The media continues to obsess over the so-called “wage decoupling” issue, the gap between the growth rate of median wages and the growth rate of GDP. Let’s start by asking why people are even interested in comparing these two growth rates. What are they supposed to show? Why not compare wages to average global temperature or the average score in an NBA game?
I suppose people must have in mind some sort of concept that GDP is like a pie, and workers are not getting their fair share. But if that’s your concern then why not compare wages to total per capita income? After all, GDP includes things like depreciation, which nobody is “getting”. Income is the “pie”, not GDP.
Because depreciation grows faster than GDP it turns out that there is a “decoupling” between total income and total GDP. Total combined income earned by workers and capital rises more slowly than GDP. That’s not a scandal.
Scott Alexander valiantly tries to make sense out of this mess, but ends up being sort of overwhelmed by too much information, too many claims. Tyler Cowen makes this observation:
On Scott’s broader points (not discussed in my excerpt), I think he is underemphasizing the possibility that productivity may be measuring better than it really performed, and thus there is not so much decoupling at all.
In a perfect world, the mis-measurement of productivity would have absolutely no impact on estimates of “decoupling”. Indeed, let’s take a step back and consider the absurdity of the various measures of decoupling that people use, which all seem to rely on comparisons of real GDP and real wages. Obviously the variables that ought to be used are nominal wages and nominal national income per capita. And any gap between nominal wages and nominal national income per capita would be completely unaffected by the mis-measurement of productivity.
Of course that’s how things would be done in a non-idiotic world, where people don’t use different price indices to deflate national income and wage income. We don’t live in that world, and thus Tyler’s point may have some validity. These comparisons typically deflate wages by the CPI, which rises faster than the GDP deflator. But why?
It would be better if Tyler simply referred to the idiocy of using two different price indices rather than the mis-measurement of productivity, which isn’t really the issue at all.
Scott Winship has the best explanation of this mess, and he shows that there is almost no long run decoupling between average wages and national income, at least in the way people think of the term. (I.e. more of the pie going to greedy corporations. He excludes owner-occupied rent, which most people don’t even think of as income.) On the other hand, there is a big decoupling between median wages and average wages, reflecting greater inequality of wage income.
If growing wage inequality is what people are concerned about, then that’s what they should talk about. Any time you see an article in the NYT discussing the gap between GDP and wages, just stop reading. You will only make yourself dumber by reading to the end.
PS. I’d like to direct your attention to a new book that just came out, called “The ABCs of Austrian Business Cycle Theory“. A better title might have been “The ABCs of Business Cycle Theory”, as you can tell from the product description:
The theories of the so-called “Austrian school” are brought into dialogue with those of John Maynard Keynes, Milton Friedman, as well as modern-day thinkers who are synthesizing the best ideas into a more nuanced but still partial understanding of the business cycle.Part 1: Robert Wenzel, AustrianPart 2: Brad DeLong, NeW KeynesianPart 3: Scott Sumner, Market MonetaristKey terms are also explained in simplified “A-B-C” form, to dispel the jargon that often obscures the relatively simple underlying problem of monetary disequilibrium that is implicated in each analysis.
READER COMMENTS
James D
Feb 27 2019 at 2:56pm
I think this from Arnold Kling sums up the difficulties of the “decoupling mess” nicely:
This, in addition to the “compared to what?” question discussed above makes it, though perhaps an interesting question and certainly advantageous talking point for some, less than useful as a descriptive matter.
Here is the link to Kling’s post: http://www.arnoldkling.com/blog/wages-and-productivity/
Scott Sumner
Feb 27 2019 at 6:58pm
James, Actually, productivity is not the issue here at all. The whole debate is a horrible mess.
Ahmed Fares
Feb 27 2019 at 11:26pm
On the other hand, there is a big decoupling between median wages and average wages, reflecting greater inequality of wage income.
This reminded me of an article by Greg Mankiw. Here is a slice:
There is heterogeneity among workers. Productivity is most easily calculated for the average worker in the economy: total output divided by total hours worked. Not every type of worker, however, will experience the same productivity change as the average. Average productivity is best compared with average real wages. If you see average productivity compared with median wages or with the wages of only production workers, you should be concerned that the comparison is, from the standpoint of economic theory, the wrong one.
https://gregmankiw.blogspot.com/2006/08/how-are-wages-and-productivity-related.html
Benjamin Cole
Feb 28 2019 at 9:08am
Not sure about this, Careful studies show declining share of labor income in relation to GDP, even with depreciation etc., accounted for.
Corporate profits are setting all-time records in relation to GDP, and are much higher today than in the Reagan days. I think this is great, btw.
Maybe declining real wages should not be a surprise. The baby boom entered the labor market, and women too. The southern border, for better or worse, was wide open and workers (not thugs) crossed over by the tens of millions. Factories went offshore. Huge trade deficits do not increase the demand for US labor. Even call centers went offshore.
There has been a 50-year glut of labor in the US.
The screw may be turning. The southern border may be tighter (no one seems to know). Women seem to have topped out in LFP. The baby boom is wearing adult diapers.
Labor costs are high everywhere in the developed world.
Maybe we can see a couple generations of real wage growth in the US. If not, get ready for AOC in the White House.
Scott Sumner
Feb 28 2019 at 12:45pm
Ben, Labor share of national income is about the same as in 1965.
Benjamin Cole
Feb 28 2019 at 7:43pm
Scott—
Re labor share of income. It seems to me that NBER put out at some studies finding a declining share. I do not know if I saved links but I will look for them.
In any event, the topic of how labor is doing in America is like the topic of global warming. People ( even experts ) decide what they believe, and then hunt for the evidence.
The layman has little chance. I thought Scott Alexander did yeoman work.
mike
Mar 1 2019 at 1:46am
Am I looking at a different chart? (serious question … I am assuming labor share of non-farm business sector is essentially labor part of gdp…or is the chart excluding govt, exports , something else etc.?)
Because this chart shows in 1960’s it was 62-64%, and now it is 56-58% in the 2010’s… 5% of the economy is $1 trillion dollars… a HUGE deal. Again, your analysis is usually very solid so I doubt you overlook a $1 trillion number…so where is my very quick analysis wrong?
Scott Sumner
Mar 1 2019 at 3:03pm
Mike, This blog post contains my assumptions.
https://www.themoneyillusion.com/the-very-real-problem-of-wage-inequality/
mike
Mar 1 2019 at 1:48am
https://www.bls.gov/opub/mlr/2017/article/estimating-the-us-labor-share.htm
Not sure but not showing the image I tried to include in my comment… so here is the weblink with the graphy
Warren Platts
Mar 24 2019 at 11:49am
Mike and Scott, here is a FRED chart that shows the same thing:
https://fred.stlouisfed.org/series/W270RE1A156NBEA
Labor’s share of national income declined from a high of 51.6% in 1970 to a low of 41.9% in 2013. Since then it has bounced back up to 43.0%. That’s about a 20% drop.
The figure for labor includes ALL compensation, including any employer paid benefits.
But of course that figure includes CEO pay, that we all know has gone through the roof. The share for the median wage earner has thus gone down even more.
Benjamin Cole
Mar 1 2019 at 12:14am
The Micro-Level Anatomy of the Labor Share Decline
Matthias Kehrig, Nicolas Vincent
NBER Working Paper No. 25275
Issued in November 2018
NBER Program(s):Economic Fluctuations and Growth, Labor Studies, Productivity, Innovation, and Entrepreneurship
The aggregate labor share in U.S. manufacturing declined from 62 percentage points (ppt) in 1967 to 41 ppt in 2012. The labor share of the typical U.S. manufacturing establishment, in contrast, rose by over 3 ppt during the same period. Using micro-level data, we document a number of striking facts: (1) there has been a dramatic reallocation of value added to “hyper- productive” (HP) low-labor share establishments, with much more limited reallocation of inputs; (2) HP establishments have only a temporarily lower labor share that rebounds after five to eight years to the level of their peers; (3) selection into HP status has become increasingly correlated with past size; (4) labor share dynamics are driven by revenue total factor labor productivity, not wages or capital intensity; (5) employment has become less responsive to positive technology shocks over time; and (6) HP establishments enjoy a product price premium relative to their peers that causes their high (revenue) productivity. Counterfactual exercises indicate that selection along size rather than shocks or responsiveness to them is the primary driver of the labor share decline.
Scott Sumner
Mar 1 2019 at 3:03pm
Ben, Not sure how your comment relates to this post.
Warren Platts
Mar 24 2019 at 12:18pm
I decided to take a crack at the median wage issue using Scott’s suggestions to use nominal Gross Domestic Income and wages, and constructed the following chart that has nominal median usual weekly wages X 52 weeks to get an annual figure divided by nominal GDI.
https://fred.stlouisfed.org/graph/?g=nnZO
The data for weekly wages only goes back to 1979, but then a median worker’s share of GDI was 4.85 “nano-GDIs” (nGDI). That is, said worker earned 4.85 billionths of the GDI in 1979. This figure declined steadily to a low of 2.28 nGDI in 2017. That represents a 53% decline. And since the data goes back only to 1979, the overall decline since the early 1970’s must be even worse.
Thus to get back to zero, the median worker wages would have be more than doubled. Since the 2017 annual median wage was $44,720, if the wage share of GNI had stayed the same since 1979 (perhaps not coincidentally the peak manufacturing employment level), the median annual wage would have to be $95,150… Since the “disconnect” happened when Bretton Woods broke down, allowing our trading partners to manipulate their currencies, the true figure is more like $100K.
Thus I think it is undeniable that the disconnect between productivity and median wages is real. It used to be considered a surprising fact in economics that the share of national income was relatively stable (Bowley’s Law).
That is where the old rule of thumb that wages were proportional to productivity came from. Production growth in the long run is driven by productivity thus if the labor/profit ratio is constant, the wages should rise directly in proportion to productivity.
Warren Platts
Mar 24 2019 at 1:10pm
Oops. I see you want to take out depreciation. So I redid the graph by subtracting out “Current-Cost Depreciation of Fixed Assets” which amounts to about 15% of GDI so I think that is the right figure. However, it doesn’t make a difference. Share of the median wage earner goes down 52.5% rather than 53%.
https://fred.stlouisfed.org/graph/?g=no33
Warren Platts
Mar 24 2019 at 2:59pm
OK, this is slightly more complicated than I realized at first. I should take into account the number of workers in order to come up with a GDI/worker figure, and compare that to the median wage. So here it is:
https://fred.stlouisfed.org/graph/?g=no6u
It also shows the BLS labor share chart that includes CEO and movie and basketball star pay. Thus the median worker’s share decline is worse than the overall decline in labor’s share. Median worker share is down by 26.4%.
Thus the economic loss to the median worker only amounts to $16,000 per year…
Comments are closed.