Comments Guidelines

All comments are pre-moderated. No spam, slurs, personal attacks, or foul language will be allowed.

Monday, March 12, 2012

Basics: Real Wages Remain Below Their Peak for 39th Straight Year

The release last month of the Economic Report of the President has elicited a great deal of commentary, but none that I have seen touches on what I consider the best measure of long-term income trends, real weekly wages of production and non-supervisory workers, which is contained in Appendix Table B-47, "Hours and earnings in private non-agricultural industries, 1965-2011." According to a Bureau of Labor Statistics staffer I spoke to some years ago (so the percentages may have changed slightly), this covers 62% of the entire workforce and 80% of the non-government workforce. This lets us focus on average workers and excludes what is happening to high-salary workers. Using weekly rather than hourly real wages takes out the impact of varying hours worked per week over the years. The table below extracts from B-47 to reduce its size. The inflation is adjusted using 1982-84 dollars as its base.

Year          Weekly Earnings (1982-84 dollars)
1972          $341.83 (peak)
1975          $314.75
1980          $290.86
1985          $285.34
1990          $271.12
1992          $266.46 (lowest point; 22% below peak)
1995          $267.07
2000          $284.79
2005          $284.99
2010          $297.67
2011          $294.78 (still 14% below peak)

Thus, we have 39 straight years where real wages have yet to get back to their 1972 peak and, indeed, they are a long way from that peak still. This is doubly surprising when we consider that productivity has been increasing steadily throughout that period, approximately doubling from 1970 to 2011, as shown by the Federal Reserve Bank of St. Louis' data:

FRED Graph

Due to the convergence of rising productivity with falling wages, we should not be surprised to see that labor's share of non-farm income has fallen:

FRED Graph

There are other ways to track the income of average workers. The Economic Report of the President highlights median household earnings in its Figure 1-1, but these data are affected by changes in the number of incomes per household, primarily the result of increased women's labor force participation. The bottom line is that the increase in incomes per household obscures the fall in individual income for most workers.

Some conservative economists, such as Martin Feldstein, have argued that we should instead use real compensation instead of real wages and use the same inflation adjustment ("price deflator") when analyzing trends in compensation and productivity. There are multiple problems with his analysis. First, he claims that the growth of compensation (1.7% per year) is almost as high as the growth of productivity (1.9% per year) over the 1970-2006 period, but ignores the power of compounding. The 0.2 percentage point difference may not sound like much, but over 36 years productivity grew by 96.9% (1.9^36) and compensation by only 83.5% (1.7%^36) using his preferred measure (and I can't comment on its correctness), so workers' compensation should still have grown by substantially more than it did.

An even bigger problem is that the compensation series includes all workers, not just average workers, so it lumps CEOs and janitors together in a single measure even though we know that the 1% soaked up most of the income gains before and after the Great Recession. Moreover, the nonwage portions of compensation, such as health insurance and defined benefit pensions, have eroded much more for average workers than for the 1%. Thus, this measure is completely unsatisfactory for understanding what has happened to the average worker.

So, we come back to Table B-47 and the real wages of production and non-supervisory workers. The fact that, adjusted for inflation, wages still remain almost 14% below what they were 40 years ago, despite a doubling in productivity, is a national disgrace. It is one of the roots of the increase in multi-income households, in higher levels of indebtedness needed to maintain consumption levels, and of the sharp increase in inequality we have seen over recent decades.


  1. There must be something in the air. Via Kevin Drum (, I see that Lane Kenworthy has posted a similar finding with regard to household income:

    Worth reading!

  2. I provided the strategic vision and rhetoric back in '95 to the Clinton White House via Dep. Chief of Staff Harold Ickes for a campaign called WORK=FAIR which focused on this divergence between productivity output and real earnings decline and Ickes liked it but it gained no traction for years, even though Sweeney at AFL-CIO and many other big politicos had a look at it. The main problem with capitalizing (pun intended) on this data for progressive political gains has been bad messaging and the Democratic Party has mostly Bob Schrum to blame for that. Gore came close in 2000 to using the correct rhetoric but kept it too general "people vs. the powerful" and Obama has let the fraudster bankers off-the-hook so far. The key thing about this data is it shows what I call a REWARD GAP - hard work has not meant fair rewards for nearly 40 years and that has fractured what it means to be an American. Here is a link to the Wayback Machine's WORK=FAIR page:

    1. Thanks for the link. A "reward gap" sounds like a good term. What other rhetorical strategies do you think would be appropriate today?

      Please feel free to email me.

  3. Hardwork almost always makes someone else rich. Fortunately for the powerful, it's easy to exploit the human tendency toward guilt and fear. I guess the elite realized they don't need a very large number of petit bourgeois to maintain their status.

  4. "Using weekly rather than hourly real wages takes out the impact of varying hours worked per week over the years."

    But this may understate how bad it's been, as it seems that the number of hours most people work per week (at least male full time) has increased substantially over the last 40 years.

    Can you get per hour figures?

    1. In fact, real hourly earnings only fell 5.3% between 1972 and 2011, from $9.26/hr. to $8.77/hr. Most of the fall came from hours worked per week, which fell from 36.9 hours in 1972 to 33.6 hours in 2011, a decline of 9%.

      I think you're right that a lot of companies added hours on to their full-time workers so they could keep from hiring more full-time workers with benefits, but it appears to have been more than offset by shorter hours for other workers.

    2. It's very hard to get a non-misleading figure on the hours worked of a typical full time worker. I think a lot of the statistics mislead by putting the part-time workers in with the full-time.

      You need more specific time series of numbers, like college educated full-time workers. I bet they work a lot more hours today than in 1972, especially if you count the work they do on their computers when they're supposedly at home engaging in leisure.

    3. The average workweek has fallen from some 38 hours in 1965 to 33 hours in 2016. It stems largely from the changing composition of employment. In the good sector the average workweek has remained over 40 hours. But in services the average workweek is now 32.4 hours versus some 37 hours in 1965. services share of employment has risen sharply.

  5. The gap between wages and productivity is a reflection of power relations between the mass of workers and capital. I'm sorry if the language sounds antediluvian, but so many people who write about the issue struggle so hard to come up with economic explanations for an obviously political problem that I feel duty bound to keep them from straining themselves.

    We entered a new historical regime in the 70s, not simply because the ideological appeal of Communism evaporated but because there were no other non-capitalist power bases on hand, at least in Europe and the U.S. We're in a time when there is no effective check on the tendency of oligarchs to increase their relative wealth. That's a historical novelty. Previously, capitalism was disciplined by landed wealth, military aristocracies, nationalism, and religion as well as by the need to buy off or fend off the revolt of the underlings. Why have the one percent taken an increasing share of the pie? You might as well ask why a dog licks his balls.

    1. This has an aura of plausibility about it, but so too does an alternative explanation that has been aired, that the difference is technological prowess. In other words, the fastest and best adapters to (and creators of) the technologies of the last few decades have reaped the gains, and the less adapted have level-pegged. It seems to me that to choose between these hypotheses (and any others that may be mooted), one needs evidence. I don't have evidence of the latter (other than anecdote about technology tycoons and workers), but do you have evidence for your power relation thesis being in fact the cause? Without this, it's just the usual political rhetoric (on all sides), useful only to bludgeon each other with. The advantage of Mr. Thomas' posting is the evidence given (and the supporting and counter evidence in the replies).

  6. I've been collecting the charts and analyses of others on this subject for 3 years here: No matter which data set you look at the story of the middle class stagnation and the gilding of the top 1% is the same.

  7. Excellent! Someone else sees the importance of this metric! I did something on this a while ago at
    [And read Mason Gaffney's addition as to why the increases in recent years are apparent only!]

  8. Hello to all. I noticed this blog and think it worthwile.
    I am an old economist, teacher for many years, and also an expert witness in cases of economic loss in the courts.
    One of our tasks is to establish a growth rate of income for injured parties so that a present value of their future income can be calculated.
    The numbers that you cite here have long been part of my thought process mostly for the reasons you identify. I used these numbers from table B-47 for many years.
    In more recent years, the profession of experts in the courts has generally shifted to growth rates for income which are supposed to be adjusted for increasing benefits to workers. These numbers have yielded a higher growth rate for income.
    I am looking at a recent report done by a highly respected economist which identifies a growth rate of .59% over the last 10 years. This data comes from the Bureau of Labor Statistics series entitled "Private industry. Total Compensation all workers"
    Of course that's only one half of a percent , but better than a big drop. It seems to me that whatever your objections to the admittedly tricky adjustments to your B-47 wage series, one cannot omit benefits as a part of wage compensation and be accurate.
    A second major concern I have with your analysis is the assumption that productivity increases of US workers should be paid out to those same workers as higher compensation.
    In a globalized world, swimming in low wage workers, it may well be that the only reason US workers can keep their jobs is the long term productivity gain.
    Thanks for your good work and this useful discussion. I look forward to your thoughts.

    1. Thanks for the comment. As I said in my post, "Total Compensation all workers" lumps CEOs and janitors in together, when we have good reason to think that CEOs got bigger wage increases and did a lot better on benefits than average workers did.

      But you are right that the ideal would be to look at the same people as in Table B-47, but with their benefits added in. I recall the Economic Policy Institute putting together some data that attempted to do this. I will try to track it down and link it.

      Regarding your point about productivity, in a closed economy we would certainly expect productivity gains to be translated into wage increases. Questioning that begins to move us into the realm of explaining why we see this pattern of wage decline, i.e. I think globalization is certainly part of the explanation, though economists have argued quite a bit over just how much of the change globalization accounts for.

  9. interestingly, if one matches those dates with presidential terms, one finds a very strong correlation with:

    [1] weekly earnings ALWAYS DROPPED during republican presidents [except Bush#2 an increase in 0.20=20 cents]

    [2] weekly earning USUALLY INCREASED during democrat presidents

    ... so does the data imply that if the the middle class wants an increase in weekly earnings that they should vote democrat ?

    ... as an aside,
    based on CIA data, the USA "GINI" index is WORSE than:

    Iran, Cameroon, Uganda, Nigeria, Guyana, Russia, Senegal

    on the bright side, USA GINI is slightly better than Rwanda

  10. I did something quick on this using real hourly wages.

    1. Thanks for the link. It corresponds with my intuitions as well, in particular that AFDC/TANF levels should affect wage rates.

      I'm curious why you logged the wage rates, since they don't go to a very high level. On first glance, that seems unnecessary to me. Maybe there's a technical point I'm missing?

  11. Hello Kenneth,
    I found a really useful link which I believe you will want to see. It certainly influences our discussion of real wages and what can be expected from the US or other econ systems. It is called non-excludable observations and it is the report on income inequality.
    This data certainly support your suspicion that we ought to take the highest paid workers out of the equation when we discuss worker wages in general.
    My email address is . I have some trouble with new fangled directions so I am the economist who filed anonymous because I couldn't get the machine to work any other way.
    In any event we seem to have a great deal of interest in common concerning political economy and I hope for a mutually rewarding discussion over time.

  12. Kenneth, for those of us employed during both sides of the post-war 'long wave' [and the few who spent decades studying political economy [not neoclassic econ], I can guarantee the decline in real wages was perfectly evident no later than the mid-1980s - the question was of causality which, IMO, has had to do with the relatively strong nonfinancial Rate of profit decline post-1968-70 and capital's attempt to overcome this. [Please note that I'm not referring to mass of profit or share but mass relative to total capital. Additionally, Fred Moseley has pointed to the rise in unproductive labor].

    Thank You


  13. I feel sorry for teens starting out now grown adults will want there pay..

  14. Kenneth - two tours in the armed forces and 20+ years public sector was disastrous for my income. Despite common mis-perception, the security of a steady paycheck only means public service is bounded by a fixed income, slow to respond to economic realities of the free market system. In almost 45 years in the job market I have not seen the disposable income situation for lower and middle class Americans in a more inequitable state.

  15. Almost any of the BLS hourly and weekly wage series are going to be distorted by the changing composition of employment.

    Maybe the best series to use is the employment cost index for wage and salary workers. I think of this as what it cost employers to pay a worker in the same job over time. It is a nominal measure, but it would be easy to deflate it.

    Let me give one example of how the data is impacted. In general individuals wages grow as they age --on average a 50 year old makes more than a 30 year old. From the 1960's to early 2000s the baby boomers moving through the age spectrum should have generated an upward bis to the average hourly earnings data. So the average hourly earnings data was biased upward. Census publishes average wage data by age that shows what I mean. Now, as the baby boomers age out of the labor force they are being replaced by younger workers so the average age of the workforce is now falling and this will biases the average wage data down. Because of this the historic average hourly and weekly wage data actually overstate the average gains and the problem of wage stagnation is worse than the data shows.

    I made the comment earlier as anonymous on the changes in the workweek because I could not figure out how to use the sign in feature. Wordpress does not accept me.

  16. You should look at this: