CEPR - Center for Economic and Policy Research


En Español

Em Português

Other Languages

Home Publications Blogs Beat the Press Is Productivity Being Translated Into Pay Increases?

Is Productivity Being Translated Into Pay Increases?

Wednesday, 17 July 2013 12:44

Jim Tankersley has a post in Wonkblog asking whether there has been a divergence between pay and productivity over the last three decades. The post notes a study from James Sherk at Heritage which makes several valid points. First, part of the gap between average pay and productivity is explained by a growing share of compensation going to health care benefits. Second part of the gap is the result of the fact that productivity is measured in gross output, whereas only net output is available for consumption. Third, we use different deflators to measure output than consumption. The consumer price index, which is used to measure real wages, shows a higher rate of inflation than the implicit price deflator. If we use the same deflator to measure real wages and output, then this also eliminates much of the seeming gap between productivity and pay.

I had made these points myself a few years back. My conclusion was that we were really looking at a story of upward redistribution from middle and lower income workers to those at the top, doctors, lawyers, and especially Wall Street types and CEOs. Distribution from wages to profits was not a big part of the picture.

But that was back in 2007. The picture looks a bit different today. The graph below shows the labor share of net income in the corporate sector. This is a bit simpler than constructing productivity and pay data, but it should get at the same issue. I have pulled out depreciation and also indirect taxes, so the division is simply between labor income and capital income. I also show the share of labor compensation in after-tax income in the corporate sector.


In the data in the graph it certainly looks like we are seeing a redistribution from labor to capital at least in the years since the crash. For the last three years the labor share of before-tax income was lower than at any point hit in the 1980s and 1990s. The labor share of after-tax income is more than two percentage points lower than at any point in the 1980s and 1990s. That looks like a fairly serious redistribution.

We can throw in the usual qualifications about the data being erratic and cyclical, but it's pretty hard to find a way to make this redistribution disappear. It may prove to be the case that if the unemployment rate falls back to more normal levels then workers will get increased bargaining power and will be able to recapture more of the gains from productivity growth, but that is not happening now.

Comments (25)Add Comment
Low labor share has created a constraint on productivity, output and employment.
written by Edward Lambert, July 17, 2013 1:56
I suppose a falling labor share will eventually take the economics profession by surprise when they realize its effects. My own work into effective demand has opened my eyes to economic constraints that others are not seeing.

Basically, real GDP has a constraint of $14.1 trillion if counter-measures are not successful.
Unemployment has a constraint of 7.0%.
Productivity is constrained by effective demand.


Unless we find a way to raise labor share, the economy will stay weak for a long time.
written by AlanInAZ, July 17, 2013 2:49
The chart seems to show a trend for weakening labor share starting around 2000. A chart Dean included a few posts back showed China exports starting their big run-up around the same time. I don't think this is a coincidence.
I Agree - But for the Lawyers
written by sherparick1, July 17, 2013 3:25
I think lawyers, except for those in the commanding heights, have joined the rest of the middle class/working class in seeing their incomes drop and productivity gains shift to the 1% as more legal work is outsourced via the internet and automated search engines and discovery software replace attorney billable hours. See http://www.lawyersgunsmoneyblo...ermination and http://nashvillepost.com/blogs...ry_decline
written by skeptonomist, July 17, 2013 3:30
Corporate profits (FRED: A055RC0A144NBEA) as a fraction of GDP shot up starting around 2003 and reached a peak in 2005:


They almost doubled from near a long-term average value. They had actually gone down slightly by 2007, and after a brief dip through 2009 they are back up to record levels. I can think of no measure of wages that would show a comparable increase. Unless this is some accounting artifact, this is a matter of profits capturing much more than productivity increases since 2002.
To depersonalize it a bit, too much income goes to the legal and financial sectors
written by Rachel, July 17, 2013 9:02

It isn't just that some lawyers are tremendously overpaid. It's also that we have too many of them. I imagine that we could do with fewer people in the financial sector as well, not just with fewer of the ones making preposterous salaries, but fewer overall, especially if we had a small transacttion tax to encourage efficiency.

But one reason that law and finance are overpopulated is because, for purposes of reducing competition, those in other professions such as medicine, in particular, have created such high barriers to entry in this country.
must look at global issues....
written by pete, July 17, 2013 11:21
Essentially, in the U.S. and Europe there is a much higher K/L than everywhere else. Thus, while labor is very productive in US/Europe vis a vis the rest of the world, this is an anachronism which has led to temporary high returns to labor in US/Europe vis a vis the rest of the world. Now with property rights more established in other places, more or less, capital is flowing to where K/L is small. The trend is clearly visible in the chart, and probably if you flipped it it would reflect growing labor gains in the rest of the world as workers move from planting rice to putting Iphones together. Socially (globally) this has to be seen as a good thing...global labor is doing quite well, thank you. Unfortunately, it does not look good for low skilled labor in the U.S. Bottom line: for those in the U.S. who are not going to make it with education, we need to track them into trades earlier where they can build at lease some human capital. I was asking a sonogram tech at a hospital about the education...pure certificate, like 9 months, not even a 2 year degree. Why can't that be taught to 17 year olds instead of telling them they are failures if they don't go to college....geesh. And for goodness sake bring back unpaid apprenticeships and internships. Must build the human capital.
Productivity has a Denominator
written by John, July 18, 2013 2:29
Productivity is production per worker or production per work hour. So a store can reduce the number of workers by setting up self-service computerized checkout stands. A hedge fund that buys up a comfortable high-tech company will make it more productive by downsizing or "right-sizing" or "reengineering" its workforce.

They will also reduce benefits. If the company has paid off its initial debt, and is sitting on a hoard of cash in preparation for future expansion (or simply a rainy day), the hedge fund will declare a special dividend for stock-holders. Citing some theory that there is an optimum debt for a company to hold (efficiency or something), it will borrow anew and send the money off as more special dividends.

Then the company is put back on the market, as a shadow of its former self -- almost a carcass of its former self. It may have higher productivity but lower production. Maybe the higher "productivity" has gone instead into the hedge fund owners' cash.

But back to productivity with a denominator. If the productivity gains all go to the top 20%, increased productivity means fewer workers and fewer work-hours. And because of lower (real) wages, those fewer workers try their best to get as much of the work-hours for themselves, trying to work overtime.
You've made the folks at Heritage very happy. Bravo.
written by Marko, July 18, 2013 2:30
Why not simply plot the ratio of nominal total comp to nominal gdp and skip the deflator fudging.
The chart does show nominal data
written by Dean, July 18, 2013 4:06

I suggest you read the post and look at the charts. If the data make Heritage happy, then this is our problem. No one pays me to lie and it frankly is not a good political strategy for progressives in any case.

The right has the money to force people to take its lies seriously. We don't.
Where in the globe is wage share rising, Pete?
written by David M, July 18, 2013 8:51
Not in the largest labor market in the world. China's wage share has decreased for 22 years in a row:

Wages may be rising in the developing world, but not as fast as the rate of exploitation.
written by Lrellok, July 18, 2013 11:12
Heritage Point 1) As you have already discussed on multiple occasions, healthcare services are massively overcharged. To call "health insurance" wages implies that where the employee paid cash wages equal to the value of the health insurance, they would derive as much utility from both. But if healthcare is overcharged, this cannot be true, they must derive more utility from the wages then the health insurance.

Point 2) I am not fully understanding how that is relevant. Wages at optimal equilibrium should reflect the marginal additions made by labor to production. If the value of wages has no relation to the value the job adds to anything, that alone is a recipe for gross inequality, and gross market failure beside. Yet you seem to be accepting that point out of hand. Why should we be looking at consumption at all, much less net production. Wages are supposed to reflect a potion of gross production precisely because labor adds a portion to gross production. The end user price must functionally include all prices up to that point, all material, capital, and labor inputs. For this not to be true, someone must be operating at a net loss (cough LABOR cough).

Point 3) I concur with Marko, and you have not adequately addressed his points in the above article. Why not stop using deflators at all? I have done alot of research into Wages and productivity and generally make a point of using nominal data where ever possible, via the mechanic of dividing the wages by the productivity to get a ratio. This happily skirts the entire deflator debate, and still shows horrifying lags in wage gains. Also, why are you crediting heritage with any shred of honesty, i generally assume nearly anything they produce is methodologically garbage. If they are happy, it is because they manipulated the data enough to show the answer they wanted.
written by Peter K., July 18, 2013 11:13
As candidate Obama pointed out in a BusinessWeek interview, the last time labor shared in productivity gains was the late 90s. I believe it wasn't out of the goodness of Greenspan's Randian heart that he allowed unemployment to reach 4 percent. He was fighting international financial crises. If we had proper currency-fiscal-monetary policies, demand management policies, we could reverse the trend. Big if.
Why average?
written by keenan, July 18, 2013 1:41

This may be a naive question, but why use “average hourly wages” and not “median hourly wages" in the chart comparing wages with productivity (Tankersley article, chart 3). If I understand Baker’s point, he’s saying salary gains went mainly to the top earners. So wouldn’t this be evident by basing the comparison on median salaries?

written by Lrellok, July 18, 2013 2:14
Who's No. 1?
written by lbmexec, July 18, 2013 2:43
Keenan and I are competing to see who can ask the most naive question; here's mine: Does it necessarily follow that whenever productivity rises, labor should share in the gains? Definitely yes if you're requiring 50 people to do the same work that once took 100 people. But if you bought a machine that doubled your output, those remaining 50 jobs might require less effort and/or skill. My understanding is that the productivity gains of the past 60+ years have come primarily from automation. It would be nice if owners shared their gains voluntarily, but if you expect them to do so and are disappointed when they don't, then neither Keenan nor I win the game.
I Didn't Use a Deflator
written by Dean, July 18, 2013 5:11
please read my note until you understand that point -- there is no deflator there. I took nominal over nominal. If you don't like what the division gives us, there's nothing I can do about that.

In terms of health care being waste, i say that all the time -- it's a different question. In terms of the median worker as opposed to the average, READ THE POST.

The question is whether income is going from wages to profits, not whether workers are enjoying the good life.
David M....
written by pete, July 18, 2013 8:47
again the pardox. Texas wages are supposed to be low but population is flowing there. China wages are supposed to be low in the cities but folks are flocking there. What gives? Rice planters, small farmers etc. who worked 80-90 hour weeks are now able to come to Shanghai or GuangZhou and make a lot more money. What the data suggest is that the hourly wage is not rising. That is weird since many companies are shifting production out of China due to high wages. Really, what all this data shows is that narrowly defined "labor" is a poor indicator as to how well off folks are doing. Nobody says China was better off in 1976.
written by Lrellok, July 19, 2013 6:03
Professor, with all possible respect, nominal what? What you are dividing by what is not clearly cited in your post. I see a graph, but i have no figures or citations telling be the source data for that graph. I know full well that if i divide Wages by GDP i see a clear downward trend starting in the early 70's. I know if i divide corporate profits + owner income by GDP I see a clear upward trend starting in the early 80s. Yet you are asserting not only is there no relevant gain in corporate profit relative to production prior to 2008, which i know to be false, but that wages did not start dropping prior to 2007, which i also know to be false. I can see no methodological reason why gross and net aggregate production should not be EXACTLY the same number, all inputs must be internal to the market. I have run wage data a dozen different ways, going to far as to divide GDP per employee and compare it to median income, which also shows a clear drop starting in the 1970's. Where the heck are you getting these numbers from?

David M Your causation is inverted, labor is cheap because people are flocking to the cities. The act of working is socially coercive, people are forced, via threat of starvation, to continue producing labor at the same quantity regardless of changes in price. Thus, labor markets cannot be considered "open" and quantity determines price, not vice versa.
written by David M, July 19, 2013 11:00
Pete, I said in my comment that wages were rising in China. But this thread isn't about wages, it's about wage share, which is falling--i.e., wages are not keeping pace with productivity growth, and the difference can't just be explained by increased investment. This is happening not only in the developed countries, but around the world.

Lrellok, I think you meant to reply to Pete (who was replying to me).
The middle class rises to the upper class
written by AlanInAZ, July 19, 2013 12:45
This post was referenced in Mankiw's blog and I think deserves comment at some point

written by David M, July 19, 2013 1:24
Alan, an easy rejoinder to the AEI post if to take into account the fact that median income has increased over the same period, from roughly $42K to $54K at its peak, or about a 25% increase. So of course households are moving upwards, if all you're looking at is total inflation adjusted income. This is why quintiles are preferred over absolute income levels to understand distribution.

But that means that someone making $25K in 1967 was making 60% of the median, while someone making $25K in 2000 was making only 46% of the median. Overall the AEI chart demonstrates widening income disparities, more than disproving them, since the level of $70K has grown, this shows that most of the income gains have gone to the top earners.
written by AlanInAZ, July 19, 2013 4:37
David M

I realize the chart doesn't disprove the outsize benefits accruing to the very top, but it does seem to show that more households are enjoying higher incomes than in the past with steady increases through the years. My thought was that this reflects the trend towards multiple household incomes (husband, wife, children) that started in 1960's. As women's salaries picked up then household incomes would also rise as shown in the charts.
Labor Compensation over Net Corporate Income
written by Dean, July 20, 2013 5:36

If go to the BEA Website's interactive data section http://www.bea.gov/iTable/iTab...D=9&step=1
Go to Table 1.14 you get a table giving gross value added in the corporate sector. This is where there is a division between wages and profits (as opposed to government or non-corporate business).
You have to take net output to be serious because no one eats depreciation, which does increase as a share of output over this period. It also makes sense to pull out taxes on production (mostly tariffs), since neither workers or capitalists see this.

So you take net value added (line 3) minus taxes (line 7) as your denominator. The numerator is labor compensation (line 4). It's a very simple calculation.

Again -- this has zero, nothing, nada to do with median wages. We are asking a different question. Was there a redistribution from labor to capital. The data shows that there was not much from 1980 to 2007 as the charts indicate. It is a different world since 2007.
written by Lrellok, July 20, 2013 5:20
You are equating profits with value added? Why? Did rent seeking suddenly vanish from all American corporate industries? And why would you discount non incorporated industries from adding value? Why would you limit "capital" only to corporate equities? And you still have yet to answer why you are counting "Compensation" as wages when companies keep defaulting on their pensions and medical services are massively overpriced. Also, where do "bonus packages" figure into this, if an executive gets a hundred million in stock options, is that counted as salary, compensation, or capital? Surely it cannot be counted as value! The number of constructional flaws i am seeing here are horrifying. TO say nothing of the fact that if your construction where true, it would mean that technology had exactly zero effect on wages prior to 2007, since compensation to output was indicated labor was equally efficient though out the whole period (back to 1930 if you look). While i whole heartily oppose the "blame the robots" methodology, i am curios if you realize you have just destroyed it?
Look at the Post Again and Again
written by Dean, July 21, 2013 12:07

the structural flaws you are seeing are all your one invention. Please learn what these categories mean if you want to argue over them. Value added refers to incomes generated in production. I have conflated nothing -- I am using terms in accordance with their definition.

Compensation refers to money paid to workers whether in pensions, health care, or wages. It does NOT refer to promised benefits. This is money that goes out the door from the company. And no, companies have not taken this money back.

Please do some homework if you want to have a debate on this topic.

Write comment

(Only one link allowed per comment)

This content has been locked. You can no longer post any comments.


Support this blog, donate
Combined Federal Campaign #79613

About Beat the Press

Dean Baker is co-director of the Center for Economic and Policy Research in Washington, D.C. He is the author of several books, his latest being The End of Loser Liberalism: Making Markets Progressive. Read more about Dean.