Some announced catastrophes have this peculiarity: they never arrive, and nobody notices. US health spending is the most expensive example. It was long described as a fiscal time bomb: an exponential curve that would, sooner or later, devour everything. Fifteen years on, the curve has bent, the gap with the forecasts runs into hundreds of billions, and we still reason as though the explosion were imminent. This piece is about the cost of a narrative nobody bothered to update.
1 The fact
A gap of nearly a trillion dollars, in a single year.
The number
What was spent, against what was expected
In 2024 the United States devoted about $5.3tn to health care, some 18% of gross domestic product and nearly $15,500 per inhabitant. These are considerable sums. But set against what was expected, they tell a different story: according to work by David Cutler and Lev Klarnet, the gap with the official projections drawn up in 2010 amounts to nearly a trillion dollars of savings for that year alone. An independent analysis already put the shortfall at about $704bn in 2023. Spending kept rising; it simply rose far more slowly than everyone had taken for granted.
2024 gap vs 2010 projections
≈ $1tn
in savings for that single year (NBER, 2026).
Gap already seen in 2023
$704bn
below the path projected in 2010.
2 The unchanged narrative
Cuts are still justified by an explosion that did not happen.
The mismatch
The data moved, the discourse did not
The striking thing is not the slowdown: it is how little it changed the public debate. Budget decisions on health programmes continue to be justified by the urgency of containing spending described as out of control. The unsustainable-trajectory argument remains central, even though that trajectory has been contradicted by fifteen years of data. This is not to say the issue is trivial: spending remains high, and how to fund it is a real question. But it keeps being handled with an outdated curve, as if navigating with a chart the coastline stopped matching.
4 What slowed
Several causes, none of them spectacular.
The explanations
A slowdown without a miracle reform
The bend does not come from a single decision, but from a cluster of causes recent work identifies. Medical technologies emerged that improve health while costing less, contrary to the received idea that innovation always makes care dearer. Supply adjusted more than expected over the long run. Population health improved on certain fronts. And reimbursement arrangements changed in ways that restrain demand and make it more price-sensitive. None of these factors is spectacular; their sum, over fifteen years, produced the trillion-dollar gap.
5 The cost of an old story
When decisions rest on a curve the facts have abandoned.
The consequences
An overstated urgency leads to poorly calibrated choices
An outdated diagnosis is not neutral: it distorts trade-offs. If spending is believed to be exploding, broader and faster cuts are justified, and collateral damage is accepted that an up-to-date diagnosis would not warrant. The target can also be wrong: reacting to an imaginary surge in volumes means neglecting the documented problems, such as the level of prices charged or the share of administrative costs. The obsolete narrative does not merely make the debate inaccurate; it steers real decisions, committing budgets and affecting lives, on the basis of a path the facts have left behind.
6 The method lesson
A projection should be revised, not recited.
The intellectual discipline
Compare the forecast with what happened
The real lesson is methodological. A projection is a useful tool provided it is regularly confronted with what actually occurred: that comparison, and it alone, reveals whether the model held. Yet forecasts often circulate in public debate without anyone going back to check their accuracy; they then turn into beliefs, passed on long after being disproved. Faced with a long-term figure, the right question is not "what does it predict?" but "what did the previous version predict, and did it come true?" A forecast that is never evaluated is not information: it is a habit.
7 The mirror image
Extrapolation errs in the other direction too.
The symmetry
Underestimating is an error of the same kind
It would be tempting to conclude that projections always exaggerate. That would be the same mistake in reverse. Extending a trend can also lead to serious underestimation: the spread of the internet, the fall in renewable energy costs and the speed of adoption of certain technologies were persistently undervalued by models extending a past rhythm. The error is not in the direction of the forecast, it is in the confidence placed in it. A distant path, whether it heralds the worst or the best, deserves the same caution: it is a scenario, not a piece of data.
8 The counterpoint
What this reading should not be taken to mean.
The analysis's honesty
Less than forecast does not mean sustainable
Saying the explosion did not happen is not saying all is well. US health spending remains the highest in the developed world, represents 18% of GDP and still grows faster than the economy: in 2024 it rose 7.2%. Problems of access to care and out-of-pocket costs remain entirely unresolved, and population ageing mechanically pushes upward in the decades ahead. An observed slowdown is never guaranteed to continue: it can stop. The point is therefore not to deny the fiscal stake, but to ask that it be debated from today's figures rather than yesterday's fears.
9 Check the curve before acting
The conclusion: verify what you think you know.
The meaning of the episode
Narratives outlive the data that founded them
The story goes beyond US health care. It says something about how a society pictures its future: an alarming forecast, once installed, transmits itself and becomes a commonplace, while its refutation, technical and unspectacular, does not circulate. The narrative thus outlives the data that founded it. That is true of public spending, but also of pensions, climate, demography and debt: whenever a distant figure serves as an argument, it is worth asking where it comes from and when it was last revised. The useful question is not "what will happen?" but "the last time we predicted, were we right?"
The compass
①
The explosion did not happen. In 2024 US health spending ($5.3tn, 18% of GDP) came in nearly $1tn below the 2010 projections, after a $704bn gap as early as 2023.
②
The error comes from extrapolation. Extending a past trend means betting nothing changes: over twenty years, a modest annual gap becomes a gap of trillions.
③
The narrative outlived the data. Decisions are still made in the name of a disproved curve, while spending stays high for other reasons. This piece frames a debate; it is not advice.
Read alongside: The virtuous path debt contradicts by itself (long-term fiscal projections) and When debt crowds out health and education (budget trade-offs). Neighboring notion: debt dynamics. Reference: abbreviations & acronyms (GDP, bn, tn).