First posted 9/19/11 on The Health Affairs Blog
Health spending share of GDP and PGDP. A monthly time series of the health spending share of GDP, from January 1965 through June 2011, is presented in the exhibit below. Recessionary periods are denoted by shaded vertical bars and the black line plots the health spending share of “potential” GDP (PGDP) — defined as what GDP would be at full employment. Since health spending is largely unaffected by recessions, the black line can also be viewed as a rough estimate of how the health spending share of GDP would have grown over time if the economy was always at full employment. Its rate of increase reflects the extent to which the growth in health spending is excessive relative to the growth in the full employment economy (further discussion of health spending share of PGDP). The deviations about this line in the health spending share of GDP are due to business cycles.
See Note 1 below for sources.
As an illustration, consider the period from October 2007 (just prior to the most recent recession) to June 2011. The health spending share of PGDP grew from 16.3 percent to 16.8 percent and we attribute that to excess growth in health spending. The share of GDP grew from 16.3 percent to 18.2 percent and we attribute the 1.4 percentage point growth differential to the recession and our slow recovery from it.
Excess growth in health spending. Over the 46-and-a-half year period represented in the exhibit, health spending grew at a 9.5 percent annual rate compared to 7.1 percent for PGDP for an excess growth of 2.4 percentage points. Let’s apply the customary shorthand and denote health spending growth rate as PGDP + 2.4. The historical pattern of excess spending growth is represented by the black line in the exhibit below. Because PGDP growth is quite stable, the jagged nature of this line reflects fluctuations in the growth rate of health spending. While the pattern is quite noisy, it is clear that 1992 marks a downward shift in excess spending growth. From 1965 through 1993, excess spending is almost always above the long run average of 2.4 percentage points. From 1993 forward, excess spending is almost always below 2.4 percentage points.
See Note 2 below for the definition of “other.”
The chart also identifies the contributions of the components of health spending to excess growth. For the sake of illustration, consider the contribution of hospital spending which is shown in blue. Over much of the period, hospital spending was a major contributor to excess growth. However, there are two periods in which it was actually a negative contributor – 1984 and 1994-98. During these periods, hospital spending grew more slowly than PGDP and pulled excess spending down.
Four periods of analysis. Take a closer look at the individual components. The colors suggest that we divide the pre-1993 period into two segments. The first segment, running from 1965 through 1983, is a time in which hospital spending (blue) was the largest single contributor to excess spending. From 1983 through 1993, spending on physician services (red) became more dominant.
Similarly, we can divide the post-1993 period into two segments. The first, which runs from 1993 through 2004, shows the emergences of prescription drugs (yellow/green) as a major driver of excess spending growth. The period from 2004 to present is characterized by particularly low excess spending growth. Here are a few observations about each period, supported by the third exhibit below.
1965 – 1983. Health spending grew at PGDP + 3.1. Over half of the excess spending growth was attributable to hospitals (blue). The combination of nursing homes, home health, and other residential services (dark grey), physician services (red), and other (light grey) each contributed similar amounts. Spending on prescription drugs made a small negative contribution – it grew by less than PGDP. This is the period immediately following the creation of Medicare and Medicaid and the ability to pay for hospital and nursing home care by the poor and elderly increased dramatically.
1983 – 1993. During this ten year period, health spending grew at a staggering PDGP + 3.5! In 1983, Medicare payment for inpatient hospital services was changed to prospective payment using DRGs. As a result, the contribution of hospitals to excess spending growth declined and actually was negative in 1984 (as noted above and shown in the previous exhibit).
All else equal this should have reduced the excess growth in health spending. However, the reduction in hospital spending growth was more than offset by growth in spending on physician services and “other”. The shift to more outpatient care due to patients being discharged “quicker and sicker” undoubtedly contributed to the increased growth rate of spending on physician services (data on individual physician incomes show strong growth during this period). The contribution from “other” spending was led by the net cost of private insurance, other professional services, and public health. Spending on nursing homes, home health, and other residential services continued to grow rapidly during this period.
1993 – 2004. During this period, health spending grew at PGDP + 1.6, a large reduction from previous periods. The excess growth is dominated by prescription drugs and “other” with hospitals and physicians contributing very little. Note that 1992 was the year that Medicare adopted the RBRVS-based physician fee schedule.
This period consists of the managed care era – running through the year 2000 or so and characterized by very low growth in hospital and physician spending — and its backlash when hospital and physician spending rebounded. I have chosen to combine them because I think the accelerated spending in the latter portion was largely the result of the controls put on in the former. It also shows that, even when the backlash effects are included, the net result of managed care was a substantial “bend” in the rate of excess health spending. Finally, the entire period was characterized by very high growth in spending on prescription drugs.
2004 – June 2011. During this period, health spending grew at PGDP + 0.8. This represents a further “bend” in excess health spending as defined here. While this period encompasses the recession (December 2007 through June 2009), the bend clearly began well before the recession. For example, health spending grew at PGDP + 0.9 between January 2004 and December 2007. Tom Getzen, who has studied the effects of business cycles on health spending, suggests that the lagged effects of the recent recession have helped keep excess spending down in recent years and that we should not assume that this low rate of growth is the new normal.
Concluding observations. Our use of PGDP rather than GDP to measure excess health spending produces a long term spending rate of PGDP + 2.4 from January 1965 through June 2011. This is similar to the GDP + 2.5 long term growth rate often cited by health economists and is no surprise.
However, it seems less well known that in the earlier years (1965 through 1993), the excess growth was much higher (PGDP + 3.2) and that it has recently been declining as shown in the third exhibit. I also find it interesting to see that policy responses to excess growth for hospital services, physician services, and prescription drugs seem to have been fairly successful. I am speaking of Medicare prospective payment for hospitals in 1983, the Medicare physician fee schedule in 1992, and the private sector (and Medicaid) push to generics in the 2000s. In each case, the excess growth was greatly reduced following the policy implementation.
In a call with Henry Aaron last May, he mentioned in passing that bringing health care costs under control could involve a series of incremental progressions, much as happened with reductions in the prevalence of smoking. This brief history seems consistent with Henry’s observation.
Note 1. Annual NHE estimates through 2009 are from CMS and converted to monthly by Altarum. Monthly estimates for 2010 to present are from Altarum. Monthly GDP estimates through 1991 are Altarum estimates and from 1992 to present are from Macroeconomic Advisors. Monthly PGDP estimates are based upon CBO quarterly estimates converted to monthly by Altarum.
Note 2. “Other” consists of various components including public and private costs of administering insurance, public health, structures and equipment, research, and health care goods and services beyond those included elsewhere.