The cost disease of services

Suppose there are two industries in a nation.  One has rapidly rising productivity growth and the other has stagnant productivity growth.  Which one would tend to grow and see rising wages?  Ironically, the industry with stagnant productivity will probably grow and wage will rise in both!

This odd tale has really happened.  Manufacturing (and agricultural) productivity has skyrocketed over the past two centuries whereas the productivity of services has been excruciatingly slow or even nonexistent.  As a result, the percent of the economy devoted the manufacturing and agriculture has plummeted and been replaced by that laggard, services.  Meanwhile wages in services have risen just as fast as in manufacturing and agriculture, so there is zero correlation between productivity growth and wage growth!  It doesn’t seem fair.

Derek Thompson at The Atlantic examined this phenomenon:

[The] National Journal [had a] special report on the rise and fall and rise of manufacturing. The spectacular graphic compares employment by sector in 1947 and 2007 and its most important lesson is a whopper. Manufacturing and agriculture employed one in three workers just after World War II. Today, those sectors employ only one in eight.

Where did all the making-stuff and growing-stuff jobs go? They went into services.

…The big story about American jobs in the post-war period is this: The manufacturing/agriculture economy shrunk from 33% to 12%, and the services economy grew from 24% to 50% [plus government, food services, energy and mining, and construction have all stayed about the same]. …as manufacturing and agriculture got more efficient, they required fewer American workers, while the services industry (which had slower efficiency gains since it has more person-to-person work) required more employees to keep up with the rising demand for consulting, nurses, teachers, computer technicians, and so on. This isn’t a sad story, or a happy story. It’s just what’s happening…

Closing thought: Why isn’t anybody talking about the tragic decline of agriculture? [Agriculture’s] share of workers has fallen by 80 percent in the last 60 years. Nobody seems to think that’s much of a tragedy, but we do consider it tragic that manufacturing has lost 60 percent of its share over the same period. Are we being hyperbolic about the decline of manufacturing…?

…Manufacturing jobs have declined as a share of the economy. But manufacturing hasn’t declined as an industry. It’s grown. By a lot. …Output has sextupled.

Here is the data from FRED. Manufacturing production (blue line) has grown about 150% since 1975 even with dips during recessions.  Manufacturing employment (red line) was about 50% bigger at its peak in 1979.  So American manufacturers are producing a lot more output with a lot fewer workers.

American manufacturers can afford expensive American workers because their productivity keeps rising. In 1960, labor costs in manufacturing typically accounted for around 30% of total manufacturing costs; by 2000, they were generally down to 12-15%.

Farming is a similar story.  In nearly all developed nations, farm production is three times bigger than in 1950 and the number of farm workers has plummeted to the point where only about 1% of American workers is in agriculture and they only contribute about 1% of GDP.  Whereas productivity in making stuff has skyrocketed, the productivity of producing services like psychotherapy, live concerts, education, and church services has not risen and the relative scarcity of services compared with the abundance of low-cost stuff at our disposal (literally) has caused the relative cost of services to soar.  This is often called, “the cost disease of services.

To understand the cost disease, start with a simple observation: whatever the economy’s average rate of productivity growth, some industries outpace others. Take car manufacturing. In 1913 Ford introduced assembly lines to move cars between workstations. This allowed workers, and their tools, to stay in one place, which cut the time to build a Model T car from 12 hours to less than two. As output per worker grows in such “progressive” sectors, firms can afford to increase wages.

In some sectors of the economy, however, such productivity gains are much harder to come by—if not impossible. Performing a Mozart quartet takes just as long in 2012 as it did in the late 18th century. Mr Baumol calls industries in which productivity growth is low or even non-existent “stagnant”.

Employers in such sectors face a problem: they also need to increase their wages so workers don’t defect. The result is that, although output per worker rises only slowly or not at all, wages go up as fast as they do in the rest of the economy. As the costs of production in stagnant sectors rise, firms are forced to raise prices. These increases are faster than those in sectors where productivity is improving, and faster than inflation (which blends together all the prices in the economy). So prices of goods from stagnant sectors must rise in real terms. Hence “cost disease”.

The cost disease of services does not mean that services have become unaffordable.  Much to the contrary, services are more abundant today than ever before because 80% of Americans now work in services whereas 200 years ago it was closer to only 20%.  However, the opportunity cost of buying services is much, greater today than in the past.  Whereas 200 years ago, a simple shirt might have been worth as much as a week of college education, today a week of college is worth many, many shirts.  Education has always been expensive, but everything used to be expensive and today we live in a world of material abundance where services still remain expensive despite the fact that they are more abundant than ever before.

The real price of services is rising because the real value of labor is rising which means that workers can still afford about the same amount of services as ever before.  In other words, when workers buy services, we are still giving up about the same amount of work time as in the past on average.

Suppose there is an economy that only produces hammers and therapy.  In a society that requires 40 hours to produce a hammer, each hammer will cost about 40 hours of therapy and an hour of therapy will be worth 1/40th of a hammer.   When manufacturing productivity increases to 1 hammer/hour, there will be a cost disease of services because an hour of therapy will rise 40-fold to cost a full hammer per hour.  Because people don’t need lots more hammers when their price drops (demand is inelastic) and therapy is a normal good that we will buy more of as incomes rise, people end up buying a lot more therapy despite the massive rise in its opportunity cost.

This basic story actually happened for manufactured goods and food versus services. Just substitute physical goods for hammers in the above example and substitute services for therapy and you have the basic economic history of the past two centuries.

Thus the cost-disease of services cannot make services less affordable than in the past because they become more abundant.  They only seem less affordable in comparison to their opportunity cost.  THE reason services are rising in cost is because wages are rising in cost and that means we will always be able to afford services.  The cost-disease is a very good thing because it is really just an odd way of talking about rising wages.  Inequality will still be a problem as some wages rise more than others, but the “cost disease” of wages is a very good thing.  The cost “disease” simply means that as goods become more abundant, we can have more goods and more services too, but that services will seem more expensive by comparison with goods.

The real problem with the cost disease of services is that productivity growth in services is so slow. As more and more of the economy is devoted to this sector with slow productivity growth, overall productivity growth will also slow down and this has been happening.  Slow productivity in services (>80% of the economy) is slowing overall productivity growth in rich countries and this will also mean slower income growth unless we can find a way to boost productivity in services as well as in manufacturing.

As of 2021:

  • 1.7% of Americans produces all our food, agricultural goods, mining and forestry.
  • 7.8% of American workers produce all of our manufactured goods
  • 4.7% of workers do all of our construction
  • and 5.8% of Americans are self-employed in nonagricultural work which is also mostly services and construction.

So altogether, only about 9.5% of American workers produced all of our physical goods (other than construction) in 2021!  The other 90% of us have been working in low-productivity-growth jobs.  Unfortunately, productivity growth in construction has not been much better than in services, and that is one reason why housing is not getting a lot cheaper like manufactured goods are.

The “cost disease” also explains a big reason why all industrialized nations have seen a major growth of government — the government mostly produces services!  So whenever there is a big growth of private-sector services, you should expect to see a similar growth of government services

The above graph omits data for France and Germany during the world-war years or they would have had a huge spike then too.  Another distortion in the graph’s data is that US spending appears smaller than the reality because this graph does not include state and local government expenditure.  That is much bigger in the US federal system than in most countries because government is much more centralized in smaller countries. There is no point having a layer of state government in any place where the entire country is smaller than the state of California, so in Europe, the federal government also fills the roles that state governments are in charge of in the US.  

I am not worried about the “cost disease” of services for wellbeing because it means that people can afford more services and more goods as long as rising inequality doesn’t make some essential services, like healthcare, unaffordable for vulnerable people…  Oops, that IS a problem! 

In addition to the inequality problem being a real disease, we have had a major problem with the cost disease of housing because whereas the cost of almost all physical goods has been plummeting in the US, the cost of housing is the one major good that has been going up in cost.  This is one of the biggest economic problems in America because unlike food and clothing, housing is a major expense for every American family–it is the single biggest expense most households have.  Marketplace reports that, “the average U.S. renter now spends 30% of their income on rent, a new all-time high” since Moody’s started collecting the data, two decades ago.  In New York City, renters pay more than 68% of income, which is the highest in the nation. 

Housing cost is the root cause that explains why there is much more homelessness in high-cost places like California and New York than in lower-cost places like Texas and Illinois.  

Other potential reasons for homelessness have very, very little effect.  Even if you could eliminate mental illness, drugs, poverty, geography (nice weather), and generous welfare programs, you would still have a massive homeless problem in places like New York, Hawaii, and California, where rents are very high. 

In poor countries today (and rich countries like the US back when they were poor), most people spend the vast majority of their income on food and clothing. In 2021, Kenyans still spent more than half of their income just on food alone and Americans were almost burdened that much in 1900 with 57% of expenditures devoted to food and clothing:

Thanks to the dramatic increases in productivity, the cost of food and clothing have plummeted and American households now spend less than 13% of their income on food and clothing despite a large rise in expenditures on eating out which is now greater than the amount spent eating at home.  Food and clothing are the two categories where increasing productivity produced the biggest source of increased economic welfare* because they are necessities.  But housing is also a necessity and America is getting worse and worse at building housing and other infrastructure.  Because Japan is able to build lots of new housing, housing in Japanese cities is more affordable than in US cities despite the fact that Japan has a much higher population density than the US.  And Japan is building lots of new housing despite the fact that Japan doesn’t need new housing as much as the US because Japan’s population is shrinking.

So forget the supposed cost disease of services.  That is a good thing caused by rising productivity in most goods.  The real cost-disease problem in the US is the cost disease of housing.  


*Increasing productivity in public health also created a tremendous increase in welfare, but that wasn’t mainly just due to a drop in price as is the case for food and clothing.  In healthcare, it was the invention of completely novel technologies that have doubled lifespan over the past two centuries. 

Posted in Globalization & International

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 92 other subscribers
Blog Archive