SSDI all year for his family of four! I have no doubts the income distribution is trending towards the top, with the top 1% taking the lions share. I’m just picking up the 1%’s crumbs but at the end of the year my household income is still more than the combined total of my entire former nuclear family. It’s totally screwed up. The middle class is disappearing as the lower and upper class expands. I jumped classes while everyone Kelsey in my family fell down a few notches. Welcome to America!

          • Up the Down Escalator on March 19, 2014 at 8:26 pmi

            I currently make 5 times the nominal money I was raised on, which with inflation is still ~150% more money. I chose a tech career that has high demand, though I certainly had no knowledge about the likelihood of it panning out when I went down that path in college. I was raised in what I would call upper lower class (if such a category exists) – disabled father and working-class mother who made the money, was on food stamps off and on, school lunch program, etc. My father, divorced, is now destitute and half on the street, sometimes in halfway houses. My mother is about to cross the 65 threshold, she went back to school 15 years ago while I was in school to finish a degree; the degree just barely kept her in the pool for a low-skill job since college is the new high school. She will never be able to retire because they are going to garnish her social security to pay off her student loans, which she can never afford to pay off. My spouse graduated in 2009 and, yeah, never got a job in her field like many others who graduated at that time. She has done a brilliant job working to recast her career in a consulting way, but gets very few hours of work. Given that there are few jobs in her profession and glut of trained people, she is staying home to raise our child which does afford us not needing to pay for childcare. However, I pay both of our student loans, leaving us with almost no disposable income at all . We are currently trying to buy a bottom of the barrel townhouse 1 hour commute from my job on the edge of the city we live in. I guess 250% of what you are raised on still isn’t enough. My close childhood friends are 50% above and below the median – some in construction, some blue-collar contracting, one went into accounting doing well (but also married a lawyer), one started his own business doing very well. I’m the only college grad of my close high school friends, I have some college friends who are drowning in student debt and will never recover. I’m pretty much on par with Debt Serf – one of the lucky ones with a “good” job, really just a cobbler in the feudal society of our New Gilded Age.

        • Longtooth on March 20, 2014 at 2:49 ami

          You don’t have to rely on the median family income to know that it’s not a compositional change. Just use Piketty & Saez’s income data over time (real income with and without capital gains). That shows that the lower 90% of income flat-lined beginning in 1973 (as did it flat-line for the top 10% as well), but that in 1983 the top 10% income gains increased took off, while the lower 90% income gains remained flat-lined to and including the present.

          Real Income Growth Rate, CAGR,% 1983 – 2012
          P0-90 0.157%
          P90-95 1.85%
          P95-99 2.56%
          P99-P99.5 2.957%

          Ref: Piketty & Saez Table 6

        • Longtooth on March 20, 2014 at 3:14 ami

          Also, BTW, there’s lots and lots of data on the composition of median family income gains if really wanted to know whether or how the median family income’s were constituted over time. The largest reason for median income gain was the two income earning family with females entering the workforce to a greater and greater proportion of the work-force (albeit at lower wage/salary rates). The other major change was due to the shift from a goods producing to services providing economy, with higher paying good producing jobs being replaced by lower paying services jobs.

          Also, on a national composite scale, the kinds of composition related demographic / earning income shifts you’re supposing might be the reason incomes increased when median family income didn’t is statistically remote to be generous.

          Besides, how does your supposition play into the fact that productivity growth rate remained constant while median family income basically flat-lined for the last 40 years? Somebody’s on the losing end of that deal aren’t they? Maybe you can enlighten us on who that might be and why?

    • amy on March 19, 2014 at 7:32 ami

      I wonder if there are other significant factors that are affecting family income?

    • Mike on March 19, 2014 at 4:32 pmi

      Check out a similar graph at
      Instead of a single curve this graph shows curves for 4 different percentile groups. While this does not eliminate data composition problems, it does mitigate them.

    • Shannon Brownlee on March 20, 2014 at 1:52 pmi

      There is one other place productivity gains are going (instead of wages): healthcare insurance premiums. Total benefits are tracking productivity gains more closely than wages, but workers don’t see those gains because it is being siphoned off into escalating healthcare costs. Yet, the healthcare sector itself is not seeing increasing productivity. Especially if we were to measure productivity in terms of better health.

  2. Bruce on March 18, 2014 at 10:57 ami

    Hypothesis: this is a retrospective view. Developing world wages are rapidly catching up to developed world, and as they do the median income curve will again track productivity.

    • Rich C on March 18, 2014 at 2:31 pmi

      That’s certainly an interesting hypothesis, but I’d point to a few reasons to be skeptical:

      1. The gap opens up in the early 1980s, when fully 1/3 of the world’s population, at least, was living under some form of state-socialism with very little exposure to trade. The timing works a lot better for the apparent increase in the gap circa 2000 (when China accedes to the WT0).

      2. My understanding is that while you are correct that wages in historically poor countries are closing the gap on wages in historically rich countries, it is also true that this “convergence” is happening more slowly than the convergence in productivity per hour. Now, maybe that’s not actually true, but I’ve seen analysis of wage-productivity gaps in a lot of developing countries (esp. China) and they are big and grew over the past decade and a half. I’ll see if I can find the source I’m thinking of, maybe at the University of Texas Inequality Project.

      3. Even if wage levels converge, the huge increase in the share of income going to owners and top managers implies, via Thomas Piketty, an accelerating level of wealth inequality going forward.

  3. Macroeconomist on March 18, 2014 at 11:26 ami

    Guys, really? I expected better from you. These figures are clearly not adjusted for the distinction between the GDP deflator and the CPI, which accounts for a huge fraction of the discrepancy.

    I’m a huge fan of your work in general, so this is disappointing…

    • Rich C on March 18, 2014 at 2:54 pmi

      Not sure why you say this. Back in 2007, Dean Baker published a paper in which he adjusted the wage productivity gap for depreciation, benefit costs, differences in the relevant deflators, and a few other things, and found that between 1973 and 2006 productivity grew 47.9%, while hourly comp grew only 20.1%. In essence, productivity growth was more than twice as fast as wage growth.

      Anyway, it looks like Mian & Sufi’s productivity measure increases by around 50% from 1980-2005 (hard to read off the graph with any precision), so they might very well be incorporating a variety of adjustments proper for this type of comparison, including depreciation and a uniform deflator.

    • joe on March 18, 2014 at 2:59 pmi

      First, there is no reason to use the CPI to adjust for output since you’re talking about a different basket of goods. Using the same deflators for income and output would give a misleading result.

      Second, the difference between the CPI and GDP deflator does not account for a “huge fraction”. If you look at 1945 to 80, and compound the growth and then compare the GDP deflator to the CPI you get the CPI growing 4.6% more. Then do the same for 1980 to 2012 where median income stagnates, you get CPI growing 17% more.

      So even if it was appropriate to deflate output using the CPI (which it is NOT), it would only account for a small fraction of the discrepancy.

      US 1945 to 1980
      Consumer Price Index 4.44% (1.044^35 = 4.51)
      GDP Deflator 4.27% (1.0427^35 = 4.32)

      US 1980 to 2012
      Consumer Price Index 3.25% (1.0325^32 =2.78)
      GDP Deflator 2.72% (1.0272^32= 2.36)

      4.52/4.32= 1.046(compounded over 35 years the CPI is 4.6% larger than the GDP deflator during 45 to 80).

      2.78/2.36 = 1.17 (compounded over 32 years the CPI is 17% larger than the GDP deflator during 45 to 80).

      • Matteo on March 20, 2014 at 12:03 ami

        The issue is not that CPI should be used to deflate output, but that the wage share measure has not been controlled for the ratio between the CPI and the value of output. I guess it depends on what one is investigating. For primary distribution, the appropriate concept is product wages, as one is tracking the contribution of labor to total value added. CPI matters if one wants to see whether subsequent changes in relative prices in the economy as a whole allowed wage earners to gain as consumers what they might have lost on the shopfloor.

  4. Jerry on March 18, 2014 at 11:31 ami

    How does the decline in family size impact the data conclusions above?

  5. Maxwell Strachan on March 18, 2014 at 11:37 ami

    I love this blog. Would you be interested on crossposting it on HuffPost Business. You have my email if so. Thanks!

  6. Victor on March 18, 2014 at 11:38 ami

    I’m a little fuzzy on that Y axis. How could productivity per hour and median family income be on the same graph. So let just say Productivity = 150 and Family income = 100, is the graph saying that in one hour our economy produces 150% more capital than the median family makes in a year?

    • Rich C on March 18, 2014 at 2:17 pmi

      No, the graph is reporting each time series as an index, with 1980 as the base year (that’s why it says “1980=100 over by the vertical axis). So, what it is saying is that between 1980 and 2012 output per worker hour (that’s productivity) doubled, while the median family income increased maybe 10% or 15%.

  7. Anon on March 18, 2014 at 11:47 ami

    It’s well known that an increasing share of total compensation for many households has gone towards employer-provided/sponsored health insurance over the past couple/few decades. So what does this graph look like when real median family total compensation is used instead of real median family income?

    • Rich C on March 18, 2014 at 2:22 pmi

      FYI, EPI and many others have plotted similar graphs with median compensation (including employer provided health and retirement benefits)rather than median family income or median hourly wages…. the effect is to push the beginning of a clear, long-term gap between productivity and what-workers-get-paid forward a few years. So instead of the wage-productivity gap clearly opening up around 1980, it opens up around 1983. Overall, you see the pattern: from the late 1940s to the early 1980s, both grow in line, from the early 1980s on productivity keeps growing while the “wage” measure more or less stagnates.

    • JakeS on March 19, 2014 at 6:20 ami

      That question is tangential to the point being made so long as the health insurance sector continues to be a giant rent-seeking scam.

      It does not change the fact that employee remuneration is being sucked up by rent-seeking bottom-feeders. It just changes the identity of the bottom-feeders (and even then not by that much – it’s still the FIRE sector…).

      - Jake

    • Michael on March 20, 2014 at 6:48 pmi

      The CBO study looks at everything, including transfer payments and taxes.

      “CBO finds that, between 1979 and 2007, income grew by:

      275 percent for the top 1 percent of households,
      65 percent for the next 19 percent,
      Just under 40 percent for the next 60 percent, and
      18 percent for the bottom 20 percent.”

  8. Anonymous on March 18, 2014 at 12:02 pmi

    I think we’ve seen a lot of similar metrics. The real question is what can be done about it? How much of a leveling effect is a result of technology lowering the value of a lot of positions? For the record I am in an entry level position.

  9. Bryce Carlson on March 18, 2014 at 12:02 pmi

    I think we’ve seen a lot of similar metrics. The real question is what can be done about it? How much of a leveling effect is a result of technology lowering the value of a lot of positions? For the record I am in an entry level position.

  10. sonal sodhi on March 18, 2014 at 12:37 pmi

    is the median real family income adjusted for labor input i.e. for number of hours worked, increase in female labor force participation in 80s/90s, as prod. is real gdp per hour but real median family income doesn’t seem to be adjusted for labor input….thanks…

  11. Brenb on March 18, 2014 at 12:46 pmi

    If we are looking at median income, shouldn’t we be looking at median rather than average productivity?

    • JakeS on March 18, 2014 at 4:11 pmi

      You would need to disaggregate productivity down to individual worker resolution in order to compute the median productivity. I am not aware of any way to measure individual productivity. Indeed I doubt that it is possible to do so except in the vagues, most qualitative manner, since the typical value chain is made up of a number of incommensurate and individually indispensable links.

      The typically used proxy for productivity is remuneration. But not only is it a rubbish proxy (by that measure Charles Ponzi was among the most productive men alive in 1929), it is clearly unsuitable for studying whether remuneration is tracking productivity: It affirms the null hypothesis by construction.

      - Jake

    • Larry Signor on March 18, 2014 at 4:20 pmi

      How, exactly would you define “median productivity”? One would assume the authors are looking at productivity/hour of labor and compensation/hour of labor. I see no problem with the equivalencies. The productivity function has continued it ascent and the compensation function stalled in 1980. Pretty straight forward.

  12. Sandwichman on March 18, 2014 at 1:25 pmi

    There’s an even more important economic chart — after all “productivity” is an abstraction. What is the relationship, then, between income, productivity and resource (especially energy) consumption?

  13. FRED FREEZAGLI on March 18, 2014 at 1:28 pmi

    Two major problems with this chart:

    1. It does not talks about total compensation, that covers health ins and other benefits.Richard Burkhauser from cornell talks about it in his research:

    2.It does not talks about the productivity of the median family but on the average productivity per hour of all americans combined. meaning, the hi-tech programmer and the dude who work at McDonald’s.

  14. Jim Moore on March 18, 2014 at 1:41 pmi

    Would you mind citing the source for the productivity data? Thanks.

  15. George N. Wells, CPIM on March 18, 2014 at 4:37 pmi

    Looking at the chart, and knowing the history there really isn’t anything unexpected. From 1945 through the early 1960′s the top marginal rates were confiscatory. This, in my opinion, served to assure that profits were invested in research, plant, equipment and labor. Then the top rates were lowered (by elimination of top brackets) and the gap starts to form but the top rate is still high. Then comes the 1980′s and the Reagan tax plan and now handing the profits to the executives makes more economic sense and the rest is starkly shown by the graph. Companies have shut down research, refused to build domestic plant and laid off labor. Many companies now make more money from their internal hedge-funds than from their operations departments. The real question becomes: is this sustainable?

  16. Craig Jackson on March 18, 2014 at 9:15 pmi

    Could this difference be attributed to the fact that the productivity of the service sector, where most of US workers work, can claim much less productivity gains, while the manufacturing sector continues to garner outsized productivity gains through the use of technology and low priced foreign labor? Make a similar chart using both manufacturing output per hour and the business sector output per hour. You see clearly that manufacturing output per hour has increased at a much higher rate. The service sector was supposed to save the day, but all it does is lower the risk somewhat and reduces the gains.

  17. Paul Orwin on March 18, 2014 at 11:00 pmi

    I don’t usually get into snarkfests, but maybe y’all should learn how to read a graph before criticizing it? it is obvious that both data sets are normalized (thats what 1980=100 means). there is no perfect way to aggregate income (or productivity) so you can always nitpick around the edges, but the big picture story is as plain as the nose on my face – the gains in output were basically spread across the income spectrum from the end of ww2 until 1980, and then the labor share stopped (or at least slowed down). stop trying to pick apart the graph, and deal with the reality that it is displaying.

  18. JasonM on March 19, 2014 at 12:19 ami

    I see the effect of this graph in my work everyday. I am a producer on a TV show. We have many creative people (writers, producers, editors) and many administrative people. The administrative workers’ wages have stagnated, and in many cases are startlingly low. Creative employees whose work is seen as more skilled and harder to replace have wages that are 3-4 times that of a clerical worker. Meanwhile the clerical workers continue to get increasingly efficient and productive due to technology. One person can do the job of two people every 3-4 years. This season our TV show has eliminated 3 positions. So there’s a huge downward pressure on wages.

  19. […] If you must know only one fact about the U.S. economy, it should be this chart: (HouseofDebt) […]

  20. […] the full post here.  And we'll keep an eye on House of Debt for ongoing […]

  21. Debt serf on March 19, 2014 at 7:35 ami

    Too bad so many of you are missing the forest for the trees. The data is in front of you, the stories of pain and downward class mobility and stagnant incomes are right under your nose. The server who gives you your food, the window washer in your office building, the cleaning crew in your suite, the mechanic who changes your oil, the cashier who rings up your groceries. Go take a look at how they live and you’ll see the data reflected in real life.

  22. McMike on March 19, 2014 at 8:07 ami

    The fact is, just about any thing you can measure got worse starting around the time Regan took office.

    One heck of a coincidence.

  23. McMike on March 19, 2014 at 8:08 ami

    er, Reagan.

  24. Fernando Rizzo on March 19, 2014 at 8:29 ami

    You can look at this for a comparison of the productivity vs hourly wage growth.

    Or at this…
    The compensation-productivity gap: a visual essay, Susan Fleck, John Glaser, and Shawn Sprague
    Available at

  25. Ben on March 19, 2014 at 9:09 ami

    A few more papers on the issue for those interested:

    Pessoa and Van Reenen have a good analysis of this trend in the US and the UK that came out recently:

    My think tank (ITIF) also put out a paper in 2007 that looks at the issue, with analysis of the effects of household composition:

  26. Robert on March 19, 2014 at 9:14 ami

    I do not think this is a good chart. There are too many problems with the data. Anyone who wants to talk about decoupling of wages and productivity needs to read the paper by Pessoa at the LSE. Then come and make an argument.

  27. Lance Paddock on March 19, 2014 at 9:30 ami

    A few observations:

    1. While I believe the graph illustrates a real issue, it is not nitpicking to ask for any such claims to include all forms of compensation, account for changes in family composition, etc. The increase in the number of households relative to number of workers could explain a significant portion of these results. It may not change the overall point, but it certainly could change our perception of the size of the issue and change our opinion about how policy should react to it.

    2. The discussion of the gap opening in 1980 is just an example of political bias coloring the reading of the graph. While Reagan and his policies might be responsible for much of this issue, the graph itself does not support such a statement. Sure the gap opens then, but the trend which can be observed by when the slope of family income changes occurs much earlier, around 1968. It just so happens that the income line was well above the productivity line at that point and that positive gap wasn’t entirely eliminated until about 1980, but that trend was established in the late 1960′s. As an aside and only intended as speculation prior to actually examining the data, anecdotally that is around the time when family composition began to meaningfully change. This supports the idea that about that time family composition began to affect this aggregate.

    3. My guess is that the issue is a lot more complex than a simple issue of Reagan’s policies, that some of it (maybe even most?) is explained by changes in compensation and family composition. None of that means we shouldn’t be concerned about the increasing take of the 1% to use a cliche’, but we should use a more granular and robust analysis and avoid simple political readings that show more about our biases than about what the data says.

  28. tom on March 19, 2014 at 10:07 ami

    I have to say that this chart isn’t a fact, it could easily be very misleading.

    1. It compares real hourly production to average household earnings. Average hourly production could lead to higher hourly pay which then leads to fewer hours worked which leads to “stagnating” incomes when in reality consumption of leisure has increased. This is easily fixed using hourly wages vs hourly production. You would dock points from an econ 101 student for making this comparison compared to a much more straight forward comp.

    2. Is “real income” compensation or dollars earned? Obviously one only tells a portion of the story.

    3. Does “real income” include transfers?

    • Richard on March 19, 2014 at 12:34 pmi

      Tom, have you paid any attention to recent American history?

      1. So you tell me. Has real median hourly pay increased in recent decades? Have hours worked decreased in recent decades?

      2. There is a difference, but as people above here have noted, it’s not enough to wipe out the gap.

      3. Considering that the vast majority of transfers are SS and Medicare, I’m not sure why this matters. To working-age Americans, their productivity has increased while their wages have not. Try to find out per capita non-SS/non-Medicare transfers if you like. I’m quite certain they’re not big enough to matter.

  29. Jesse on March 19, 2014 at 10:31 ami

    Until median income recovers, there will be no sustainable economic recovery.

    One may argue the causes, and I think that ignoring offshoring and the redistributionist tendencies of a corrupt financial system among them, but the fact remains that without changing something, the ‘new normal’ will grow increasingly unstable.

  30. Daublin on March 19, 2014 at 10:36 ami

    While I am sympathetic to the issues being raised, this graph has a number of problems. It ignores medical insurance as compensation, and it plots *family* income. Family size has decreased over the last few decades due to an increasing divorce rate; as such, family income has a built-in decrease just due to higher divorce.

    The graph also compares a median to an average.

    The story of the last few decades is that technically capable people have individual productivity that has skyrocketed. Historically these people were maybe 10x more productive than a median worker, and now they are maybe 100x times more productive than a median worker. These super workers also earn more income.

    When you plot a median, changes in the upper parts of the data are not reflected. As a result, the line for median income is unaffected by super-workers. On the other hand, when you plot an average, the increased earnings of super-workers do indeed show up in the graph.

  31. David Ayer on March 19, 2014 at 10:46 ami

    1. The income rolloff actually begins as the “Great Society” programs of the mid-1960′s kicked in.

    2. Income and compensation are not the same thing – an example being corporate health care plans.

    3. If productivity has skyrocketed then obviously goods are either much better using constant dollars or better goods are much cheaper. And goods such as computers and communications gear are both.

    - David Ayer,

    • David Ayer on March 19, 2014 at 11:33 ami

      “better goods are much cheaper” should have been “equivalent goods are much cheaper”

  32. McMike on March 19, 2014 at 11:47 ami

    Of course one cannot blame Reagan solely or entirely.

    But it remains reasonable to assert that just about any measure we can think of started to diverge, went parabolic, took a southern dip, inverted, reversed course, fell off a cliff, or otherwise appeared to go to hell within proximate association chronologically to the Reagan era. Reagan’s administration is, like it or not, a major inflection point economically for nearly every single measure.

    In other words: that is an ex-parrot.

    And we can generally add, the trends continued on that vector thereafter.

  33. McMike on March 19, 2014 at 12:02 pmi

    Re: “super workers”. Are you referring to Wall Street traders and other members of the top 5% or so?

    Otherwise, your thesis doesn’t hold. The second 20% has not really seen appreciable income growth. The middle certainly not. So are you arguing that all that super productivity is coming from the top 5%?

  34. Cocoabean on March 19, 2014 at 12:40 pmi

    Good to hear you admit that “the highest-paid workers are getting a larger share of income”.

    It’s no specifically “higher-paid workers”: it’s any worker whose job is guaranteed or protected from competition by government. Or which benefits from government-granted monopoly access to a resource. Natural resource extraction: oil, gas, forestry, fishing. Government employees of all kinds. And the ancillary industries and companies that supply these sectors, plus their pensioners.

    They’re all doing very well here: they buy the new F150 pickups, the motor homes and the Netflix subscriptions. They have the smartphones and bills for every member of the family. And the $400,000 mortgages, the multiplicity of credit cards, the boats and the trophy wives…they eat out frequently.

    The “new rich” are not all banksters: they’re often your neighbours. Thank you, government.

    • Richard on March 19, 2014 at 2:28 pmi

      Ah, the deluded libertarian.

      Here’s the thing, though: CEO’s in virtually every industry have seen their compensation increase several times more than the median worker’s in that industry. So are you saying that every industry is protected by the gubmint?

  35. Jim Shannon on March 19, 2014 at 12:48 pmi

    You are all missing the cause!
    The Governments manipulation of the TAX CODE to benefit the CentaMillionaire$ and Billionaire$ has given the 1% enormous incentive to squeeze evey last nickle out of the consumer.
    The consumer -70% of GNP – has been consumed by the greed of the “Taker Class”!
    A WakeTFUp and realize it is all divide and conquer politics to We The People keep on ignoring the real cause of the demise of the 99%!
    It’s the TAX CODE and the RICH are RICH because the CODE is the GRAND DECIDER!
    There is no “FREE MARKET” for the working poor and that’s a clear fact the RICH understand!

  36. Rodger Malcolm Mitchell on March 19, 2014 at 1:48 pmi

    Lest anyone doubt the premise of this post, merely look at:

  37. Paul on March 19, 2014 at 2:32 pmi

    There is a very simple explanation why productivity (per hour worked) has increased more than median Family Income, AND WHY THIS SHOULD BE THE CASE.

    In the last 30 – 50 years, in North America and other developed countries, automation has caused a replacement of workers with machines (robots and other mechanised means in the production of goods). The average worker (other than knowledge workers, had nothing to do with this, other than that more and more menial, repetitive, boring and dangerous jobs have been replaced with machines. Obvious that output (productivity) per worker has increased under these circumstances. Who can we thank for this increase in automation. First of all of course the people who invented these new means of production, and secondly, the people who put up the capital to build these new means. The result: compensation of knowledge workers increased, as did the remuneration of capital providers. The untrained are sadly not in a position to claim the benefits of this change in productivity. Lesson learned: to increase your chances of success you will need to study and train. Finding a job flipping hamburgers out of highschool is not going to cut it any more. Sad but true, and just the way it happens to be.

  38. Rodger on March 19, 2014 at 2:51 pmi

    From the graph, the median income slope dramatically changed in the early half of the 1970′s. Why ? What were the main economic events of that time ? I bet if you researched the growth of statutes, rules, and regulations at that time impacting businesses, you will find the slope of the graph also started upward in a more steep fashion than say in the 50-60′s. As I remember, we entered the age of the regulatory authority, with the pages of laws, rules, regulations impacting private enterprise changing significantly from its historic norm.

    So if the regulatory burden started a massive increase, could it be that it increased the expense load on a business that then caused the business to stop passing on the benefits of improved productivity to its workers because the business had to protect its bottom line or else it couldn’t invest for its future, so in order to feed the government regulatory monster, businesses had to give directly or indirectly more and more of their economic value add to their new partner, the government.

    Pls consider adding to this graph the total pages of federal and state laws, statutes, rules, and regulations, and also simply the total expenditures by all government entities. I suspect it will become pretty clear where the increased economic value add went…to the governments, local, state, and federal !

    • McMike on March 19, 2014 at 6:07 pmi

      So why doesn’t the slope bend back when the conservatives took over and posed with chainsaws over stacks of federal regulation?

  39. Chris on March 19, 2014 at 6:28 pmi

    I think Mian & Sufi should read Brynjolfsson & McAfee. Chapter 9 of their latest book. They point out that, “Most of the growth in productivity directly translated into comparable growth in average income.” The divergence between median and average incomes began in the mid-70s. And the cause has been the increasing use of technology.

    Technology has been increasingly employed as a substitute for less skilled, routine workers. Those who design, program, maintain, and operate that technology, or who use technology to augment their labor in various other ways, have become more in demand. Those with greater skills, especially cognitive skills and skills in information technology, are more productive due to skill-biased technical change.

    It wasn’t Reagan that was responsible for this, but in large part the personal computer revolution that began in the early ‘80s.

    Very early in my career (in 1993) I was working as a financial analyst for a very small consumer finance company. After less than a year there I obtained a 20%+ raise in my pay after introducing into the finance department a couple of new information technology innovations: Microsoft Excel for modeling the budgets of the owner’s joint companies, and a “banded” database report writer that could produce reports for the finance/accounting staff that was previously dependent on relatively unresponsive programmers in the IT department. I’ve seen the same demand for technical/analytical skills in one form or another at every place I’ve ever worked over the last few decades, including now at my own small startup company.

    Figure 9.2 in Brynjolfsson & McAfee shows the wages for full-time, full-year male U.S. workers from 1963-2008. Those with graduate degrees increased the most, college graduates the next most, and so on, while the wages of those who dropped out of high school actually decreased. I think it’s safe to assume that in an economy driven by skill biased technical change, the average incomes of each group closely tracks the productivity of each, just as growth in overall average income tracks the growth in average productivity.

  40. Marcello Pavan on March 19, 2014 at 7:19 pmi

    There are much versions of the same graph.

    1. indexing 1980=100 is not a good choice. If you look at the graph, you see the growth in P and W track each other in lock step, then deviate around 1972 or so. So it has nothing to do with Reagan.

    2. Other versions show ‘average’ as well as ‘median’. (you can use google as well as I can) of course due to inequality, the ‘average’ curve slopes a little higher, but not by much, certainly nowhere near enough to alter the conclusions.

    3. Family composition has changed in one major way – now most families have 2 wage earners, due to the influx of women into the workforce starting in the 70s. That makes the implications even worse, since now it takes MORE people per household to make essentially the same money as 40 years ago.
    number of people/family hasn’t really changed a ton.

    4. again, other versions of the same graph show the median, as well as the 20th, 40th, 60th, 70th, etc percentiles, where below 50 shows the curve sloping _down_ and above 50 shows the curves sloping ever more upward. Showing clearly that the benefits of the aggregate productivity is NOT going at all to ~40% of the population, only barely to the middle-to-upper-middle classes, and predominantly to the highest income tiers.

    if you think its those highest income tiers are generating all the productivity, your ideology is too far gone to have a sane discussion…

    • Daublin on March 20, 2014 at 4:29 pmi

      Marcello, I would be interested in a citation for average income. From what I understand, average income is roughly the same as average productivity, so the two values are closely tied in any given year.

      The reason for this is because gross productivity is defined as the amount of wealth that is created in a given year. Average productivity is simply gross productivity divided by the number of workers. Meanwhile, average income is simply total income divided by the number of workers. However, where does total income come from? It’s the amount that all that produced value is sold for.

  41. Lucais Sewell on March 19, 2014 at 9:23 pmi

    Strangely, there appears to be no word in English for Verteilungsspielraum – “distributional leeway” – i.e. the amount by which productivity increases justify increased wages. How can it be that this basic term in German economics has no English equivalent? Is anyone aware of an English term?

  42. Jay on March 19, 2014 at 10:01 pmi

    I note the red line is Median FAMILY Real Income, my emphasis on “Family.” The mid 70′s is when law changes led to the beginning of increasing spousal entry into the workplace. So a variant that I would find useful is the how median hours worked by both spouses changed over the last forty years. If the hours trended up, then the reality is even more extreme than the chart indicates.

    • Jay on March 19, 2014 at 10:04 pmi

      Just noted that Marcello Pavan made my point somewhat elequently a few hours before mine.

    • Daublin on March 20, 2014 at 4:35 pmi

      I can’t speak to the law, but actual female employment has been increasingly steadily since the 40s, with no particular spike in the 70s.

      • Jay on March 21, 2014 at 11:30 ami

        Haven’t looked at that data, but if true I’d wager that female employment prior to the Equal Credit Opportunity Act (1975) was primarily part time and for “pin money.”
        That Act required lenders to include the income of both spouses when considering credit qualification. Prior to that lenders discounted spousal income on the assumption that it’s continuity was unreliable. This opened the door to two-income families being able to “afford” (i.e. borrow more) a better home, which ultimately meant most families had to have two incomes to compete.

  43. Barry "The Economy" Soetoro on March 20, 2014 at 1:56 pmi

    71 one replies and not one person notes the obvious? Ending Bretton Woods was the biggest mistake of the last 50 years.

  44. James Sherk on March 20, 2014 at 2:48 pmi

    If you compare average hourly compensation with average hourly productivity, using the same measure of inflation for both, that gap essentially vanishes. Workers pay has risen in line with their productivity, and they are in fact enjoying the fruits of their labor:

    Lesson: Apples to Apples comparisons are important.

    • Jay on March 20, 2014 at 4:06 pmi

      Average, smaverage.

      On average Bill Gates and I are quite wealthy. While median has it’s limitations, it’s a lot more informative than average.

      At least that’s the case if you care about “real” persons more than “paper” persons, ie corporations.

    • Richard on March 20, 2014 at 4:45 pmi

      A flatlining in median family income is still important. It means that a rising tide has not lifted all boats (which also means it has lifted some boats tremendously) which means rising inequality.

  45. Simon on March 22, 2014 at 8:22 ami

    Especially the last abstract sounds like you would agree with this analysis here, arguing that inequality, debt-led consumption and the financial crisis are interlinked:

    Maybe also with this extended version also discussing the Global Imbalances, and developments in China, and Germany:

    Would you?