A new article by Adam C. Powell, Ph.D. a Senior Lecturer at MIT Sloan who co-teaches Healthcare Lab, argues that high U.S. health spending is not a directly fixable problem, but rather is a byproduct of the structure of the healthcare system and the economy built around it.
For more than 50 years, American presidents have worried about the cost of healthcare. Every few years brings a new plan to reduce costs. None has stuck. In 2024, U.S. health spending grew 7.2% and reached 17.2% of the economy, compared with an average of 9.3% across other wealthy OECD countries.
The usual explanation is waste: inflated prices, unnecessary tests, piles of paperwork. Those problems are real. But in "Structurally Expensive," forthcoming in Seminars in Roentgenology, Powell argues they are not the whole story. Much of what drives spending, he writes, is built into the system and has not yielded to the cost-cutting tools tried so far.
A service you can't automate
Start with what economists call Baumol's cost disease. Baumol’s theory holds that sectors such as manufacturing can increase productivity to reduce costs and raise wages, whereas sectors that rely on human judgment, such as healthcare and education, cannot achieve the same productivity gains. For example, a factory can make more cars per worker every year. A doctor examining a patient or a nurse at the bedside can't speed up nearly as much, because the work moves at the pace of human judgment. Yet hospitals still have to pay those workers enough to keep them from choosing software, finance, or engineering instead. So healthcare gets steadily more expensive relative to everything else, even when it is run well.
Better tools, more work
In most industries, new technology cuts costs by replacing labor. In healthcare, new technology more often expands what's possible. Drawing on radiology, Powell notes that imaging studies once made up of tens of images now often contain hundreds, and some contain thousands. Scanning got faster; reading the results did not. Sharper images also turn up incidental findings that lead to more tests, more specialist visits, and more follow-up.
He expects artificial intelligence to follow the same pattern. AI is useful for triage, measurement, and quality control, but it is more likely to add services than to replace radiologists. Medicare already pays separately for some AI-assisted analyses, in addition to the payment for reading the scan.
A physician pipeline with a bottleneck
Becoming a U.S. physician takes four years of college, four years of medical school, and then residency. Three in four graduates carry student debt, with an average of about $200,000. Meanwhile, a 1997 federal law capped the number of residency positions Medicare funds, and that cap stayed essentially frozen for nearly 25 years. The U.S. now has 2.7 practicing physicians per 1,000 people, versus an OECD average of 3.9. Scarce supply means high pay.
Healthcare is the jobs program
Healthcare and social assistance is now the nation's largest employment sector, with about 24 million workers, and that sector is the top employer in 38 states. Cutting spending at scale means cutting paychecks and sometimes closing the biggest employer in town. Because every state gets two senators, most of the Senate represents places where healthcare anchors the local economy. The costs of cutting are concentrated and visible; the savings are spread thin.
What doesn't explain it, and what can help
Powell also takes on popular explanations. Population aging accounted for only about a tenth of spending growth between 1996 and 2013. A single-payer system could trim administrative costs and prices, but by Powell's reading of Congressional Budget Office estimates, even the most favorable scenario would cut national health spending by only about 11%. That, he argues, is not enough to close the gap with peer countries and leaves training, wages, and chronic disease untouched. Meanwhile, peer countries are also facing healthcare cost growth due to Baumol’s cost disease.
His conclusion is realistic rather than fatalistic: America cannot make healthcare dramatically cheaper without changing what it does, whom it employs, and how it trains its professionals. The better path is local stewardship. That means cutting low-value tests, using price data to steer patients away from overpriced facilities, letting highly-trained clinicians focus on work only they can do, and judging new technology by whether it actually saves time. The gains are modest. But they are real and within reach.