How can faster productivity growth *not* raise middle class incomes?

by on April 22, 2015 at 11:34 am in Economics | Permalink

Jared Bernstein worries that it won’t.  But how can it not?

Let’s say an entrepreneur introduces a new product, call it the iPhone.  At first the relatively wealthy are the main buyers and at first a lot of the gains accrue to the IP holders.  Over time, however, the IP rents erode, smart phones become more popular, even with stupid patent laws smart phone innovations trickle down into cheaper models, and so on.  The middle class ends up better off.

Of course not all productivity gains are innovations.

Let’s say Ford can suddenly make cars — a known product — more cheaply.  Even in a world with market power, this expands output and lowers car prices, thereby raising real wages.  You might think labor won’t “get its fair share,” but still it should raise real wages, and that’s ignoring the possibility that wages in the auto plant might be bid up a bit.

So when some economists argue that faster productivity won’t raise middle class incomes, which model do they have in mind?

Even if you think income is shifting from capital to labor, how does the productivity gain worsen this shift?  (That is, real wages still ought to go up.)  Inquiring minds wish to know.

I am, by the way, familiar with a variety of diagrams showing a disconnect between productivity and real wages.  I’ll be happier with those when a) I know they are using common deflators, and b) they take adequate account of the possibly dysfunctional productivity in our hard to measure sectors of health, education, and government consumption.

Addendum: Read Scott Winship.

1 JFA April 22, 2015 at 11:43 am

Doesn’t this complaint go against Tyler’s arguments in The Great Stagnation? He thinks median earnings have been stuck since 1973, but productivity has increased since then. Where am I reading Tyler incorrectly?

2 mulp April 22, 2015 at 1:31 pm

When productivity increases, jobs get cut, wages slashed, profits go up creating wealth, and greater wealth spurs consumers spending other people’s money to consume more.

Just remember, workers are not consumers, so cutting labor costs increases profits increasing wealth increasing consumers spending other people’s money.

Consumers are not workers because they spend other people’s money.

In other words, incomes are irrelevant to growth, but very relevant to profits, so lower incomes are good for growth.

Simple free lunch economics. Far superior to the economics of the 30s-70s based on labor income driving GDP growth because debt is something to be reduced.

3 E. Harding April 22, 2015 at 3:19 pm
“Far superior to the economics of the 30s-70s based on labor income driving GDP growth because debt is something to be reduced.”
-AKA oil and natural gas. In the long run, nominal variables are irrelevant.
“When productivity increases, jobs get cut, wages slashed, profits go up creating wealth, and greater wealth spurs consumers spending other people’s money to consume more.”
-[citation needed]. You are just engaging in the Old Labor equivalent of praxeology.

4 collin April 22, 2015 at 2:26 pm

Then why have real wages stagnated during the last 40 years despite higher productivity? For the most part non-managerial wage have fallen the last 40 years with interruptions in the late 90s and some increases the last 2 years.

Secondly, why do the most high productive economies in the world (Singapore, US, Germany, etc.) have such a low birth rate?

5 collateral April 22, 2015 at 2:35 pm

Real compensation hasn’t stagnated, only real cash wage. The shift in compensation from cash to non-cash compensation has to do with distortions created by the tax code.

I don’t see what birthrate has to do with anything.

6 Bill April 22, 2015 at 3:00 pm

Non-cash compensation is a way for employers to lower compensation costs, no? An employer is obviously more monopsonistic than individual employees, and thus can obtain lower unit prices for goods than the wages they would have to pay for individual employees to purchase the goods. Furthermore, the employees may want to purchase more expensive goods, and supplying cheaper substitutes can mitigate that.

7 PavJ April 22, 2015 at 3:05 pm

I tend to lean towards the idea that increasing healthcare costs have eaten into what would have been wage compensation. The insurance is getting expensive and workers value holding onto their healthcare over having more cash in hand.

8 E. Harding April 22, 2015 at 3:20 pm

-That ended in the mid-1990s.

9 collateral April 22, 2015 at 3:54 pm

I haven’t seen any data series that shows real median total compensation per hour. Do you have a link? The best proxy I’ve seen is a real total compensation for production and nonsupervisory workers in nonagricultural establishments and that shows growth since the mid 90s (e.g.

10 E. Harding April 22, 2015 at 4:38 pm

I mean the fringe benefits becoming a larger portion of compensation. That did end in the mid-1990s. But real wages have been increasing since then, undoubtedly, though by no means as fast as in the 1960s.

11 collin April 23, 2015 at 9:57 am

Why does the birthrate have to do with wages? If wages for the middle class is always increasing (as productivity has) then why does the average (in the developed world) family have less children? If you ask the average family, you hear we can’t afford more children.

It hits the great modern contradiction is the richer societies get, the less they can afford children. And if wages were increasing more, then average families would be bigger.

12 W.C. Varones April 22, 2015 at 11:46 am

I’m no Piketty commie, but housing price increases thanks to the Fed’s asset bubbles have eaten up a lot of whatever share the working class should have gotten from productivity gains.

13 CC April 22, 2015 at 1:02 pm

Housing prices have increased because we’ve drastically reduced the supply via zoning/regulation, not because of monetary policy which affects the economy and prices broadly.

14 JonFraz April 22, 2015 at 2:18 pm

There was no great holocaust of houses leading up to the Bubble, and in fact we were building new houses like mad for a while.

15 E. Harding April 22, 2015 at 3:21 pm

As Kevin Eerdman points out, the per capita housing stock was not increasing at this time.

16 collateral April 22, 2015 at 2:42 pm

There’s more than one cause. I’d credit massive subsidization in the mortgage market (via Fredie, Frannie, Ginnie, FHA, VA, the tax code, and fed MBS purchases) over zoning in terms of nationwide prices. Local markets sometimes have the effects the other way around (e.g. San Francisco).

17 rationalman April 22, 2015 at 9:40 pm

Collateral has somewhat right, but the private markets, in their mismeasurment madness, and musical chair games, were the primary culprit behind the madness.

As to E. Hardings point: not true in inland Calif areas. Per capita housing stock went through the proverbial roof.

Lastly, give W.C. Varones the tinfoil hat award. He deserves it.

18 Floccina April 22, 2015 at 2:53 pm

The stock of houses is bigger and better now even though zoning and regulation have had a negative effect but even these are not totally without benefit.

19 Bill April 22, 2015 at 3:14 pm

That depends on how you define “better”. The housing stock may be bigger, but a lot of that stock is in the exurbs. The market seems to think that older and smaller houses in city centers or close-in suburbs are “better” than big, new houses out in the exurbs, judging by prices. And the houses in the exurbs are further from employment and services, and require car, gas, and long commute times.

20 louis April 23, 2015 at 5:03 pm

House prices are not housing costs, they are asset prices. Look at rents instead.
To the extent the house price / rent ratio has expanded, it is because interest rates have fallen. And if you choose to buy your home, this means you can borrow to fund its purchase more cheaply, with the rate locked in for 30 yrs.

21 Brian April 22, 2015 at 11:46 am

In the case of the iPhone, or any other new good, are you certain that cheaper models increase real wages. Pre-iPhone nobody had one and, therefore, were spending $0 on them. Post iPhone (most)everyone has one and is spending something greater than $0 for the privilege. It’s not clear to me that the middle class real wages have increased as a result without incorporating a whole host of potential offsets to the fact that we’re now spending more on phones than we once did.

22 Rowz April 22, 2015 at 12:31 pm

Remember that all of us are free to simply avoid buying iPhones, leaving us unaffected by their introduction. It is therefore not possible for iPhone introduction to reduce real wages.

23 j r April 22, 2015 at 12:44 pm

I tend to agree with Tyler here, but it’s not “not possible.”

It is possible that iPhones introduce a Red Queen situation. In other words, not having a smartphone might be an effective reduction in real wages.

24 Alex Godofsky April 22, 2015 at 12:55 pm

Yes, but as actually applied to the real world that’s absurd. It implies you look at 2015 technology and say “man, I hate this stuff, I’d be so much happier if it just didn’t exist and we went back to 1975”. It amounts to “well I guess technological progress is a bad thing and we should suppress it.”

25 Urso April 22, 2015 at 5:05 pm

I would love to not have an iPhone. But if I didn’t carry one, I would be fired. This would not have been a true statement in 2004. qed

26 Okay April 23, 2015 at 2:18 am

Alex, you have a naive view of how these optimization processes work. None of them optimize for human happiness (why would they?)

A satisfied customer is a dangerous thing. That’s why there are multi-billion dollar marketing industries specialized to use every human psychology trick to create new unscratched itches that didn’t exist before.

27 E. Harding April 22, 2015 at 3:22 pm

“It is possible that iPhones introduce a Red Queen situation. In other words, not having a smartphone might be an effective reduction in real wages.”
-How’s that possible?

28 j r April 22, 2015 at 4:10 pm

“How’s that possible?”

The same way it is possible that having only a high school diploma in Detroit in 1960 means something quite different from having only a high school diploma in Detroit in 2015.

There is a such thing as a positional good.

29 spencer April 22, 2015 at 12:50 pm

No, there is always the possibility that the consumer gives something else up for their iPhone and it rsults in lower real wages.

30 E. Harding April 22, 2015 at 3:53 pm

Not possible.

31 Bill April 22, 2015 at 3:32 pm

Not necessarily. There is a sort of network effect where everybody adopts something and you can’t opt-out selectively without opting-out of other things. My employer requires that I basically have texting, email, internet access at all times. Many jobs are like this these days. Contemporary social life demands these things as well. You can’t opt-out of smartphones without also opting out of employment prospects, jobs, promotions, social scenes, etc.

32 RZ0 April 22, 2015 at 11:48 am

In The Great Stagnation he argues that there have been no world-changing inventions since television because no one has invented foo powder. (I read this book on my telephone, by the way.) In Average Is Over he argues that technology is changing so quickly that we could all be replaced by computers.
Here, he argues that his thought experiment is superior to the evidence that contradicts it because. Just because.

33 rayward April 22, 2015 at 11:51 am

There’s an assumption that productivity growth and increased output go together. Not necessarily. Productivity growth could simply increase output per worker while producing the same total output; same output, but with fewer workers. That could increase the wages of the workers who remain but leave total wages flat – the displaced workers having zero wages.

34 KPres April 22, 2015 at 12:45 pm

Your scenario has flat total wages with fewer overall hours worked. Sounds like an increased standard of living to me.

35 Pshrnk April 22, 2015 at 1:21 pm

Fewer workers needed and more people unemployed/needing jobs. Seems a simple formula to decrease the wages of the fewer workers.

36 Thiago Ribeiro April 22, 2015 at 9:51 pm

Indeed, reserve army of labor.

37 vv April 22, 2015 at 11:51 am

When productivity grows exponentially but consumption only grows linearly.

To take one example –> there’s perhaps a linear increase in movies being ‘rented’ economy-wide, but an exponential decrease in movie rental employees (blockbuster at peak = 60K employees; netflix = 3K employees)

You see a similar [exponential productivity boost] vs. [linear consumption increase] to Uber, Amazon, Google, telecom… and many many other sectors of the economy.

38 8 April 22, 2015 at 12:03 pm

Why would wages increase when immigration, social and trade policy all work to drive wages lower?

39 Pshrnk April 22, 2015 at 1:23 pm

To drive whose wages lower? We talking American or Worldwide?

40 j r April 22, 2015 at 1:54 pm

Doesn’t matter if it’s American or worldwide.

America is not one big labor market!

41 DC April 22, 2015 at 12:05 pm

The question seems to assume a closed system. i.e. the production, the IP, and middle class are in the same bucket. You can certainly imagine a situation where all the production has moved out of the US leaving no jobs but cheaper imported goods for the middle class to buy with no income. The argument is probably correct when you regard the entire world but not necessarily any particular part of it.

42 Nemo April 22, 2015 at 12:24 pm

By “real wages” you seem to mean that even though the actual salary may be lower (due to; e.g., open borders putting downward pressure on actual wages) the overall basket of goods and services are effectively better. I don’t think this is what the average person thinks of when they think of better “wages”.

43 Alex Godofsky April 22, 2015 at 12:58 pm

Yes, that’s precisely what “real” (as opposed to “nominal”, not actual) means in economics.

44 Clover April 22, 2015 at 12:26 pm

We seen in the last few decades that productivity gains have not flowed to the workers. One reason is automation. You didn’t mention how Ford would make cars much more cheaply. It might be through laying off a large portion of the workers and replacing them with machines. I do not see why we should assume that automation, immigration and outsourcing won’t continue to assure that productivity gains flow to capital and not to labor.

45 Sebastian H April 22, 2015 at 12:41 pm

I’m not an economist so I apologize if I’m using terms incorrectly.

If the money gains from productivity go mostly to the rich isn’t it possible that they will then bid up the prices in certain crucial goods such that the real wealth is not made up in the kind of iPhone like products you are talking about. I’m specifically thinking of education, housing and medical care. (Medical being a tricky area. Innovative pills may be accessible. Innovative treatments may be less so. Once standard care (think rehabilitation or personal care) may be priced out.

46 Lord April 22, 2015 at 1:05 pm

This is it. Innovation doesn’t end, and IP strengthens through obtuse laws. Prices don’t fall, but gains are turned into asset price rises. To a small extent this does increase middle class wealth, but most of it stays at the top.

47 ladderff April 23, 2015 at 3:57 pm

Yeah you’ve pretty much got it.

48 Ryan McKean April 22, 2015 at 12:45 pm

You really need an “upvote” button.

49 meets April 22, 2015 at 9:06 pm


50 Yancey Ward April 22, 2015 at 12:53 pm

There is a blindness in a lot of people, seen in some of the naysayers of Cowen’s comment. I first noticed it about 15 years ago when some prominent economists started pining for the world of pre-1980.

51 The Anti-Gnostic April 22, 2015 at 1:16 pm

I’m beginning to think the policy choices are on a curve between paying our lower-tier citizens at the cash register and they work, and we protect their employment prospects with border controls and tariffs, or we open the borders and eliminate tariffs, and then we pay our lower-tier citizens welfare and they don’t work. And I think paying people on the left side of the curve not to work has bad consequences.

I was talking with some Syrian friends recently, and they mentioned that Lebanon bans Chinese imports, under the rationale that if a single Chinese factory were to target a particular Lebanese market, then a whole town of Lebanese lose their livelihoods. Now, I am sure there are all the data and models in the world showing how free trade is a trillion-dollar bill lying on the sidewalk, but I can tell you the Lebanese aren’t buying it.

Of course, a lot of them are over here instead of over there, too.

52 spencer April 22, 2015 at 12:58 pm

What is this about using the same deflator?

The BLS data on productivity and real income uses different deflators to measure real output and real compensation they also provide the data so that it is very easy to deflate compensation with the output deflator.

There are legitimate reasons why you should use different deflators for compensation and business output.

53 spencer April 22, 2015 at 1:02 pm

One of the reason the BLS shows real compensation lagging productivity is that BLS does not include stock options in the compensation data. Does anyone have an estimate of how big an impact this has? The gap between productivity and compensation began about the same time that stock option started becoming significant.

54 Pshrnk April 22, 2015 at 1:27 pm

Middle class have a lot of stock options do they?

55 JFA April 22, 2015 at 1:14 pm

@Spencer – is there a reason why you think stock options would have a large impact on the measurements? And what do you mean by stock options “becoming significant”? Here’s what you get if you google the question: 9 million workers out of about 150 million. So my estimate of the impact of stock options on median compensation: 0.

56 spencer April 22, 2015 at 4:03 pm

The mid 1970s was when the scale of stock option use started rising sharply and the bull market of the 1980s was what really drove it becoming widespread.

The BLS also reports in the spring 1999 Labor Review:
The ECI currently excludes compensation in the form of
stock options. Traditionally, the incidence of payment in the
form of stock options has been low, and stock options were
not believed to have the potential to affect the
ECI greatly.
However, in light of the apparent growing use of this form of
compensation, the Bureau fielded a nationally representative
survey to determine the incidence of new stock option grants
in 1999. The survey, of about 2,100 establishments, was fielded
between February and June of 2000.
The results of the survey showed that 1.7 percent of all
private-industry employees and 5.3 percent of employees in
publicly held companies received new stock option grants in
1999. As expected, grants were m
ore prevalent among higher
paid employees (12.9 percent of all employees earning $75,000
or more), larger establishments, and certain industrial sec

57 Devin Helton April 22, 2015 at 1:15 pm

It depends on the limiting reagent in production. The limiting reagent may be labor, natural resources, or some input owned by a monopoly rent-seeker. If natural resources or monopoly control is the limiting factor, then increasing labor productivity will not increase overall output. If some new oil drill replaces the need for human mechanics, but is no more efficient at producing oil, then overall output does not increase. Increasing labor productivity will merely drive down the price of labor, output will be the same, and so wages will be lower. Those who own the monopolies or natural resources will be better off.

So for instance, new tech might make home building use lots less labor. But if most of the cost of the home is in the lumber and the land, then the overall output will not rise much. Wages will be pushed down, as the construction workers are made redundant. But home prices remain high. The existing owners of land are better off, but the non-owning laborer is worse off.

For further reading, read the section about the “Plebs” in my blog post on the Plight of the Four Economic Classes.

58 grad student April 22, 2015 at 1:24 pm

spencer: the BLS does include the majority of stock options in the compensation data. Specifically, it gets corporate compensation data from the national accounts compiled by the BEA. The BEA’s measure of wages is taken from state UI administrative data that is generally understood to include the difference between market and exercise price of nonqualified stock options (NSOs) at the time of exercise. (This is closely related to their tax treatment by the IRS.) Incentive stock options (ISOs) are not included, though. The impact of stock options on the compensation data is visible around 1999-2000, when there is a big increase in compensation partly attributable to options.

As an incredibly wonky side note, the world would be a better place if people understood how the BLS labor share data (and their labor productivity & costs release in general) is mostly sourced from the national accounts. For instance, here is a fred graph that shows how the BLS labor share for the nonfinancial corporate sector is simply line 20 of NIPA table 1.14 divided by line 17:

The nonfarm business labor share, which is more popular, is a little more complicated because it also relies on some internal black magic in computing the contribution from the non-corporate business sector. (How much of a proprietor’s income is “labor” versus “capital”?) The nice thing about connecting the labor share, productivity, and compensation data to the national accounts is that they’re embedded in a much larger framework and one can get a much better sense of how everything fits together (rather than plucking two inconsistent measures from fred at random and riffing on how the 1% is responsible for any discrepancy, which seems to be the current progressive MO).

59 spencer April 22, 2015 at 3:51 pm

JFA, I suggest you go to this link at BLS dated 2013

Does the Employment Cost Index include stock options?

The ECI does not include stock options. We have conducted tests in order to determine the prevalence of stock options across the economy and currently collect stock options access as an emerging benefit

60 yo April 22, 2015 at 1:35 pm

Perhaps it’s diminished, sometimes even negative returns? My model goes thus: Somebody makes cars cheaper, more affordable. More people use them and cut time off their daily commute, which means they can work longer hours (or enjoy more leisure). Somebody else comes up with another idea to make cars even cheaper, and roads start getting crowded. People spend more time on their daily commute than they used to, and they also do more damage to our ecosystems. Same thing for the internet. Wikipedia, Excel, and Ebay might have been productivity-enhancing at first, but now there’s facebook and free to play games, which are much more effective competing for/damaging our attention than the available productivity tools.

61 Doug April 22, 2015 at 1:54 pm

What if iPhones and their apps substitute for traditional (low-middle wage) service sector labor? It seems like a lot of low-skill native-born non-ghetto/hillbilly workers had a competitive advantage in jobs where their primary role was basic, understandable and polite communication. Call center workers, taxi dispatchers, hotel receptionists and concierges, the people who answer phones for takeout orders, catalog order takers, bank tellers, etc. Consumers increasingly prefer to deal with apps than actual people. Watch a millennial grown when he actually has to call in an order, rather than place it over the Internet.

62 KevinH April 22, 2015 at 2:02 pm

I think this is a good example of where linear and monotonic thinking in Econ for the last 50 years is preventing a comprehensive understanding of the problem.

Taking a big enough view and sure, $1B in lost wages is made up for by $1.1B in added productivity. At the same time take any individual and give them $100 less in pay but make the products they buy a total of $101 cheaper and they are better off. However, there are many intermediate scales where this benefit is not linear or monotonic.

Think about someone who is paying off debt. A initial hit in income has the possibility of never being beneficial at any future point if there if the rate of real wage growth to ‘trickle down’ to them is slower than the interest rate on their debt. Similarly, the fact that economic and social networks are not uniform or static mean (and in many cases depend on past economic success) that distributing losses and gains non-uniformly across the network and across time can have persistent effects.

The mathematical theory of ‘strange attractors’ would have plenty of example of this type of non-stable and quasi-stable behavior and would probably serve as a pretty good starting point for modeling behavior of multiple economic agents. Without our limited understanding of social-economic critical points, I think it’s a bit premature to say “But how can it not?”

63 edarniw April 22, 2015 at 2:20 pm

Suppose an entrepreneur introduces some new product, say the computer, which entirely displaces some other product, say the typewriter. Suppose further that the production of computers is more capital intensive than the production of typewriters. If the nominal flow of funds accruing to computers is the same as it was formerly for typewriters (which have been entirely displaced), then wouldn’t it be possible for real wages to have fallen?

64 Ryan April 22, 2015 at 2:28 pm

Don’t most business produce output based on demand? I don’t think it is a given that simply because a business can produce more, more efficiently, that they will. After reading Scott Winship, he indicates that what Tyler suggests is not a given. To me, it boils down to what the particular productivity gain is and whether it eliminates labor in the process. Using the iPhone as an example, mobile computing has certainly opened a lot of avenues for employment and business. But, not all labor is fungible, so there’s certainly a ‘catch-up period’ where existing labor is left in the dust or until it is retrained. In that interim, transition period, middle America might be better off with the new technology, but I don’t see any clear evidence that their wages/compensation would increase. That said, I don’t see the introduction of the iPhone as something that would _decrease **wages**_ either. It may decrease your overall household budget but not your wages per se.

To reduce what I’ve said into a sentence: If the productivity gain is effectively robots replacing humans, then unless new areas commerce lead to a new demand for labor, I don’t see the argument that it is a given middle-america **income** will increase. Standard of living, sure, but we’re talking compensation, right?

65 E. Harding April 22, 2015 at 3:28 pm

Agreed. The era in which computers most strongly contributed to RGDP growth was the Windows 95-Windows XP SP1a era. After that, diminishing returns set in and much of the rest has been simple growth in consumer surplus.

66 mulp April 22, 2015 at 2:37 pm

“Let’s say Ford can suddenly make cars — a known product — more cheaply. Even in a world with market power, this expands output and lowers car prices, thereby raising real wages. You might think labor won’t “get its fair share,” but still it should raise real wages, and that’s ignoring the possibility that wages in the auto plant might be bid up a bit.”

Let’s evaluate this from current free lunch economic principles.

Price defines value – if the price goes up on land, a house, art, it is more valuable.

Thus declining prices for these Fords means they are less valuable.

Given people can only drive on car at a time, lower value cars will not significantly increase sales of cars. Thus the auto industry will as a whole will tend to drive down GDP, which is measured in price, which means value, so the value of GDP declines.

To lower prices, wage income in the auto industry must decline, and thus workers are less valuable, and their output is less valuable.

So, how can real incomes rise when prices and value are declining.

Of course, the Ford management and marketeers will never agree that the new Fords are less valuable, so the prices stay high, but the higher productivity means labor income has been cut, so the value of labor has fallen while the value of the cars has not. But auto workers are no longer able to afford the price of Fords given their lower wages so they are not able to buy as much value in cars.

Unless the workers spend other people’s money.

Which is what the data shows over the past three decades. Worker debt is way up. The max 3 year car loan for only new cars is now 7 year car loans and 3 year loans are offered for used cars that in the 60s were cash only. Worker savings are down. Even after three decades of constant “you need to save for retirement” and all sorts of tax incentives and mandates, fewer workers today have savings for retirement than in 1980, and worse, workers nearing retirement have lots of debt on cars and housing when their peers circa 1980 were likely debt free.

Economics needs to be zero sum if it is science. For any system, every action has an equal reaction, every input is matched by an equal output.


HW might not have been able to state the problem, but he did correctly sense that the economic theory of Reagan was voodoo, not science.

67 E. Harding April 22, 2015 at 3:30 pm

mulp, you are an economic ignoramus who does not understand that RGDP is NGDP adjusted for price changes.

68 E. Harding April 22, 2015 at 3:31 pm

Ever heard of the real wage skyrocketing of 1873-1896, when the general price level was falling?

69 E. Harding April 22, 2015 at 3:32 pm

In the U.K. and Netherlands, not U.S.

70 JFA April 22, 2015 at 2:45 pm

So I read Scott Winship, and his analysis does not support Tyler’s claim of productivity growth always increasing middle class incomes. Winship explicitly says don’t look at the median, look at the average. But if we want to know what is happening to the middle class, the median (i.e. the middle) is exactly what we want to look at. Winship shows, with a few adjustments (not all of which are unreasonable), that wages are in line with productivity growth. But Winship notes repeatedly that it is the productivity/income growth for the supervisory/middle manager/upper part of the distribution that accounts for this. This does not show that overall productivity growth increases the income of the middle class.

71 dbeach April 22, 2015 at 3:06 pm

I had the exact same reaction. The Winship article shows, I think convincingly, that mean wages do in fact track mean productivity, while median wages do not. I see two possible explanations: either only the supervisory workers have become more productive and thus they “deserve” the higher wages while their underlings do not, or the supervisory workers are simply capturing everyone’s gains. The answer is not observable, I don’t think. But either way the median worker is not benefiting from productivity growth.

72 Brian Holtz April 24, 2015 at 12:15 pm

Where in Winship’s article (or anywhere else) are there data showing that for a given broad set of workers their average hourly compensation is diverging from their average hourly productivity? To say “the median worker is not benefiting from productivity growth” is uninteresting unless you can show that the median worker himself is the one whose productivity is growing.

73 PavJ April 22, 2015 at 2:57 pm

if the productivity growth is mostly happening in the realm of non-essential consumption while the costs of essentials like healthcare and real-estate (and education) continue to rise at a rate in excess of the lower prices on baubles and toys, then this would constitute a real loss in income.

And I think it can be argued that highly unequal income distribution ends up bidding up the prices on goods with limited short run supply (like real estate). Also, the gains from rent-seeking in insurance markets also don’t really erode the way they would for new IP. So it’s not hard to imagine how the middle class might feel increasingly squeezed despite better access to bread and circuses.

74 Floccina April 22, 2015 at 3:07 pm

Wouldn’t it be as simple as this:

1. Productivity rises through the production of some great expensive good that requires same some amount of platinum and so only the to 20 percent can ever afford it.
2. At the same time that the advantage goes to producers say due to population increase and so wages of the lower 80% of workers falls?

75 Zach April 22, 2015 at 3:21 pm

Baseball salaries before the abolition of the reserve clause would seem to be a counterexample. Increases in total revenue translated into increased wages very slowly, and there were periods where the major leagues were absorbing the minors where salaries had severe negative pressure.

The recent Silicon Valley wage fixing scandal seems like another sticky point. There a big increase in productivity led directly to a conspiracy to slow the increase in wages through noncompetitive hiring. If anything, the growth in productivity increased the return for collaborating with the cartel.

Once you allow for imperfect competition, imperfect information, and cartel behavior, the simpleminded equation of productivity and wages starts to be complicated. More money in the system increases incentives to cheat just as much as incentives to play fair, and a changing “fair” wage is harder for a worker to discover.

76 Silas Barta April 22, 2015 at 3:32 pm

Typical economic measures don’t typically capture the consumer surplus (and thus resulting “real wage gain”) of innovations like this. So you’ll see wages-as-measured stagnating, even though those same stagnant wages buy cooler stuff.

77 E. Harding April 22, 2015 at 3:55 pm

Yeah. I doubt, for example, that blogging and Google Books contribute much to GDP, but they certainly contribute much to consumer surplus.

78 Floccina April 22, 2015 at 3:41 pm

BTW does Krugman make the various mistakes pointed out by Scott Winship because he is not a labor economist? His work on trade was/is brilliant but should we expect his labor economics to be well done. Should we ignore him on labor issues not related to international trade?

79 dbeach April 22, 2015 at 4:34 pm

It’s not clear to me that they’re really mistakes, at least the mean/median problem. It can be true that mean wages track productivity, but if you care about the median worker, it’s not clear why that should matter. It just shows there’s a distributional problem, which is the point.

80 Floccina April 22, 2015 at 6:00 pm

The first error though, using household income verses individual seems like an error no labor economist would make.

Perhaps Mian and Sufi’s and Krugman’s most egregious no-no is that they suggest that annual family incomes should track productivity.

81 Bill April 22, 2015 at 3:53 pm

Let’s say Ford can suddenly make cars — a known product — more cheaply. Even in a world with market power, this expands output and lowers car prices, thereby raising real wages. You might think labor won’t “get its fair share,” but still it should raise real wages, and that’s ignoring the possibility that wages in the auto plant might be bid up a bit.

When’s the last time a major car manufacturer lowered prices? I think these lower car prices can only be posited through manipulation with hedonic deflators. Stick a touchscreen that would have cost a ton of money 10 years ago in a car console today and cars today are magically cheaper than they were 10 years ago, because they now have touchscreens. But this is completely arbitrary. You can’t get a V8 or even a V6 today without literally paying tens of thousands of dollars more today.

82 Floccina April 22, 2015 at 4:37 pm
83 Bill April 22, 2015 at 4:55 pm

That’s my point. Cars haven’t gotten cheaper. You can only impute lower prices by manipulation with hedonic deflators. You can’t get a V8 engine today without paying literally tens of thousands of dollars more than you had to decades ago.

84 Floccina April 22, 2015 at 5:58 pm

@Bill I may have read it wrong but do think that there were a couple of years where car prices fell.

85 Cooper April 23, 2015 at 12:32 am

you think cars are getting worse?

Middle class cars today can park themselves. They can access global radio stations. They are safer, stronger and more efficient than ever.

I think cars are one of the bright spots of tech innovation.

86 Merijn Knibbe April 22, 2015 at 4:34 pm

The British ONS tries to estimate public sector productivity. Their shot at this (and nobody should be fast to dismiss this: the people doing this are highly trained professionals who have given this subject a lot of thought, thre might however be subtle and not so subtle differences in the concept of productivity used by the ONS and the ‘common parlance’ concept):

“Estimated growth rates of total public service productivity are 2.5% in 2011 and 1.2% in 2012, the
two largest growth rates in productivity since the series began in 1997.

There are four consecutive years of total public service productivity growth from 2009 to 2012.
Average total public service productivity growth in 2011 and 2012 is 1.8% per year, compared to
an average of 0.4% per year in 2009 and 2010.

Positive productivity growth for total public services in 2011 and 2012 is driven by falling volume
of inputs, the first two falls in the volume of inputs since the series began in 1997.

Total public service output growth slowed down substantially from 4.0% in 2008 to 0.4% in 2011
and 0.1% in 2012, the smallest output growth rate in the series. Output includes an element of
quality adjustment for healthcare and education which is positive in all years except 2012.

Healthcare and education have the greatest combined share of estimated government
expenditure on public services at 54% in 2012; therefore, growth rates for healthcare and
education have the greatest influence on estimated growth rates for total public services.”


87 Jon April 22, 2015 at 4:35 pm

How about if a monopolist is able to create their product more cheaply (with fewer employees)?

For example, say that De Beers figures out how to mine the same quantity of diamonds for a lower cost. I wouldn’t expect that to have any real effect on prices or most consumers, but there would be some laid-off miners.

88 Floccina April 22, 2015 at 5:16 pm

Monopolists produce to where average cost equals price. DeBeers is a special case not only because it is a monopoly but also they sell a product that gets its value from being expensive (it is veblen good).

89 Miguel Madeira April 22, 2015 at 9:43 pm

“Monopolists produce to where average cost equals price.”


90 Floccina April 23, 2015 at 9:58 am
91 bobE April 22, 2015 at 4:43 pm

Can’t “productivity” be not real productivity? (Asi I mean Cowen’s first law.) Just as you say, dysfunctional productivity.(whatever that means) What are we measuring?

92 M April 22, 2015 at 5:46 pm

If your productivity increases by producing products that are a lot less durable and have to be replaced more often, then you could be standing still while productivity is still growing.

Or likewise if they have the same durability but are junked more often. Then you have a waste problem to deal with, but you’re not actually materially better off.

How do you economists get around that kind of measurement issue, where a good is being replaced by a lower quality good while maintaining the same market value? (which obviously happens all the time)

Likewise, how do you also measure the different “real” value of incommensurably different baskets of goods people have access to? (How many iPhones today is worth how many hamburgers ten years ago? etc.)

Productivity in terms of labour is also a per hour worked sort of rate thing? So if you’re producing the same amount but individuals employed work fewer hours, at higher and more stressful intensities, and get proportionately lower wages, then obviously people are worse off (albeit with more “time off”).

93 ThomasH April 22, 2015 at 6:18 pm

“Jared Bernstein worries that it won’t”

No, he does not. Bernstein said, “To be clear, faster productivity growth would be a highly welcomed development, and would surely help boost median incomes to some degree.”

94 Seth B. April 22, 2015 at 6:45 pm

Jeff Sach’s new robot paper answers this question

95 josh April 22, 2015 at 8:06 pm

Not having an iphone today is like being at a bar and not drinking. You are obviously much better off than the alcoholics (not everyone at the bar in the analogy is an alcoholic), though they don’t seem to notice. I imagine this is true for many of the things that are supposedly raising the “standard of living” for people who for some reason bread below replacement level. When in history has that ever been a sign of general prosperity?

note: I am not saying that the opposite is a sign of prosperity, but how confident are we that women who have 1 or zero children enjoy a high “standard of living”.

96 David B April 22, 2015 at 8:59 pm

The Scott Winship article is silly.

The argument that he makes is based off of a mathematical identity. Productivity = Income / Hours. He shows that, for the MEAN (not the median) of ALL workers (not just production workers), HOURLY wages are equal to productivity. Well, this is simply the definition. That doesn’t mean there’s not a problem!

The whole issue that concerns many lefties is that median income is not tracking productivity increases. This can be dryly stated as “all the productivity gains are going to workers with above-median annual income”. Now, is somewhat difficult for me to imagine that productivity improvements have simply not taken place in industries with median wages. With the advent of information technology, many or most jobs have changed dramatically and become significantly more productive.

Now a story wherein managers/programmers are capable of more thanks to information technology is not ridiculous to me–managing 12 people rather than 10 thanks to Outlook scheduling, or selling to a broader market. But the argument that management has become more and more adept at rent extraction due to weakened worker bargaining power–or perhaps substantially better at price discrimination due to improved analytics, or whatever other hypotheses are out there–these are not to be discarded out of hand. But I don’t think trotting out “mean hourly income = mean hourly income” graphs is really showing anybody much of anything.

97 Tom Warner April 22, 2015 at 9:17 pm

This seems very easy to answer: if the rewards for productivity gains accrue to others besides the middle class.

There are at least two obvious ways that can happen: other classes can gain market power, or redistribution can favor other classes. Getting into the details, the declining value of US citizenship and residency versus the growing power of globalized capital mostly explains why this has happened since the 1970s.

My question would be different: is the fact that the rewards of productivity growth have been accruing mainly to the capitalist class for the past four decades any reason to assume that will continue?

98 GC April 23, 2015 at 4:26 am

How about productivity gains coming from reducing human resources need?

Ford makes a cheaper car with new robotic technology that erases all its manpower needs.

Now, the worker could buy a much cheaper car, so his real salary would be up.. only, he doesn’t have a salary anymore. The he gets a fast food job (before that too gets shifted to robots) and his real AND nominal income is down.

99 Miguel Madeira April 23, 2015 at 5:13 am

Scott Winship.:

“But economic theory does not predict that when the economy’s productivity rises that the median worker’s pay should increase.”

Case closed.

100 AcidDC April 23, 2015 at 9:18 am

Productivity growth can raise incomes by making certain products less expensive, but that’s not really the same as raising middle class incomes. Cheaper iPhones are great, but even if they were free, that wouldn’t solve any of the issues the middle class is having with stagnant wages and rising housing, health, and education prices.

101 Denis Drew April 23, 2015 at 9:50 am

It’s nice to know that incomes in the poorest parts of Africa and Asia have gone up — because everyone there now has a cell phone. Doesn’t do much for the other aspects of $2 a day existence.

Doesn’t do much for a minimum wage several dollars below its 1968 high here — double the per capita income later. Doesn’t do much about the 1% share of income growing from 8% to 25% while bottom 50% share shrinks. Wont stop the 1% share of income from going to 75% out of 150% of today’s income a few decades hence.

102 prior_approval April 23, 2015 at 1:58 pm

98 comments, and not a single mention of how growth based on financial engineering productivity has nothing to do increasing middle class incomes?

103 louis April 23, 2015 at 5:12 pm

Easy answer: Cowen is likely to be right as long as demand is equal to potential output.
In a recession, increased productivity increases potential output, but because there is still slack in the labor market, and no guarantee that demand will increase to match the increased potential supply, the labor share could be worse off on net. Worker bargaining power is eroded, and profits increase due to falling labor costs. Sound familiar?
Listen to Delong and make Say’s Law true in practice and productivity gains are a Good Thing again.

104 Mark Cancellieri April 23, 2015 at 8:51 pm

The real issue isn’t whether there is a disconnect between productivity and real wages, but rather productivity and total real labor costs.

It is very possible to experience productivity increases and experience weak real wage gains if labor costs other than wages grow rapidly. One good example is the ObamaCare health insurance mandate. Another example is payroll taxes. Another example is labor regulations. And so on.

All these non-wage labor costs cause employers to bid lower on wages in labor markets, which is really too bad, because the vast majority of Americans have been hoodwinked into thinking that they are getting a free lunch, especially when it comes to “free” health insurance from their employers.

Comments on this entry are closed.

Previous post:

Next post: