Real Compensation per hour was over-valued from the 1950s to the 1990s

I want to show a graph with a large discrepancy.

First, we start with the two equations for productivity. Real output (Y) divided by total labor hours (L)... and real compensation per hour (W) divided by labor share (e). These two equations are equal.

Y/L = W/e

I rearrange to solve for labor share (e).

e = WL/Y

This makes sense. WL is total labor income, which is then divided by real output (Y) to give us its percentage of total output (Y).

Let's graph this equation using actual data.

Graph #1

So this graph is in effect showing labor share over the years. As everybody knows, labor share has been falling. Now let's add to the graph the line for labor share of income (Business sector: 2005=100)

Graph #2

Why aren't the lines the same? Oh yeah, I forgot to convert the labor share index with the "effective" conversion by multiplying by 0.78.

Graph #3

We can see that the lines match up much better now, but only at the beginning before 1950 and recently since about 2000. Why the huge difference from 1950 to 2000? This is the discrepancy that I want to show.

If the red line is true, labor share started to fall hard after the Volcker recession and has kept falling ever since, except for the dotcom bubble years. Also, labor share increased in the 1950's much more than is shown by just the labor share plot.

This is a serious discrepancy that needs an explanation.

There are 5 options to explain this discrepancy.

1. Total hours worked was over-valued.
2. Real compensation per hour was over-valued.
3. Real GDP output was under-valued.
4. Labor share itself was under-valued.
5. A combination of the options above.

Whatever the explanation, we would need to see an acceleration of a value during the 50's and 60's, and a deceleration of the value during the 80's and 90's.

Based on working with the labor share data, I would say that labor share itself was under-valued a bit during the 60's, but only by a percentage point or two... not nearly enough to account for the discrepancy. So what other option above is the real culprit?

In my view, the real culprit is...  real compensation per hour (FRED code = RCPHBS).

Graph #4

As we can see in this graph #4, real compensation rose steeply during the 50's and 60's. Then backed off from that growth rate during the 80's and 90's. And as I noted before in another post, professor Martha Olney at UC Berkeley did a video of lecture 23 of her Economics 113 class this spring of 2013. She was talking about productivity. And productivity is calculated as real compensation per hour divided by labor share. So if real compensation is over-valued, productivity would be over-valued.

Here is what I said in my previous post about what she said in her lecture.

"In the video she shows a chart of productivity from 1960 to 2012 about the 2:00 minute point. She mentions that productivity grew in the 60's at 2.8%, from the mid 70's to mid 90's 1.5% and from 1995 to 2004 3.3%. Her point is that there was a slowdown in productivity from the mid 70's to the mid 90's, but nobody has been able to figure why.

She showed that it was not caused by age, gender, oil, manufacturing/services, immigration, education, R&D, choice to work less, and more. None of these factors showed any correlation to the the changes in productivity."

My view now is that there is something wrong with the data for real compensation per hour itself. No one has been able to find a real world explanation for the apparent slowdown of real compensation in the data. Graph #3 above would imply that there is an error in real compensation data.

In the following graph I have adjusted only real compensation per hour in order to undo the discrepancy.

Graph #5

The orange line shows what real compensation would look like in order for the red line in graph #3 to move down and equal the blue line. There isn't a dramatic difference as far as the shape of the official data (blue line). There is still a bit of a deceleration of real compensation from the 70's to the 90's. The orange line looks more reasonable, but it would need some econometric testing.

My basic conclusion based on my working with the labor share data and the remark by Professor Olney is that the real compensation per hour numbers are over-valued. Something is wrong in the way they were calculated before the 80's. And then the numbers were slowly brought back into line.

One takeaway from this analysis... Real compensation has increased more than many say since the 1970's. For example, many say that real wages have been stagnant since the 70's as productivity grew. But apparently real wages were over-valued in the 70's. Which means that real wages have grown more than people think.

Comments

[Fred]

Very interesting - thanks! So the significant stagnation is really a more recent phenomena. I wonder what caused the jump in real compensation about 1998-2004.

[Edward Lambert]

From 1998 to 2002, money was flowing more readily. Natural interest rate jumped up during that time. Liquidity increased throughout the economy. The liquidity made its way into wages while inflation stayed low. If inflation had accompanied the rise in wages, "real" wages would not have risen as much.
The answer is that inflation stayed low as wages rose with excess liquidity.

[Jim Hornsby]

When I read this again today, I wondered about the indexes used to compensate for inflation.

I found this article:
http://www.bls.gov/opub/mlr/2011/01/art3full.pdf

The date was January 2011.

Page 57
"There are two main components that account for the magnitude and direction of the compensation–productivity gap. The first is the difference between the price indexes used to adjust for inflation in the BLS hourly compensation and productivity measures. The Consumer Price Index and the implicit price deflator comprise different baskets of goods and services; if consumer prices rise more quickly than output prices, purchasing power falls and the compensation–productivity gap grows."

"The second component, “labor share,” is the share of output accounted for by employees’ compensa- tion. Labor share is a measure of how much of the economic pie goes to all workers. When labor share is constant or rising, workers benefit from economic growth. When labor share falls, the compensation– productivity gap widens. Concurrently, nonlabor costs—which include intermediate inputs into pro- duction and returns to investments, or profits—rep- resent a greater share of output. Because real hourly compensation and labor productivity, which is output per hour, both include hours worked in their calcula- tions, changes in hours worked have no impact on the gap."

Page 58
"The real hourly compensation measures used in this essay differ from other compensation measures published by BLS. The productivity program develops the measures on the basis of BLS and BEA data, making adjustments for coverage. The IPD for non-farm business output is published by the BEA. The consumer price index prepared by the BLS productiv- ity program is the quarterly average of the monthly BLS Consumer Price Index for all Urban Consumers Research Series (CPI-U-RS) for 1978 through the most recent full year, currently 2009. Changes in the BLS CPI for all Urban Consumers (CPI-U) are used to estimate an index for years prior to 1978 and for recent quarters. The productivity program’s consumer price index based on these measures is displayed in this essay."

I don't know if this would affect your analysis, but I thought you might be interested.

[Edward Lambert]

Jim,
I am reading your comments with great interest. I cannot find that source that says labor share rose during WWII, but I found another paper that shows it was low. But...
But...
That paper turns out to be really interesting because they did a study on capacity utilization and labor similar to what I am doing.
Here is the link for you to look at...
http://www.newschool.edu/scepa/papers/archive/cepa200303.pdf