The Cost Disease: Why Some Prices Never Stop Rising

The Feeling You Have Never Had Explained

You have a feeling about prices that nobody has ever explained to you properly. The things you need keep getting dearer. The things you buy for fun keep getting cheaper. A hospital bill, a semester of college, a day of child care, a plumber, a haircut - up, every year, forever. A television, a laptop, a pair of sneakers, a flight to see your sister - down, or flat enough that you notice the difference.

That is not an impression. Here is the Bureau of Labor Statistics price index for the American city average, comparing annual averages in 2000 with annual averages in 2025:

Bureau of Labor Statistics, CPI-U, U.S. city average, annual averages (not seasonally adjusted): all items, televisions, new vehicles, college tuition and fees, hospital services. Percent changes are calculated from those annual averages.

Same country, same dollars, same quarter century. Prices in general rose 87 percent - and one of these items fell by 98 percent while another tripled. What you are looking at is not one inflation rate doing its work. It is the price list inside the economy rearranging itself, at speeds that differ by a factor of two hundred.

The explanations you get handed do not survive that table. “Greed” does not survive it, because greed would have to be stronger in hospitals than in television factories every single year for sixty years, in every rich country at once. “Inflation” does not survive it, because inflation is one number moving everything together, and here things move in different directions.

There is a real explanation. Two economists worked it out in 1966 while trying to understand why theatres keep going broke, and once you see it you cannot stop seeing it.

The Four Musicians

In 1966 William Baumol and William Bowen published a study of the live performing arts called Performing Arts: The Economic Dilemma. They wanted to know why orchestras, opera houses and theatres were permanently short of money even when audiences filled the seats. The answer turned out to have nothing to do with the arts.

Take a string quartet. Four players, one late Beethoven quartet, about thirty-five minutes of music. In 1800, four musicians playing that piece needed thirty-five minutes. Today, four musicians playing that piece need the same thirty-five minutes. Better lighting, better instruments, electronic ticketing, streaming recordings - none of it changes what you bought, which is four people playing for half an hour.

Output per hour of work is fixed. In a television factory it is not: the same hours produce more sets, and better ones, every decade.

Now the arithmetic. The labor cost of one unit equals the wage divided by the units produced per hour.

Start with the factory. A worker makes 20 units an hour at $10 an hour, so labor costs 50 cents per unit. Ten years later productivity is up 20 percent and wages are up 20 percent: 24 units an hour at $12. Labor still costs 50 cents per unit. The worker got a real raise and not one cent of it went into the price, because the productivity gain paid for it.

Now the quartet. The musicians’ output per hour does not move. When their wages rise 20 percent, the labor cost of a performance rises 20 percent. Every cent of the raise lands in the ticket price, because there is no productivity gain to absorb it.

And the raise happens whether the quartet can afford it or not. No symphony orchestra pays 1960s rates in a 2020s labor market; the cellist has other offers. That is the part people miss: the musicians are not paid more because they got better. They are paid more because everyone else got more productive, and the orchestra has to compete for people in a labor market that productivity has lifted.

Baumol and Bowen saw what this meant for any trade where the work is the product:

What does that mean? In most industries, labor is a means to produce a good, so a machine that makes the good cheaper is welcome. In some industries the labor is the good. You cannot automate the nurse’s hour in the room, the surgeon’s hands, the aide who helps your father shower, the teacher standing in front of thirty students, the electrician in your attic, the person cutting your hair. Nothing about the technology of the last fifty years has changed how long any of those takes.

Why This Is Not a Curiosity

Baumol and Bowen did the empirical work as well as the theory. They pieced together cost records going back to the 1770s for Covent Garden and found that the cost of putting on a performance had risen a little over twice as fast as the general price level. For the years after the Second World War they gathered data on 23 major American orchestras, three opera companies, one dance company, and a sample of Broadway, regional and summer theatres. In every group the same thing happened: cost per performance rose far faster than prices in general.

That was 1966, and the finding has not aged. The table at the top of this piece is the same mechanism, running for another sixty years, in sectors nobody would call artsy.

Health care. Spending rose from 5.0 percent of GDP in 1960 to 18.0 percent in 2024, according to the Centers for Medicare and Medicaid Services, whose Office of the Actuary compiles the National Health Expenditure Accounts. That is $5.3 trillion, or $15,474 per person. The same office projects 20.6 percent of GDP by 2034.

College. Tuition and fees rose 188 percent against 87 percent for everything else since 2000. The work is a professor’s attention, delivered to a room of people, and a fifteen-student seminar is not the same product as a fifty-student lecture.

Child care, care homes, repair work, dentistry, restaurants. Every one of them is somebody’s time, sold by the hour, with no machine coming to make the hour shorter.

Notice what the pattern does to the political argument. If the rising cost of care were a story about bad actors, you would expect the sectors that got cheaper to be the ones behaving well. They are not. Television manufacturing has been through periods of brutal concentration, and plenty of the firms that got rich making the goods that now cost nothing were no gentler than anyone else. What changed their prices was not virtue. It was that a machine could do more of the work every year.

What may be surprising, though, is that the disease is a symptom of success. The quartet’s costs rise because productivity somewhere else raised the wages it must compete with. A stagnant economy has no cost disease, because nothing pulls wages up. This is the productivity gain going to consumers and the wage pressure going to whoever cannot mechanize. The mechanism that makes a television cost $200 is the same mechanism that makes a nurse expensive.

The Part That Is a Choice

The cost disease explains a floor, not a ceiling. On top of it, three of these sectors are paid for by somebody other than the person receiving the service, and that changes what happens next.

• The patient does not see the price. In 2022, employers and private insurers paid hospitals an average of 254 percent of what Medicare would have paid for the same services at the same facilities, according to RAND’s hospital price transparency study (Whaley, Kerber, Wang and Kofner), which re-priced $77.4 billion of spending across more than 4,000 hospitals using twelve states’ all-payer claims databases. That study is part of an employer-led initiative, so its interest is the buyer’s: employers want to know what they are paying for. Read the spread inside it, because the average hides the story. States like Michigan, Massachusetts and Rhode Island averaged under 200 percent of Medicare; several others ran above 300 percent. A spread that wide is not a disease. It is a negotiation, settled by how much competition exists and who is doing the bargaining.
• The student does not see the price either. Tuition is covered by a subsidy, a loan, or both, and the person choosing the college is not the person paying the bill.
• The price is paid by a third party who was not in the room. Insurance, government, employers, parents.

Once the person who receives the service is not the person who feels the cost, the one discipline that forces productivity on everybody else is weakened, in exactly the sectors where productivity is hardest to get. That is why the price path here runs above what the disease alone would predict. The exact split is not known, and anyone who gives you a number is guessing. Clearly both forces are real, and the argument about the size of each is unsettled.

What the Diagnosis Does Not Excuse

The magnitude really is contested, and you should hear both sides.

For. Robert Archibald and David Feldman’s 2008 study in The Journal of Higher Education set the cost disease against the competing “revenue theory” of college costs and concluded that the cost disease explanation dominates, a finding they later expanded into a book on college pricing.

Against. Bruce Kimball and Jeremy Luke went back through 150 years of American higher education in Social Science History (2017) and found little validation for the theory except in particular periods. Tyler Cowen’s 1996 paper is titled, flatly, “Why I do not believe in the cost-disease”, and his objection is worth considering: if a service is measured by hours of labor, a service that improves its quality while charging more looks like pure inflation.

So treat the mechanism as established and the size as argued about. The index measures list prices, not what people actually pay. Published tuition has risen far faster than the average net price after discounts and aid. The sticker price is a real number, and so is the discount.

Quality cuts both ways. If a hospital in 2025 can do things a hospital in 1985 could not, some of that price rise is payment for more, not the same thing costing more. The television index adjusts for that and the care index does not, which flatters the machines and penalizes the medicine.

Which brings the argument back to Baumol, who spent his last book insisting that his own diagnosis was not a license to stop trying. The Cost Disease: Why Computers Get Cheaper and Health Care Doesn’t (Yale University Press, 2012) contains a chapter whose title is the whole point: “Yes, we can cut health-care costs even if we cannot reduce their growth rate.”

The disease explains the slope. It does not explain the level. And an industry that treats a diagnosis as an alibi has misread it.

The Bill Nobody Sends for the Gains

There is one more consequence, and it is an uncomfortable one.

William Nordhaus tested the mechanism across the whole American economy in “Baumol’s Diseases: A Macroeconomic Perspective” (The B.E. Journal of Macroeconomics, 2008), using industry data from 1948 to 2001. His findings: technologically stagnant sectors show rising relative prices and declining relative real output, and, as their share of the economy grew over the second half of the twentieth century, that shift dragged down measured productivity growth for the country as a whole.

This means nothing got worse. Nobody got lazier. But as spending and employment migrated into the sectors where the work cannot be speeded up, the average productivity number for the entire economy bent downward, because the average was now being computed over more slow work.

Now add the politics. The sectors carrying the disease are the ones you meet weekly: the hospital, the school, the care home, the tradesman. Every one of those encounters produces a visible bill with a name on it. The thing that made those bills affordable relative to your income - sixty years of productivity gains in everything else, which is why a television you could not have afforded in 2000 now costs less than a week’s groceries - never sends a bill. It shows up as an absence.

So the public sees the price rises, attributes them to greed, and concludes that the market has failed. The mechanism doing the damage is the market succeeding somewhere else. That is a misunderstanding with consequences, because the wrong diagnosis produces the wrong prescription: cap the price and you get shortages, as the rent control evidence and every other price ceiling keep demonstrating. Blame the producers and you get a thicker rulebook over the sectors that are already hardest to make productive.

The Lens

Two questions, and they will serve you for the rest of your life every time something gets more expensive.

Is the work itself the product? Not “is it labor-intensive” - plenty of labor gets replaced. Ask whether the thing you are paying for is somebody’s time, such that they cannot get more productive. If it is, the price will rise for as long as everything else gets cheaper, and no amount of anger at the seller will change the arithmetic.

And who pays at the point of use? If the answer is “nobody in the room”, you have found the part of the bill that is a choice. That part is ours to get right.

Prices in care, teaching and repair will keep climbing faster than the average, and that is a signature of a country getting richer in everything else. Whether the country also gets good value for those rising prices is a separate question. It is not answered by the disease. It is answered by whoever is paying - and whether they can say no.

Related: Greedflation: What the Theory Misses, How Prices Work, The Credential Con, and The Great Escape.