The Centers for Medicare and Medicaid Services (CMS) has released its latest forecast of medical spending for the next decade. The headline number is that medical care as a share of gross domestic product (GDP) is expected to increase from its current 17.5 percent of GDP to 20.1 percent by 2025, resuming an upward increase after a several year slowdown.
Forecasting is an inexact science. To make guesses about the future, analysts typically examine the past. The history of medical spending can roughly be described using Fuchs’ law: medical spending increases have exceeded GDP increases by about 2.5 percentage points annually since 1950. In the past three or four decades, the differential has been closer to 1.0 and 1.5 percentage points. Consistent with this, the Actuaries project that the growth of medical spending will exceed the growth of GDP by 1.25 percentage points annually between 2017 and 2025. Regardless of the exact differential, the qualitative conclusion is the same: if medical costs increase more rapidly than GDP, medical care will grow as a share of the economy.
The continuity between past and future is critical to this conclusion. Generally speaking, what happened in the past is likely to repeat itself in the future. But what if the continuity is wrong? What if fundamental changes have reduced the long-run growth of medical spending relative to the economy?
Figure 1 below shows the time series of real per capita medical spending increases and real per capita GDP growth. The data are from the CMS projections just released. I start from 1975 so that I can focus on the more recent period and the projection. To aid the visual analysis, the historical growth rates are three-year moving averages.
The data from 1975 through the early 1990s are consistent with Fuchs’ law: the growth of medical spending exceeded the growth of GDP between 1975 and 1992 by 3.1 percentage points annually. The remainder of the 1990s saw the managed care revolution, and with it lower spending, largely via reduced price increases. The managed care backlash is apparent as well. Provider consolidation and the loosening of managed care restrictions in the late 1990s led to large price increases and thus high spending growth in the early 2000s.
But then the picture changes. From 2004 to 2013, the growth rate of medical spending slowed markedly. Medical spending growth was actually below GDP growth from 2011 to 2013. Spending increased in 2014, though this is largely explained by the enormous coverage expansion that took place that year. The slowdown has been a decade in the making.
The decade-long slowdown in spending growth caught most observers by surprise. Figure 2 shows the Actuaries’ forecasts for the share of GDP that would be accounted for by medical care made in different years. In its 2008 forecast, the Actuaries thought that medical care would reach 20.3 percent of GDP in 2018. A decade of slow growth and a massive coverage expansion later, the current forecast is that medical care will near that level in 2025.
Leaving aside the increase in spending due to new coverage 2014, a fundamental question emerges: At what point does one believe that a decade of slow spending growth is likely to persist? When does ‘new’ become ‘normal’?
The Actuaries’ view is ‘not yet.’ They believe that price increases will come back up, and that utilization growth will rebound. Some of this seems to be occurring. For example, the very low growth in spending in 2012 is partly attributable to the loss of patent exclusivity for many branded drugs in that year, including Lipitor, Plavix, and Singulair. We know from pipelines that this will not recur for some time.
Further, several expensive new medications have been approved, including drugs for Hepatitis C, some cancers, and multiple sclerosis. Further, trends in industry consolidation are creating conditions where further price increases are possible. Hospitals and physicians are consolidating faster than antitrust authorities can prevent them. History shows that consolidation is a recipe for higher prices.
Why Health Spending Growth Could Be Lower Than Projected
But the case is not so clear cut. There are at least four reasons to think that the Actuaries may have overstated the path of medical spending.
The Opportunity To Cut Wasteful Spending
First, it is clear that there is enormous waste in the medical system. Perhaps the most widely cited estimate is that one-third of medical spending is not associated with improved health, with estimates ranging from one-quarter of spending to over half. If the medical system eliminated this waste over the course of 15 years, that would be a reduction of 2 percentage points in spending growth annually.
Of course, the fact that waste exists does not mean that it will be eliminated. But the converse is true as well. The fact that waste is present means that it is possible to reduce spending growth substantially—even well below levels in recent years—without adverse impact on patients.
High Cost Sharing
Second, trends towards higher cost sharing in insurance have led to marked reductions in utilization. Over the last decade, high-deductible health plans have gone from 4 percent of the privately insured population to 24 percent. People on high-deductible plans use less medical care.
Consistent with these trends, medical care utilization has increased less rapidly for lower-income populations than for higher-income populations in recent years. This is part of the reason for the recent slowdown in medical spending. The trend towards higher cost sharing is likely to continue, especially if the Cadillac Tax or some replacement takes effect. Thus, spending growth is likely to fall by even more.
And even if cost sharing stabilizes, a higher level of cost sharing may be associated with smaller spending increases. To begin with, people with higher cost sharing may be more skeptical about using new services, such as specialty medications that come at a very high price but have limited benefits. In 2012, Memorial Sloan Kettering famously decided not to cover one medication because it cost twice what a competing medication did but had no better clinical profile. (The response by the manufacturer was to cut the price in half.)
In addition, people may get better at shopping for lower prices. Price shopping has been the missing ingredient in the response to higher out-of-pocket costs. While studies uniformly show that higher cost sharing reduces use of services, fewer show people buying the same services from less expensive providers. To a great extent, this is because the prices that consumers face have been shrouded in secrecy. Some evidence shows that consumers do shop for lower priced providers when given the information to do so. As cost sharing has increased, many insurers are putting together price transparency tools, which could spur greater price shopping.
Third, the payment system is changing in a way to encourage more efficient care provision. The fee-for-service payment system is rapidly giving way to a world of alternative payment methodologies. In the public sector alone, this includes the accountable care organization (ACO) program, the Comprehensive Primary Care Plus medical home model, and the comprehensive joint replacement model to bundle payments for hip or knee replacements.
Private insurers are following a similar path. The Secretary of Health and Human Services has set a goal that 50 percent of Medicare payments be on a value basis by 2018. Many of the authorities needed to make these changes have been codified in recent legislation, including the ACA and the Medicare Access and CHIP Reauthorization Act of 2015.
Physicians are responsive to financial incentives. For example, experiments to bundle fee-for-service payments into larger groups have shown significant savings, as have some of the preliminary evidence from Accountable Care Organizations. The stakes need to be significant for physicians to change their practice, but when they are, physicians respond.
The ability for public and private payers to enact such payment changes makes the forecasting task for CMS difficult. The Actuaries’ forecasts are designed to reflect ‘current law.’ This is as they should be, since the point of the exercise is to show what will happen absent changes in policy. But when current law involves discretion, it is not clear how to interpret the forecasts.
Implicitly, the current forecast assumes that payment methodology changes will not be made, or that they will not materially affect the growth of medical spending. The first is a political judgment; I have no idea if it is true. The second is possible, although I weigh the evidence as more consistent with a sizeable potential response.
Potential Administrative Savings
Fourth, there are other areas of medical spending that are amenable to interventions and that could yield significant savings. Administrative waste is a classic example. The United States spends $361 billion annually on health care administration — more than twice total spending on heart disease and three times spending on cancer. Such spending could be cut substantially; some estimates suggest at least in half. The key is to substitute technology for people, for example taking people out of the coding process and having computer interactions determine appropriate billing and payment.
Making such changes are difficult but not impossible. It will require action from the Administration. But once again, the authority to do so exists under current laws.
All of these factors suggest that the ‘new’ growth rate may in fact be the ‘normal’ growth rate, or that medical spending increases could fall even farther. Of course, the possibility of significant spending growth reduction does not mean that spending growth reduction is certain to occur. But then again, the possibility that the future will exactly resemble the past does not make that the most likely path either.