Blog
Tuesday, March 7, 2023
Did ACA Produce Free Healthcare?
An interesting observation was bouncing around Twitter over the weekend.
Fabulous point... in 2010 when Obamacare passed, health care was 17% of GDP and uninsured rate was 16%
— Timothy McBride (@mcbridetd) March 5, 2023
Today? health care 17% of GDP and uninsured rate is lowest ever: 8%!
See: https://t.co/athQpfpkww https://t.co/Q0SD1ObscA https://t.co/h8qMMiWVyg pic.twitter.com/CrEscJ8C4P
Left-leaning opiners are quick to attribute the success to the ACA, but it's not at all clear from the graph that this is what's happening.
Going back to 1990, the trend seems to be a sharp increase during recessions and then a gentle decline during expansions. This is true for 25 years up to 2015, when there's a slight bump up. Then, it begins another gentle decrease until the pandemic when it goes crazy. This graph certainly doesn't show that the ACA reduced health care costs.
But it's still an interesting observation that coverage is higher than it was despite the overall costs falling to pre-ACA levels. Despite the attempted attribution to the last major Democratic accomplishment, the cause is very difficult to pin down. The pandemic caused once-in-a-lifetime shocks to both healthcare and the economy. On the latter, there was a sharp decrease in GDP early-on, followed by an enormous stimulus that resulted in a quick economic recovery and then went on to cause a boom in demand that raised prices.
In healthcare, there was an immediate increase in demand for those who contracted Covid, but otherwise a sharp drop in healthcare demand as hospitals were not allowed to see non-Covid patients. Furthermore, policy changes in Medicaid and subsidies for the ACA upended the traditional coverage breakdown and also high employment would generally mean higher coverage through Americans' workplaces.
Two potential causes are services like telehealth, which is a lower-cost health care service, but this seems too small to account for a 2% of GDP drop in healthcare costs. A more likely cause is a shift from commercial to Medicaid. Medicaid reimbursements are infamously lower than commercial, so if more people switch from commercial insurance to Medicaid, high coverage rates would be achieved at lower cost.
The downside of this would be that providers would lose enormous sums of revenue. This can be tested by digging into the National Health Expenditures data.
The other potential cause is that healthcare coverage doesn't really mean anything. It's important to remember that having insurance does not necessarily equate to using healthcare services. Unless you have a chronic condition, visiting a doctor is a sporadic event. From 2001 to 2011 (for some reason the Obama Administration decided to stop producing these reports), the Census put out information on healthcare utilization. In 2011, the last available, 26% of people did not visit a medical provider at all; only 8% spent a night in the hospital; 57% did not take any prescription drugs.
Perhaps coverage increased, but it did not produce an increase in utilization. In fact, this theory is supported by the other graph in the Tweet.
Most of the drop in uninsured occurred in 2013. This marked the year both when subsidies kicked in for the healthcare exchanges and also when Medicaid expansion kicked in for the states that expanded.
But was there a simultaneous increase in healthcare costs? Looking back at the healthcare cost graph for 2013-2014, if anything, it shows a slight decline. This suggests that the explanation for the coverage/cost conundrum is that they are not strongly-linked. Particularly for Americans who are currently uninsured. It is a criminally under-reported fact that many who are not insured choose not to be despite having access to free or subsidized coverage.
Lastly, this entire observation may be ephemeral. The original graph of costs is not from CMS or a government agency, and thus is not the official value, but a non-profit organization's attempt to recreate it on a monthly basis. Their numbers are not quite the same as CMS for the years where they overlap, and notice that the graph goes into 2022, data unavailable through CMS. This wouldn't be the first time that pundits were quick to react to partial information that proved their point and then were proven wrong.
Wednesday, February 8, 2023
Does Access to Medicaid Improve Health?
There have been several studies and politicians over the years purporting that health insurance, in general, and specifically Medicaid saves lives. A recent working study has found that, contrary to previous research, Medicaid expansions do not save lives. This result calls into question the numerous promises that were made in the ACA debate and foundations for a single-payer health system.
- Add Bernie Sanders claims about lives saved
- Medicaid promises during ACA debate
The canonical study on Medicaid was published in the New England Journal of Medicine in 2012. It found that Medicaid expansion led to a statistically significant reduction in "adjusted all cause" mortality. While the research was conducted after the passage of the Affordable Care Act, it was still before the implementation of Medicaid expansion, and consequently focused on pre-ACA expansions of Medicaid.
The small number of expansions is one of the primary reasons that the current study finds different results. The 2012 study was based on only three states, and furthermore, only found an effect in one of those states--New York. Drawing a conclusion for the entirety of the country based on the experience of only a single state, which is obviously not representative of the other 49 states in many ways, leaves ample space for over-interpretation of results.
Of course, the famous paper on the Oregon experiment in Medicaid expansion that found no effect is cited, yet appropriately caveated. The experiment wasn't large enough to detect small improvements and it was pretty soon after expansion, before there was enough time for improvement to occur.
Then there are two more recent studies on the ACA that found improvements in mortality of 3.6% for everyone and 9.4% for 55 to 64 year olds.
The authors of the new study focused on 8 of the 16 states that expanded Medicaid prior to implementation of the ACA. In true Brandeisian fashion, each state implemented its own version of Medicaid expansion--differing in the range of incomes affected, the years of implementation, the requirements, etc., offering researchers a panoply of variation from which to discern an effect.
They found that contrary to the other studies, those based on ACA and the one based on the three states only, that Medicaid did not reduce healthcare-amenable mortality, i.e. mortality that can be improved through better or more healthcare. This would include treating heart attacks or communicable diseases but exclude catastrophic car crashes and drownings, for example.
In fact, in none of the eight states that expanded Medicaid, was there an improvement in mortality that was determinatively produced by the expansion and wasn't just statistical chance. Reinforcing that result, without accounting for statistical variation, only three of the eight states even showed a reduction in mortality; the other five states had higher mortality rates after expansion.
To demonstrate that this study was an improvement on the previous three-state study, the authors even used their data and approach in an attempt to replicate the methods and results of its predecessor, and they succeeded. They found, just like the three-state study, that New York did show an improvement, and it was larger than the others. This provides support that the previous study was lacking in sample depth and rigor which led to a faulty conclusion.
Finally, the authors address the ACA studies and conclude that it is likely that the mortality effects are more significant for older populations but also that the other results may be partly driven by the non-expansion aspects of the ACA such as woodwork effects from the marketing campaign which would mean that mortality improvements could be achieved without expansion.
How Much has Zero Improvement Cost Us?
In 2013, the year of implementation of the ACA, approximately 18.7% of Americans were enrolled in Medicaid. In 2020, before Covid and its resultant legislation caused an increase in enrollment, that figure was 22.4%. So a cursory estimate is that the ACA led an additional 3.7% of Americans to enroll in Medicaid.
Multiplying that by costs per enrollee and the total population, the cost in 2020 for expanding Medicaid was $112 billion. Applying that same approach to every year from 2014 to 2021, a combined $700 billion has been spent without achieving any mortality benefits.
From 2014 to 2021, a combined $700 billion has been spent without achieving any mortality benefits.
Even if we assume a 10% improvement in mortality, higher than either of the studies that found an effect, this works out to saving 560 lives per year and spending $200 million per life saved1, 20 times the value used by the EPA in their recently proposed social cost of carbon update.
Had a 10% improvement in mortality been achieved, Medicaid expansion would have spent $200 million per life saved, substantially more than policy-makers deemed effective.
Footnotes
1Healthcare-amenable mortality rate is 45.419 per 100,000 enrollees. So 10% improvement would save 4.54 lives per 100,000 enrollees every year. Using approach described, in 2020, there were 12 million additional Medicaid enrollees because of ACA.Sunday, January 8, 2023
The Technocrats' Utopia
Main Takeaways
- Government projections for new policies consistently underestimate costs and overestimate benefits
- Details here for three such misses
- The media and the electorate should both exercise much more skepticism and scrutiny of these numbers when considering proposals from administrations and agencies.
On January 5th, the FTC announced that it will implement and enforce a rule that will prohibit non-compete clauses in labor contracts, nation-wide, retroactively, and proactively. To support their new regulation, the FTC claims that nationwide, earnings will increase by up to $300B. Informed citizens should be well to remember that, while the FTC is basing their estimates on what they consider the best possible evidence, it is often the case that the government projections turn out to be faulty, either because the research was too thin to properly predict the effects, because the research was just flat incorrect due to the bias of researchers, or because unforeseen circumstances undermined the basis upon which the research was conducted. A good example of the last possibility is projections on Medicaid enrollment for the ACA. I do not fault analysts for underestimating how politicized the ACA would be and that right-leaning states would eschew Medicaid expansion, the Obama administration would threaten to take away all of their current Medicaid funding, or how the Supreme Court would fall. Consequently, I don't fault them for over-estimating the number of people insured through Medicaid.
Another possibility is that once a policy is enacted, particularly those that redistribute benefits to groups, it often is expanded either in the number of people eligible or the amounts provided. This has been the case with the ACA and also student loan forgiveness. In this case, the analyst's job is not to predict how the policies will change, but to predict the effect of the proposed policy directly. It should be left to think tanks and pundits to extrapolate potential effects the political world may have.
Still it is worthwhile to consider some of the policies enacted, the effects they were supposed to have, and the effects they actually had.
Electronic Logging Devices for Truckers
In late 2015, the Department of Transportation's Federal Motor Carrier Safety Administration proposed a rule that would require truckers to use electronic logging devices to track their hours driving. The idea behind it was that truckers were driving more than they were able to, and becoming tired and more dangerous. Electronic logging would force them to limit their driving and thereby increase the safety of them and other drivers. In its original regulation, the Federal Motor Carrier Safety Administration claimed that EDS would save 26 lives and prevent 1,844 crashes per year, but in the two years after implementation, and before Covid, both fatalities and the number of crashes increased. The explanation is that when truckers are time-constrained, it causes them to drive faster to make up for the time they won't be on road, undoing any benefits of the reduced time. The Department of Transportation also didn't account for the price increases it caused.
Coverage of the 2017 protest to block
Nationalization of the Student Loan Industry
In 2010, President Obama signed a law that effectively nationalized the student loan industry, preventing any guaranteed loans from being offered by private banks, only from the government. The government estimates it will save $61 billion over 10 years. In fact, by 2022, the government had cost the government and taxpayers nearly $200 billion. Plus the costs of the debt forgiveness package Axios estimated to be $300 billion. GAO report on the student loan program in 2022"A Government Accountability Office (GAO) report released in July found the Department of Education predicted that student loans would generate $114 billion for the federal government; they instead lost $197 billion — a $311 billion error, mostly due to incorrect analysis." --Foundation for Economic Education
There are certainly benefits to subsidizing student loans, as a well-educated workforce will enhance the economy for everyone, but taxpayers deserve to have an accurate expectation of those costs and benefits to decide. And if time shows that that analysis was off, they should have the chance to reconsider.
ACA
In 2019, the average monthly premium per enrollee in the individual market was $515, up from $217 in 2011.
The three examples above illustrate that the analysis the government conducts and propogates through the media to support its regulations is all too often incorrect and underestimates costs and overestimates benefits. The media and the electorate should both exercise much more skepticism and scrutiny of these numbers when considering new proposals such as the ban on non-competes or environmental regulations.