Anyone who has ever hung a shelf knows there are two ways to get it wrong. The shelf can hang too high. Or it can tilt, so the height grows from one end to the other. Two different problems — and fixing one does nothing about the other.
Now look at U.S. health care spending, which reached $5.3 trillion in 2024 — 18 percent of the entire economy. CMS actuaries project this to hit 20.6 percent of GDP by 2034 — more than one of every five dollars in the American economy. When people complain about health care costs, they are usually mashing together two different grievances: costs are high, and costs keep rising. Those sound like the same complaint. They are not. One is a level problem (the amount). The other is a growth problem (the increase). Health spending has consumed a growing share of the U.S. economy for more than six decades, and every time the trend has reversed, the reversal proved temporary. A surprising amount of what frustrates people about American health policy follows from solving the wrong problem.
High is not the same as rising
Some things make U.S. health spending high without making it rise faster each year. Administrative complexity is a prime example — not any single administrative function, but the sheer redundancy of the machinery: parallel billing and eligibility systems that do not exchange information; credentialing repeated payer by payer; claims formats and portals rebuilt for every contract. Streamlining that redundancy would save real money. Once. It would lower the shelf without changing its slope.
Eight drivers, two kinds
Currant’s new white paper, Why Do U.S. Health Care Costs Keep Rising?, sorts eight drivers of U.S. health spending by a simple question: Does this force explain why spending is high, or why it keeps accelerating? The distinction is not academic: Level drivers yield to one-time reforms, while growth drivers require ongoing intervention — so knowing which kind you face tells you what a given fix can deliver before you spend the effort.
On the level side: Administrative costs are substantial and reducible but are not the engine of acceleration. Aging matters less than intuition suggests, though it grows more prominent as Baby Boomers move through Medicare. Population health burden explains where spending is concentrated (with two exceptions, obesity and behavioral health, that behave like growth drivers — both compounding rather than holding steady). Finally, income growth and insurance coverage act as enabling conditions; they make adoption affordable without independently driving the climb. That distinction will matter when the next slowdown arrives.
On the growth side: Medical technology is the leading long-run growth driver because new capabilities command new prices and new volumes. Provider market consolidation drives spending growth because concentrated markets set prices, that power compounds year after year, and costly new technologies diffusing through concentrated markets can command prices beyond what the innovation alone would justify. Labor and workforce costs make up an emerging growth driver, with wage pressure reaching rates and premiums on a multiyear lag.
It is important to note that two growth drivers — medical technology and provider market consolidation — can also be level drivers: Each raises the level of spending even as it drives the climb. Alongside them sit price and service intensity — not an independent cause but the channel where the growth forces surface in the spending data. Treating the channel as the disease is how policy ends up chasing symptoms.
What three decades changed
Classic frameworks — Newhouse’s in 1992, its 2009 update by Smith, Newhouse, and Freeland — got the technology story right, and it holds up. What they could not fully anticipate is how much the ground would shift beneath it: hospital markets consolidating into price-setters, administrative complexity becoming more visible and measurable, and a workforce-cost shock working its way through the system. Currant’s framework keeps what those models got right and adds what three decades changed. For anyone who has to act on health care costs — payers designing benefits, manufacturers setting strategy, policymakers choosing where to spend political capital — that classification is the practical payoff.
Three times the curve bent — and un-bent
Three times in recent decades, U.S. health spending growth slowed sharply — and each time it came back. In the 1990s, the HMO era slowed spending growth to its lowest rate in decades through contractual leverage, utilization management, and other arrangements that constrained prices and volume — until the backlash arrived, the market shifted toward looser PPO products, and gains won through leverage, rather than durable changes in market structure, eroded. After 2008, the Great Recession and a thin drug pipeline flattened growth for five years — and then spending resumed as if the pause had never happened. In 2021 and 2022, health care’s share of GDP actually fell — not because of anything health policy did, but because the rest of the economy rebounded faster. It was a denominator effect widely mistaken for progress.
Three slowdowns, three different mechanisms, one common feature: None of them touched technology diffusion, consolidation, or labor costs. Each time, the climb resumed.
A fourth slowdown is coming. Watch what happens next
CMS actuaries project national health spending growth to cool from roughly 7 percent in 2025 to 6.3 percent in 2026 and remain subdued through 2028, as enhanced Marketplace subsidies expire and new Medicaid provisions phase in. Coverage will contract; measured spending growth will slow; it will be tempting to declare the curve bent.
Currant’s framework says the curve is not bending: This slowdown operates on insurance coverage — an enabling condition — and on administered Medicaid payment rules, not on the forces that generate the growth. We have seen the mirror image before: The ACA’s coverage expansion in 2014 coincided with an acceleration of spending growth, and neither the expansion nor its unwinding changed the long-run trajectory because coverage enables spending. It does not generate the growth. CMS actuaries’ own projections of GDP by 2034 point the same way. Currant’s white paper predicts the slowdown will pass, and the climb will resume — worth revisiting in 2028.
Is the climb even a problem?
Rising health spending is not automatically a problem. Economist Mark Pauly likes to point out that a wealthy country may simply be choosing to buy more health care, the way it buys more of anything it values. Spending is only “too much” if the resources would do more good somewhere else.
Two things turn this climb into a problem. First, Americans keep saying the resources would do more good somewhere else — in paychecks, in classrooms, in public budgets — and at one dollar in five, the crowd-out is no longer hypothetical. Second, and more fundamental: When growth reflects prices set by market power rather than value delivered to patients, the extra spending is not a choice at all — it is a transfer. Growth that buys genuine health is defensible. Growth that pays consolidated price-setters is not the same thing. How spending grows matters as much as how much it grows.
The attention doesn’t match the contribution
Prescription drug prices command an outsized share of the policy conversation — negotiation, importation, price caps, PBM reform. Yet retail prescription drugs account for roughly 8 percent of cumulative U.S. health spending growth over the past two decades; hospitals, by comparison, account for about 32 percent. However one feels about drug prices, the attention seems disproportionate to the growth.
The growth driver getting the least attention of all is labor — a persistent clinical workforce shortage, wage pressure still working through contract cycles, and no federal response yet scaled to the problem. This is not an argument against drug-price policy or administrative streamlining. Budgets are real; one-time savings are still savings. It is an argument for honesty about which problem a given reform can solve.
Ask which problem they’re solving
The next time someone promises to fix health care costs, ask one question: Does it touch the level or the growth? If the answer involves streamlined paperwork, transparency dashboards, or a one-time price cut, it is aimed at the level — useful, worth doing, but unable to stop the climb. Durably bending the curve means reaching the growth drivers themselves: governing how costly technologies diffuse and are priced, restoring competitive discipline in consolidated provider markets, and rebuilding the clinical workforce pipeline so scarcity stops writing the wage bill. Until policy reaches the growth, all we will ever fix is the level.