Friday, December 19, 2014
The expansion of coverage to some 10 million individuals in 2014 has implications for the growth rate in national health expenditures and on health workforce requirements. While there is broad agreement that spending and utilization will be pushed upward, there is much debate about the amounts. One recent study concluded that utilization doubles upon gaining coverage, but most recent research, including findings from the Oregon Medicaid lottery, suggests the increase is closer to one-third. In this post, I attempt to make sense of these rather divergent estimates and then draw implications. I conclude that the increase in utilization, adjusted for treatment intensity, is on the order of 50%. This translates into a 0.7% increase in overall health care utilization and workforce requirements when coverage is expanded to 10 million individuals. I note that the near-term impact on health care spending should be larger than the pure utilization effect, because providers will experience less uncompensated care but still receive substantial uncompensated care subsidies and have higher operating margins. I estimate that this adds another 0.7 percentage points, driving the total impact to 1.4 percentage points. Finally, due to lags in gaining coverage and obtaining care, these effects will not be felt in full until the latter part of 2014.

Health economists have been expecting an acceleration in health spending in 2014 due to the estimated 10 million or so individuals who have gained health insurance under the Patient Protection and Affordable Care Act (ACA). For example, the most recent 10-year projection from the Centers for Medicare & Medicaid Services (CMS) shows a 2 percentage point increase in the health spending growth rate, from 3.6% in 2013 to 5.6% in 2014, with most of this increase ascribed to an estimated 9 million individuals gaining coverage.[1]

In a recent article, I attempted to provide some intuition as to why expanding coverage to 9 million individuals (about 3% of the population) would cause the health spending growth rate to jump by somewhat less than 1.5%. In this argument, I applied an old rule of thumb, supported in a recent Society of Actuaries study, that health spending of the uninsured doubles upon gaining coverage.[2] After publication of this paper, I received a note from David Auerbach alerting me to a body of research suggesting that health care utilization increases by only about one-third when an individual becomes insured. Since spending is generally proportional to utilization, this is at odds with the doubling rule of thumb and cuts the estimated impact of expanded coverage in 2014 to something less than 0.5%.

In what follows, I have tried to resolve the conflicting findings on how insurance affects health spending and utilization. I also provide an argument for why the impact of expanded coverage on health spending will exceed that on utilization in 2014 because of excess uncompensated care subsidies. I conclude with an estimate of the likely impact on utilization and an updated estimate of the likely impact on spending.

Health insurance and health care utilization. As noted above, a recent study by the Society of Actuaries supported the rule of thumb that utilization (and spending) doubles for the newly insured. This finding was based upon analysis of data from the Medical Expenditure Panel Survey (MEPS) showing that insured individuals used about twice as much care as uninsured individuals after controlling for age, sex, income, and health status. It was assumed that the newly insured would expand their health care use to match that of similar insured individuals, and therefore their use would double.

In his blog, David Auerbach notes that those who gained coverage through random selection into the Oregon Medicaid lottery exhibited an increase in utilization of about one-third. He also cites a literature review done by the Congressional Budget Office supporting a similar conclusion. A somewhat older study by Hadley and Holahan shows a similar result but also holds a partial explanation for the discrepancy with the rule of thumb. [3] While their analysis suggests about a one-third increase in the number of office visits and hospital days, total charges are projected to increase by about two-thirds. This implies that health insurance not only increases the volume of encounters but also their intensity. The notion that the intensity of visits and days would increase for the newly insured is also supported in a paper just published by Abe Dunn, though he shows a relatively smaller impact for intensity.[4] Based upon these various findings, I think a reasonable estimate of the increase in utilization, adjusted for intensity, would be around 50%.

How does one resolve the difference between this 50% estimate and the Society of Actuaries study showing a 100% increase? The most likely explanation is that the newly insured do not utilize care at the same rate as the already insured, even after controlling for age, sex, income, and health status. This could be because of unmeasured differences such as attitudes toward seeking care and barriers to access to care. In effect, we are arguing that the newly insured use only three-fourths as much care as the already insured who have the same age, sex, income, and health status. If the Society of Actuaries had applied this assumption, their findings would be consistent with the 50% increase in utilization posed here.

The impact on utilization of expanding coverage by 10 million individuals. It appears that ACA expanded coverage has resulted in about 10 million newly insured, roughly 3% of the population. Assuming that these newly insured increase their utilization of health care by 50%, what is the percentage increase in total utilization? In order to answer this question, it is necessary to estimate the newly insured’s share of health care utilization prior to gaining coverage. The overall percent increase in utilization will then be 50% of this share.

As a first step toward estimating this share, I used data from a Kaiser Family Foundation report to estimate the uninsured’s share of health care utilization among the nonelderly population. The report estimates that in 2013, there were 40.8 million individuals with no insurance for the full year and provides estimates of the private-pay value of care utilized by this group. Combining this with similar information for the partial-year insured and fully insured, I estimate that the 40.8 million full-year uninsured accounted for 8.6% of health care utilization among the nonelderly in 2013. Next, I used CMS tabulations of personal health spending by age to estimate that the nonelderly account for about two-thirds of total health care utilization.[5] Thus, the 40.8 million full-year uninsured accounted for 5.6% of total health care utilization.

Data from a Gallup poll suggest that the newly insured are roughly representative, age-wise, of the uninsured population.[6] Thus we can estimate that the 10 million newly insured accounted for about 1.4% of health care utilization prior to gaining insurance (one-fourth of the 5.6%) so that, upon gaining coverage, they would cause total utilization to grow by 0.7%. This is about half of what I had suggested in my previous article and is as would be expected, since the rule of thumb has been cut in half.

Why the impact of expanded coverage on spending will exceed the impact on utilization. It seems reasonable to assume that the 0.7% increase in total health care utilization due to expanded coverage will result in roughly the same increase in health spending. However, higher utilization is not the only way in which expanded coverage affects spending. The second mechanism is through the increased payments for the care that was previously partially uncompensated.

Figure 1, from the Kaiser Family Foundation report mentioned above, displays per capita health care spending in 2013 for nonelderly individuals who were uninsured for the entire year by source of payment. These estimates were derived by researchers at the Urban Institute using data from the MEPS. Because the uninsured often do not pay in full for care received, per capita spending is estimated based on what would have been paid for the care under private insurance.[7] As the chart shows, actual payments for the care delivered to the uninsured come from multiple sources.

Figure 1: Per Capita Medical Spending Among the Nonelderly Full-Year Uninsured, 2013

Implicit Subsidies

Source: Coughlin et al., Uncompensated Care for the Uninsured in 2013, Kaiser Family Foundation, May 2014.

About 20% is paid out of pocket by the uninsured individuals, with another 10% paid by a third party.[8] About 29% of the care is provided by other sources such as the Department of Veterans Affairs (VA) and the Indian Health Service or paid by workers compensation. The final 41% is not paid for directly but is roughly covered by a variety of primarily public programs aimed at covering the costs of uncompensated care. Medicaid and Medicare programs such as disproportionate share (DSH) payments cover 9% and 5%, respectively, with state and local programs covering 11% and the remainder coming from physician charity care (truly uncompensated) and other sources.

With this as background, suppose that a representative group of 10 million uninsured persons gain coverage and, as a first step, suppose that there is no change in their utilization of care. Direct payments for their care (out of pocket plus third party) will increase from 30% to 100% of total payments.[9] This increase in direct payments for care, combined across 10 million newly insured individuals, would drive total health spending up by about 1% (ignoring, for the moment, any accompanying increase in utilization).[10]

The key question is what happens to the other sources of funding associated with uncompensated care. It seems unlikely that there would be any major reductions in spending by the VA or other sources of care.[11] Medicaid has announced plans to reduce its subsidies, but that will not begin until 2016. Starting in October 2013, Medicare modified its DSH payment formula, but I have not been able to determine the magnitude of any net reductions. It is not clear what steps are being taken in terms of reducing state and local uncompensated care subsidies. There is lots of uncertainty here, but assuming a 20% reduction in “VA and other sources,” no current reduction in Medicaid, an 80% reduction in Medicare and state and local subsidies, and a 10% reduction in the remainder, which is mainly charity care for which there were not payments to begin with, my rough estimate is that these sources of funding would be reduced by about 30%. With this reduction, the increase in spending associated with what might be called “excess” subsidies for uncompensated care would be 0.7%.

In summary, expanded coverage pushes spending up both due to an increase in utilization and a reduction in uncompensated care that is not fully offset by reductions in uncompensated care subsidies. My rough estimate is that each factor adds about 0.7% to spending growth. Thus, we would expect expanded coverage to add about 1.4% to the growth rate in health spending.

What does this imply about the overall growth in health spending in 2014? According to data just released by CMS, national health expenditures grew by 3.6% in 2013. Adding 1.4% would bring the annual growth rate to 5.0%, but due to lags in obtaining coverage and gaining access to care, this jump would only be felt partially in 2014. Those who have followed our monthly indicators have seen that spending on health care services showed no uptick in the first two quarters of 2014 but in the third quarter it jumped to 5.4%. This could reflect the delayed impact of expanded coverage. The indicators also show a second source of acceleration that began early in 2014 that is largely unrelated to expanded coverage. The growth rate in spending on prescription drugs has increased from 2.5% in 2013 to 10.6% for the first there quarters of 2014. This increase seems due to primarily to new products and fewer patent expirations rather than to expanded coverage. If the uptick in spending on health care services continues and fourth quarter health spending grows at the same rate as the third quarter, the growth rate for the year as a whole would be about 5.3%. This would reflect a partial year of expanded coverage effects along with the acceleration in spending on prescription drugs.[12]


[1] The report does not say exactly how much of the acceleration is due to expanded coverage, but in last year’s projection, it was explicitly stated that expanding coverage to 11 million individuals would raise the growth rate in national health expenditures by 1.6 percentage points.

[2] Society of Actuaries. (2013, March). Cost of the future newly insured under the Affordable Care Act (ACA). Retrieved from

[3] Hadley, J., Holahan, J. (2003, January–June). Covering the uninsured: How much would it cost? Health Affairs, Suppl Web Exclusives, W3-250–W3-265.

[4] Dunn, A. Health insurance and the demand for medical care: Instrumental variable estimates using health insurer claims data. BEA working paper, WP2014-03.

[5] I used their most recent data which were for 2010 and assumed that utilization was proportional to spending.

[6] The percent reductions in numbers of uninsured were quite similar by age group.

[7] MEPS captures total charges for the care delivered to the uninsured and spending was estimated by applying the payment to charge ratio from those with private insurance.

[8] In MEPS, a small amount of third-party payments are made on behalf of those who claim to have been uninsured for the entire year.

[9] “Total payments” refers to the $2,443 spent per uninsured shown in figure 1.

[10] As argued earlier, the 10 million newly insured accounted for about 1.4% of health care utilization prior to gaining coverage. The increase in direct payments upon gaining insurance covers about 70% of the dollar value of this utilization or about 1% of the dollar value of total utilization.

[11] Under ACA, enrollment in the VA health system satisfies the individual mandate for insurance coverage.

[12] Another factor causing our estimated growth in health spending to accelerate in 2014 is the CMS projected jump in the net cost of health insurance. We have no data source to track this element of national health expenditures and therefore use the CMS projection converted to monthly rates.

All postings to the Health Policy Forum (whether from employees or those outside the Institute) represent the views of the individual authors and/or organizations and do not necessarily represent the position, interests, strategy, or opinions of Altarum Institute. Altarum is a nonprofit, nonpartisan organization. No posting should be considered an endorsement by Altarum of individual candidates, political parties, opinions or policy positions.



Subject Matter Experts

Areas of Expertise