Is the Cost Disease Dead?

Even though William Baumol didn’t win the Nobel prize this year it got me to thinking about the cost disease, as did the death last week of William Bowen, the co-author of Performing Arts – The Economic Dilemma which brought the cost disease to public attention. The cost disease says that if two sectors have unequal levels of productivity growth then the sector with lower growth will increase in relative price. If in 1900, for example, it took 1 day of labor to produce one A good and 1 day of labor to produce one B good then the goods will trade 1:1. Now suppose that by 2000 1 unit of labor can produce 10 units of A but still only one unit of B. Now the goods trade 10:1. In other words, in 1900 the price or opportunity cost of one B was one A but in 2000 to get one B you must give up 10 A. B goods have become much more expensive.

The cost disease says only that the relative price of the low productivity good increases, it doesn’t say that the low productivity good becomes absolutely more expensive. The economy in 2000 is much wealthier than in 1900 so relative to income B has become cheaper. Anyone who could consume x units of B in 1900 can still consume x units of B in 2000, the only difference is that in 2000 they will be giving up more A than in 1900 so the tradeoff has become steeper even though still affordable.

Stated generically the cost disease is indisputable. But it becomes more contentious when we try to identify the A and B good. Baumol and Baumol and Bowen initially pointed to labor intensive goods, the service sector, as the low-productivity B sector. The performing arts were the key example–it took four quartet players 40 minutes to perform a Mozart composition in 1900 (or 1800) and it took four quartet players 40 minutes to perform a Mozart composition in 2000, hence no productivity improvements in Mozart performances, hence a rising cost over time since those four players could produce many more goods in say the manufacturing sector in 2000 than 1900. Health care and education are other stock examples.

Tyler offers one response to the cost disease namely that it’s true if you define the good narrowly (listening to a live performance of a 40 minute Mozart composition) but why should we define the good narrowly? If instead we define the good as “listen to music for 40 minutes” then it’s clear that costs have fallen dramatically. Not only has the cost of listening fallen, variety has increased. Costs have fallen even further since Tyler wrote. In a similar way, Tyler and I have argued that online education greater lowers costs and increases quality.

robot-playerHere, however, I offer a different and more fundamental response. Baumol pointed to labor and the service sector as the low productivity, low growth, sector. But robots and artificial intelligence mean that there is no longer a pure “labor” sector. Robots are labor made of capital. Whether we are talking about robot vacuum cleaners, AI answering machines or Dr. Watson there is much more capital in the service sector than ever before. K has become L. And when K becomes L, the productivity of L increases with the productivity of K. If manufacturing productivity improves and we are manufacturing robots then any sector that uses robots increases in productivity. If software productivity improves–if AI becomes more intelligent, for example–then any sector that uses AI increases in productivity. Any service that uses information technology inherits all the productivity growth of information technology.

At any moment there will always be some sectors that are increasing in productivity at a faster rate than other sectors–that is the nature of progress, uneven and episodic–but the time when one could distinguish a manufacturing sector and a service sector and argue that as a general rule the latter increases in productivity at a slower rate than the former is rapidly coming to a close. K has become L.

Addendum: Timothy Lee also has a piece today on the cost-disease.


... and just what is the practical value of knowledge about this "Cost Disease" ?

If it is not yet 'dead' how might it be usefully applied in the real world?

But increased productivity of labor (from digitizations, AI, etc.) has reduced the share of income going to labor, a phenomenon pronounced in the arts and soon likely in the academy. Indeed, for musicians live concerts have become the dominant means of producing income. For professors, they will compete with digital replicas of themselves for income. Ritual reality competing against actual reality for a share of income.

Virtual reality.

I kinda liked the "Ritual reality".

"...has reduced the share of income going to labor.."
Has it? Or has it been redirected to other industries and that labor. Has income been reduced, overall, to labor?
When farming moved from having millions of people in he fields, or 70% of the population, in the two first decades of the 20th century, to less than 2% by 1990, did less income go to labor or has it moved to labor that produces and services the farming equipment, the increased manufacturing of raw materials into the products for producing the farm equipment, to other industries that has been made possible due to less resources being dedicated to the growing o food?
Maybe I misunderstood your statement. I'm not arguing. I am just trying to better understand your statement.

Yes, it has declined. See, e.g.

Actually, the decrease in labor in the agricultural industry has moved to other industries to such a great extent it changed the U.S. from an agrarian economy to an industrial economy. The same this has been happening again in the manufacturing industry. Decreases in labor have moved to the informational industry and once again our economy adjusts. There is some disparity on if it has occurred yet, but we either are in or are headed towards an informational economy. Think about how much wealth today is generated by the manipulation of information as opposed to manufacturing. This is why the argument about manufacturing jobs leaving the country (being a bad thing) is really moot.

Remember though, labor comes at the expense of leisure.

Ideally zero income would go to labor and no labor would have to be performed to satisfy all utility. This still sounds crazy today, but considerably less so than in 1816.

I think people's worry is that if you reduce how much you need to spend on Wal-Mart crap by $X, then you just bid up the price of your mortgage by $X, so that you are buying the same amount of Wal-Mart crap and living in the same house as before, nothing about your life has changed, even though some prices changed.

Key to this is the idea that:
1) There exist necessary goods people have to have regardless of price (real estate, healthcare, education credentials).
2) Those goods don't increase in supply when price increases, or increase only in ways that aren't actually useful (more but useless healthcare for example).
3) In such a scenario all dollars freed up by robots making Wal-Mart stuff goes straight into bidding up these items.

At least that is the only version of this that would be worth talking about.

But, wouldn't the price increase then direct home suppliers to increase supply, thereby putting downward pressure on the housing market?

They could increase supply, or they could increase "quality" to compete for buyers. I'd argue the latter is much more profitable. This may not work in the rural areas, but here in DC every new housing development is "luxury". I want to live close to work/transportation, but I don't even have the option to save money by going non-granite counter tops, etc.

The unfortunate problem in increasing supply, in an area like DC, are the restrictions and enormous costs in increasing supply, such as height restrictions, rent controls, and expensive permit/political processes.
Less cumbersome, laced with red tape, and less expensive is to upgrade existing property.

The building cost isn't the bottleneck though.

I think what happens is, when you try to increase supply, you have to locate the new houses near the desirable locations, where the economic activity takes place. But all the buildable locations are no good because you still have to drive along the choked roads to get anywhere good, eliminating the utility that can be gained by adding more that way.

So it ends up that you just have an arms race for the available home plots that eats up the excess wealth.

(I know the common response is that "well issue more permits" but I think this can't be fixed until fundamental infra is improved.)

While not denying zoning can be an issue, I have discussed at length how it either isn't the main issue or is not an issue anyone has offered a realistic solution to (besides whining).

To clarify, I was agreeing that it's not a simple matter of "let them build more".

My gut feeling is that it wouldn't pan out, after considering all factors quite the way you expect. The effect is probably non-zero and positive (i.e., I'd guess you're right, but maybe the effect isn't that large).

The concern might be that if this goes hand in hand with arguments for raising protectionist trade barriers, that this is part and parcel with arguments about trade, where if we went the Trump route, there are very few economists who would predict overall benefits for Americans, with the negative effect increasing over time due to lack of exposure to competition and a higher ability to be complacent about productivity due to reduce trade exposure.

Also, I think the Walmart crowd might not be bidding up mortgages all that much. But perhaps the same logic applies in the top 10-20% who might not often shop at Walmart but whose savings from technology and cost reductions has fewer MORE places to go (poorer people might more likely buy something else with their savings) and thus bid up housing prices.

All the ways in which things that look good at first sight lose their lustre when considering how they contribute to bidding up fixed asset prices, especially in real estate. At least with real estate, it shouldn't be too difficult to keep revenues higher than costs so you'll always be earning a positive return.

"Wal-Mart crap"

Its largely not "crap", its food and household products that people need at reasonable prices.

Wal-Mart and Target products overlap by 90% or so. Target may have slightly better clothes quality, (maybe, they buy from the same third world places as Wal-Mart) and a better presentation. Otherwise, little difference.

I suggest you just do not like the Wal-Mart customer base.

>1) There exist necessary goods people have to have regardless of price (real estate, healthcare, education credentials)

You don't think demand curves slope downward across real estate markets? I have to live somewhere, but when I lived in New York I "had to" have a 600 square foot apartment but now that I live north of Chicago, I "have to" have a 2200 square foot house. Roughly the same with health care and education credentials.

"The performing arts were the key example–it took four quartet players 40 minutes to perform a Mozart composition in 1900 (or 1800) and it took four quartet players 40 minutes to perform a Mozart composition in 2000, hence no productivity improvements in Mozart performances, hence a rising cost over time since those four players could produce many more goods in say the manufacturing sector in 2000 than 1900."
If they could get a job producing goods in manufacture, which they can't unless they are called Li, Wang, Zhang and Tao.

The Ramones were working on improving productivity in the arts:

But the noise prevents me from hearing the words.

How does the history of mobile phone service fit into this scenario? The widespread adoption of cell phones may have been one of the most rapid acceptances of new technology ever. It wasn't cheap. Cell phone contracts drained money away from other consumer purchases, probably had a negligible effect on productivity and may well have been a significant cause of the Great Recession, since cell phone service quickly became more important than transportation and even housing to many people.

Baumol's cost disease works well in the abstract, but in practice desires are flexible, adaptable, sometimes even perverse and fashion oriented.

Not only did poorish people need a smartphone, they needed a premium iPhone.

It is a mature market now. But what was the characteristic when it was growing?

I'd suggest it was (and still is) productivity advancing. When we got our first cell phone we no longer needed someone answering the phones and dispatching. We no longer needed to go to the office to phone for orders or tracking down information. And as data was included the standard email and web improved communication, as well as made possible vertical applications that really enhance productivity. What makes an enormous difference for me is the ability to take a quick shot of what I'm working on, maybe a few with all the details, and have it available anywhere or easily sent to someone else. No more transcribing numbers while standing on my head trying to see through my bifocals while reversing what I see in an inspection mirror.

The bread and butter services and large bills that people like me paid over the years were the cash flow upon which to build out the networks.

The consumer usage is mostly entertainment. But it has replaced land line phones. Does anyone get one of those anymore?

> probably had a negligible effect on productivity

Yeah, no.

The productivity gains associated with this one are hard to quantify, but seem to me far greater than costs.

Though certainly I see many people using phones and plans that are far more expensive than they need, don't really get it. A $60 smartphone and $30 monthly contract suits me fine, and I have a significant income surplus.

Check out these two charts at this link:
Chart 3 (Services keep getting more expensive while manufactured goods get cheaper)
Looking at this are we wasting our time trying to create more inflation and should we just let the price level deflate?

Chart 10 (Average workers' wages have been falling far behind productivity growth): What caused the abrupt change in trend of worker's compensation in 1973?
Here's a good discussion:

Well, then what's *your* take on the question:
What caused the abrupt change in trend of worker’s compensation in 1973?

"Now the goods trade 10:1. In other words, in 1900 the price or opportunity cost of one B was one A but in 2000 the to get one B you must give up 10 A. B goods have become much more expensive." That seems to be something of an equivocation -- in the abstract the exchange is still a unit of labor in one good for a unit of labor in the other good. Market exchange ratios have changed but it's not really clear opportunity cost or real cost have changed at all.

"K has become L."

I prefer to think of it as: L has become K.

Nay, L has always been K.

Hardwood lumber is expensive, computers are cheap.

I'm not sure that is cost disease exactly, or that you can't grow a big tree faster, no matter tech advancement. Also no one wants a dinner table the size of a Raspberry Pi.

In the case of hardwood, the cost comes from tress being scarce compared to demand not the technology involved in cutting + transport.

For the size of the dinner table, it's the commodification of restaurants. When I grew up, going to a restaurant was a once a month luxury activity for the middle class income of my parents. It would be an interesting exercise to know what would you get today with the money parents paid in restaurant bills many years ago? Perhaps the problem is modern cheap restaurants, not restaurants per se.

I think that what I was saying, yet another way to say it is that the ratio of old trees to consumers has fallen off a cliff. That's obvious, but I still can't wrap my head around a computer "faster than a Cray" costing less than one board foot of lumber.

Restaurants are enjoying an efficiency advantage over home cooking.

We are going out to ? for breakfast, no one cooks ? at home. Hours at preparing a good stock? No one has time for that.

Spell check suggested a emoticon, ? for "pho" and I took it.

So the size of kitchens in modern houses is to be viewed as something of a luxury consumption good rather then from a production view -- otherwise that might be just wasted investment in the house (though I suspect one could argue the more modern open designs make the kitchen much less of a place to produce a meal and more of a place to congregate so it's the invasion of the living and family rooms into that space).

"It took four quartet players 40 minutes to perform a Mozart composition in 1900 (or 1800) and it took four quartet players 40 minutes to perform a Mozart composition in 2000, hence no productivity improvements in Mozart performances"

I don't think that even this is true. Improvements in the design of concert halls (and amplification, though that offends some purists) mean that the number of consumers that can simultaneously consume a live performance has gone up significantly over time.

I pay my Russian-Chinese barber much more for a haircut than he got in the old country, but he's doing the exact same work. Baumol's disease holds - until a threshold of automation/replication is reached (specific to each industry) which transforms the old "overpriced" service into the luxury consumption of human craftsmanship and relationship.

Cost disease is not dead at all. The better question is "How does the labor force equilibrium evolve in response to increasing automation?" On average, workers displaced from some sectors via capital substitution will have to take up jobs which, by their nature, are difficult to automate cheaply. If you think about it, you'll realize that these jobs will necessarily have a strong Baumol-like character, because they are relying on human factors which, for the median worker, can't be sped up. And, when more and more people are doing services which aren't producing any more output per man-hour as time goes by, then average labor productivity growth slows to a stagnating crawl. We are all living in Baumol's world now.

As incomes rise, status and positional goods will matter more. Technology cannot change this. It can only provide low quality alternatives (e.g. CDs vs. listening to live orchestras or nice wired houses in the middle of nowhere vs in the nicer parts of the most desirable citiess). So we will see more and more attempts to provide improvements through technology while allowing the base service to change.

Ask a frequent business flyer if any change in technology can compensate for the deterioration in onboard and terminal service relative to a period like the 1970s. Ask someone pushed out of a nice neighborhood due to rising costs if staying in a modern flat in an area with bad schools (i.e. bad pupils) is compensation.

Heck, just look at the willingness of elites to not have children. Absent the right environment, they'd rather go childless and have their line ended for good. Nothing is more costly (in a Darwinian sense) than that.

>Ask a frequent business flyer....

OK, you asked. First class now costs about the same as coach did in the 1970s in constant dollars, so I'm actually better off - and I don't have to fly as often because I get a lot of work done over the Internet.

Yes, excluding the original example, where people who go to live performances go because they want to experience a live performance as much as the ostensible product of the performance. Which is one reason that ticket sales for live performances have gone through the roof in recent years.

Childcare suffers from Baumol's Cost Disease even worse than healthcare. Many parents would happily leave their children at home in the care of a robot, if such a robot existed; but since the government forbids it, the robot will never be made.

(In Peter Pan, the children's nanny is a dog. J.M. Barrie was clearly familiar with the cost of childcare.)

The logistics or organizing a 40 minute Mozart quartet performance are orders of magnitude easier.

1) 1 minute: Client calls/emails person responsible for quartet schedule (whether or not in the quarter) to get quote and check schedule.
2) 5 minutes: That person calls/emails each person in the quartet to check schedule.
3) 1 minutes: The go-to person calls/emails the client to confirm.
4) On the day of the gig, each musician drives/buses to the venue.

Compare this to 1850.

I'd say this is hands down the dumbest thing I'll read on the internet today, but it's only 9:40 on the west coast and there's no rule against topping your own record.

OK, go find yourself a quartet without a phone or the internet, and arrange for them to come to the event in a way that does not include cars or buses.

What is the time difference, even if everything is perfectly and easily available for you at a free price?

Alex: You are misunderstanding or misrepresenting what Baumol/Bowen mean in the context of the performing arts. The applicable "cost" is that borne by the performers, including the sunk cost of learning to play the instruments well enough to perform Mozart in the first place. Including recordings in the argument is rhetorical slight-of-hand because we could say there is no reason to re-record these works and therefore the marginal cost is zero. It's irrelevant to the cost disease concept whether it is easier for consumers to enjoy the performing arts in 2016 as opposed to 1791. The point of the cost disease argument is limited to examining the costs underlying individual performers preparing and presenting a live concert of a work today.

Except that 1) my productivity has increased and so the opportunity cost of my time (uncorrected for income effect)
and 2) the main cost of listening to music is my time (see 1).

Music today seems to be a poor example. Better to look at education and health care. Looking at the CPI we have 241 for all items while healthcare services are at 499. A chunk of that is probably cost disease. Tuition, other school fees and childcare is 716. A chunk of that is also probably cost disease. At the same time, Educational books and supplies is at 693. I suspect that most of that is from the textbook cartel rather than cost disease since recreational books is at 99. (1997 base versus 1983 base.)

Even without artificial intelligence, the path to increased productivity in many services is plain. I got vaccinated for travel today. The actual shots took about four minutes. The pharmacist typing furiously at a keyboard took forty five. Wouldn't take that much business process improvement for a several hundred percent increase in productivity.

I'm sorry you have an idiot for a pharmacist. That is all too common, and says something about the quality of labor when an industry expands too quickly. However, I got my travel shots done in about 5 minutes at the doctor's office where the employees are screened by a doctor for their ability to perform basic tasks.

Sounds like a much better experience! No doubt it would go faster if I (or my insurance carrier) had paid more for the more personalized in-office attention. But more core to the productivity point is that it isn't difficult to visualize a system where the employee need not be bright, have much specialized training, and can handle a high number of patients. For example, I walk in, say I'm going to X country, they're prompted with a couple questions (e.g., are you going to a rural area), they scan my insurance card, the machine in the back drops out the three syringes, I pay, they give the shots. Might not be easy to actually accomplish, but easy to visualize.

Eventually biology will also become information technology driven, as we master the language of proteins.

Increasingly, the question of utility and in whose eyes it is beheld becomes paramount.

Comments for this post are closed