We have all seen the doomsday predictions. Health expenditures in the United States are a large and ever growing component of GDP. Nothing can seem to stop this perpetual growth, and eventually these expenditures could account for over 20 percent of government spending and have ruinous effects on both state and federal budgets. This was part of the motivation for the 2009 Affordable Care Act (ACA).
But before the ACA was even drafted, something important was happening to the rate of growth medical spending – it was slowing! From 2000 – 2007, health spending grew at 6.6 percent per year, well above inflation and a source of concern. However, over the next four years this rate of growth slowed to an annual rate of “just” 3.3 percent. We should note that even slower growth of a large number is a cause for concern.
The source of this slower growth has caused great debate. Theories have ranged from a slower adoption and development of new costly technologies, to a number of blockbuster drugs coming off patent, to early effects of the ACA, and finally the effects of the “Great Recession.”
In a new study published in Health Affairs, we (along with our co-author Christopher Ody) examine the role of the economy in the slowdown in health spending. In a research partnership with the newly established Health Care Cost Institute (HCCI) we obtained data on the health expenditures of a large sample of privately insured individuals. We aggregate these data to the CBSA level and then examine how the effect of the 2008 economic downturn affects health spending in these areas. We find that absent the recession, the rate of growth in health expenditures would have been 70 percent higher. This is similar to some previous studies, and quite higher than others.
An advantage of our study over the existing literature is our ability to exploit regional variation during this downturn. Previous efforts have primarily used evidence from past recessions on the relationship between aggregate economic activity and health spending to predict the role of the recession in the past few years. However, there are many reasons to think that this recession might be different. For example, it has been a longer and deeper recession than those in recent decades. In addition, it was caused primarily by a shock to the financial system and results in large losses in housing wealth. A second advantage of our study is that we examine individuals that retain private insurance. Therefore, our results show that the downturn in spending from the economy was not caused simply by individual losing employer provided health insurance.
Exploiting regional variation also allows us to control for other coterminous factors that might be affected health spending. A fair question is what is the appropriate measure of economic activity in this setting? The generally used measure in this literature is Gross Domestic Product (GDP). For two reasons, this measure won’t work for our research question. The first is practical. We don’t have good measures of this type of economic activity at the local (i.e. CBSA) level. The second is more conceptual. It is not clear that GDP reflects the economic conditions facing the average American. This is particular true following the most recent recessions which have been marked by “jobless” recoveries where employment growth lags aggregate economic activity.
For these reasons, we measure the impact of the 2008 economic downturn using the change in the CBSAs employment-population ratio. Looking just at the raw correlations, we find that every one percentage point decrease in the employment to population ratio resulted in a 0.84 percentage point decrease in medical spending in that CBSA.
While it is important to understand the past, the obvious question is: What does this mean for the future of health spending. Our results suggest that if all other things remain equal, as the economy improves we should see a return to the previous rate of growth in health spending. But are all things remaining equal? Of course not!
First, we have implemented the ACA. While we think it is naively optimistic to think this caused an immediate change in health spending in 2009, it is quite possible that in the future the incentives and programs created by this law could coordinate care in a way that leads to lower spending. On the other hand, it is also fair to note that an increase in the number of insured individuals might actually raise health spending. Just today, the Wall Street Journal reports large increases in utilization by the newly insured – including an increased used of the emergency room. This utilization increase might offset the benefits of more coordination.
Second, one mechanism that likely drives a portion of our main results is a decrease in the generosity of cost-sharing for employer provided plans. It is possible that this shift was greater in areas the suffered a larger macroeconomic shock as employers used less generous health insurance as a means of lowering compensation and absorbing the financial shock. However, if this were the case it is quite possible that future employment negotiations may increase the cost sharing in these plans in the future. Such a shift may take some time as these contracts are typically negotiated on at least at most an annual basis. Therefore, we are not sure whether this will remain a permanent feature of the market.
Our results show that attempts to claim the slowdown in health spending primarily resulted from changes in government policy have little merit. Importantly, we find this change among a group of insured individuals. This means that we are not simply finding that individuals losing insurance following job loss spend less on health services. Instead, our results show that the depth and breadth of the downturn (and likely the decline in housing wealth) affected even the health spending of those who retained insurance.
However, it is also important to note that we find that among our sample of privately insured individuals approximately 30 percent of the decline in the growth in health spending resulted from factors other than the local economy. Given what we spend on health care services each year, this point is nothing to gloss over. It is important to think carefully about what might or might not be in that 30 percent. For example, this could result from factors such as a lower rate of technology adoption or the lack of new blockbuster drugs (though if this was the cause it seems like Sovaldi alone might end this channel). But it also could be that economic trends that were not well correlated with the local economy, such as changes in the stock market, could explain a large portion of this effect. Given this fact, we might want to keep our proverbial champagne on ice for even the 30 percent decline going forward.