Explaining the Growth in US Health Care Spending Using State-Level Variation in Income, Insurance, and Provider Market Dynamics

At least 70 percent of the recent slowdown in healthcare spending can likely be explained by long-standing patterns in healthcare spending related to changes in income, insurance, and provider market characteristics.

The recent slowdown in the growth of healthcare spending has led analysts to question whether, and to what extent, the phenomenon resulted from predictable factors such as the Great Recession, or resulted from some new structural change in the healthcare sector, possibly related to the Affordable Care Act (ACA). The answer could be critical to projecting the likely future pattern of spending for medical care with the ACA in place.

A new study funded by the Mercatus Center at George Mason University offers an important potential explanation for recent healthcare spending trends. In a paper published in the journal Inquiry, professors Bradley Herring from Johns Hopkins University and Erin Trish from the University of Southern California use a method of analysis not previously applied to this question and find that at least 70 percent of the recent slowdown in healthcare spending growth—if not all of it—results from predictable, known factors such as income, insurance coverage, and provider market characteristics. The authors also apply these results to project the likely healthcare spending increases from the Medicaid expansions of the ACA.


  • At least 70 percent of the recent slowdown in healthcare spending per capita—and possibly as much as 98 percent—can likely be explained by long-standing patterns known to affect healthcare spending trends, not by new, unexplained conditions in the medical sector.
  • Breaking down those figures, roughly 41 percent of the slowdown probably resulted from the decline in real per capita income because of the Great Recession.
  • Other factors known to affect healthcare spending growth—such as changes in the number of physicians and hospital beds per capita and in the percentage of the population with insurance coverage—account for somewhere between 32 percent and 57 percent of the slower healthcare spending growth.
  • The projected expansion of Medicaid coverage owing to the ACA will likely raise national healthcare spending in 2019 to about 1 percent higher than it would have been without the expansion.


The rate of healthcare spending growth has slowed considerably over the past several years. For instance, healthcare spending per capita, adjusted for inflation, grew by just 1.48 percent annually from 2008 through 2013. That was less than half the 3.76 percent annual rate from 2000 through 2007.

Proponents of the ACA, enacted in 2010, have sought to attribute this slowdown largely to structural changes in healthcare delivery. Others have contended that the Great Recession was most responsible, along with a temporary decline in the adoption of new medical technologies and a shift of some individuals from private to public insurance. Understanding the reasons for the slowdown is important for estimating the likely growth of healthcare spending in the future with the ACA in place.


This study takes a systematic approach to analyzing the trends in healthcare spending. Using a wide range of data sources, it builds on a model examining the past relationship between healthcare spending and factors such as income, insurance coverage, and provider market characteristics, as well as geographic variations reflected in state-level measures. To account for possible differences in these effects over time, the study evaluates two periods: 1991 through 2009 and 2000 through 2009. The principal components of the model used for the analysis are the following:

  • State healthcare spending. The state-level data reflect total healthcare spending, regardless of payer. That is, they are the sum of Medicare, Medicaid, private insurance, and out-of-pocket spending. The study examines the extent to which healthcare spending changes resulted from the three variable factors (the remaining bullets) in each of the time periods studied.
  • Income. Income measures include each state’s per capita income, the percentage of the state population living below the federal poverty line, and the state unemployment rate. As expected, rising healthcare spending relates closely to the rate of growth in per capita income, as well as to increases in poverty and unemployment.
  • Insurance coverage. The data show a predictably significant relationship between healthcare costs and the percentage of the population with insurance coverage. This is likely because of the “moral hazard” effect, according to which individuals with insurance coverage tend to use more healthcare services. The effect is especially true for Medicaid and the Children’s Health Insurance Program. Meanwhile, also as expected, the penetration of health maintenance organizations (HMOs) in a given geographical area tends to reduce healthcare spending.
  • Providers. An increase in the number of physicians per capita closely relates to healthcare spending growth, while increases in the number of hospital beds per capita is associated with reductions in healthcare spending. Interestingly, medical malpractice reforms that cap noneconomic damages unexpectedly had little effect on healthcare spending growth across the entire 1991–2009 period, while reforms that replaced joint and several liability with proportionate share allocation of liability did vary significantly with healthcare spending increases.


In order to project the degree to which these known factors affected healthcare spending from 2010 through 2013, the study first compares the actual increases in national healthcare spending with a hypothetical, 3 percent real per capita growth trend, equal to the average real growth rate from 1991 through 2009 (termed a “counterfactual” trend). It then compares the difference between those two to the difference between actual national spending and the projected spending. This spending is driven by the state-level model’s effects derived from income, insurance coverage, and provider market characteristics. Finally, a comparison between the two differences shows how much these factors might explain the observed slowdown.

The results show that the decline in real per capita income because of the Great Recession probably explains 41 percent of the observed reduction in healthcare spending relative to this counterfactual. Other known factors—including insurance coverage and provider market characteristics—can account for between 32 and 57 percent of the slowdown, dependent on the particular model specification. Totaling these factors indicates that they likely account for between 73 and 98 percent of the reduction in healthcare spending growth during the 2010–2013 period. That would leave little room for reductions because of the ACA.

Finally, the study estimates how much healthcare spending is likely to increase as a result of the ACA’s insurance expansions, given the model’s findings of a significant relationship between rates of Medicaid coverage and healthcare spending. (The model also examined private insurance coverage, but did not find a significant effect on healthcare spending.) Congressional Budget Office projections show that the percentage of the US population without insurance will decline from 14.7 percent to 7.8 percent by 2019, while the number of individuals covered by Medicaid or the Children’s Health Insurance Program will increase from 12.0 percent to 15.6 percent of the population. Applying the model’s result for the relationship between healthcare spending and Medicaid coverage shows that total healthcare spending will likely be about 1.0 percent higher in 2019 owing to the insurance expansions than it would have been otherwise.

The Mercatus Center gratefully acknowledges the financial support of the John Templeton Foundation for research on healthcare policy in the United States.