Download this memo (pdf)
A central pillar of the recently enacted Patient Protection and Affordable Care Act is the individual mandate, the requirement that all individuals for whom insurance is affordable purchase such coverage or pay a tax penalty. Yet this is also one of its most controversial elements. In recent public opinion polls, the individual mandate is rated as one of the least popular elements of the new health law. And recent court decisions on the constitutionality of the individual mandate have reached mixed conclusions, with two courts upholding the mandate and two striking it down.
So what happens to health care reform if the mandate is repealed? And is there a reasonable alternative? This issue brief answers both of these questions. In particular, I consider the two most-discussed alternatives to the mandate and estimate their impact on insurance coverage, public sector costs, and insurance prices. I find that both alternatives significantly erode the gains in public health and insurance affordability made possible by the Affordable Care Act.
Reform with and without an individual mandate
We have a fairly good sense of how the world will look if health care reform includes the individual mandate. Both the Congressional Budget Office and independent modelers such as myself find that the majority of the uninsured would be covered. CBO and I both estimate that Affordable Care Act will cover about 60 percent of those who would be uninsured absent the law. We both find that there would be a very modest reduction of employer-sponsored insurance, that premiums in the nongroup insurance market for the same quality product would fall, and that there would not be much effect on premiums in the employer-provided insurance market.
These estimates are consistent because we have a clear example to draw on in this case, the state of Massachusetts, which four years ago enacted a plan that is very similar to the new federal health reform law. In Massachusetts we have seen more than 60 percent of the uninsured gain coverage with little effect on employer-sponsored insurance premiums. We have seen a steeper drop in nongroup premiums that estimates suggest for the Affordable Care Act, however. According to insurance industry figures, nongroup premiums have fallen by 40 percent in Massachusetts while rising by 14 percent nationally.
This much steeper drop in Massachusetts arises because the state has also given us a glimpse of what the world would look like if the mandate were stripped from the Affordable Care Act. In the mid-1990s, Massachusetts along with several other northeastern states passed insurance market reforms similar to those in the Affordable Care Act, eliminating or restricting the ability of insurance companies to discriminate against the ill either in prices or coverage exclusions. The result in each state was very high nongroup insurance prices as insurance companies worried that only the sick would enroll in insurance and priced their products accordingly.
We do not, however, have an example of a state that has included the other major element of the Affordable Care Act—extensive subsidies for low-income individuals to buy insurance. This will offset to some extent the “adverse selection” that drives up premiums in the nongroup market by bringing some healthier individuals into the market. The extent of such offset, however, is unclear. CBO estimates that removing the individual mandate from the new federal health law will cut the number of individuals newly insured in half (from 32 million to 16 million), while I estimate that it will cut the number of newly insured individuals by three quarters (from 32 million to 8 million). CBO estimates that the reduction in employer-sponsored insurance will double with no mandate; I estimate that it will triple. CBO estimates that premiums in the nongroup market will rise by 15 percent to 20 percent; I estimate they will rise by 27 percent. Finally, CBO estimates that removing the mandate would lower net government spending by $47 billion in 2019, or roughly 25 percent of the costs of the policy. I estimate a cost reduction of 30 percent.
So there is agreement between CBO and myself that a bill without the individual mandate will cover significantly fewer persons, with more erosion of employer insurance, and lead to significantly higher premiums. Moreover, we both agree that removing the mandate would significantly lower the “bang for the buck” of health policy, reducing coverage by 50 percent to 75 percent while only lowering costs by 25 percent to 30 percent. But there is more uncertainty and divergence in the estimates. And this is a key point to highlight about removing or replacing the individual mandate—it will raise our uncertainty about what health care reform can accomplish. One advantage of the individual mandate is that we have an example to build on; alternatives put us in a much less clear world.
If we were to replace the mandate, then there are two common alternatives that have been proposed. One is “auto-enrollment,” whereby individuals would be automatically enrolled in insurance as a default but could “opt out” if they decide they don’t want coverage. This has been called a “soft mandate” because it doesn’t force individuals to buy insurance, but it does force them to take affirmative action to avoid coverage. If some of the lack of enrollment in pensions (or health insurance) is due to “inattention,” then such a policy could greatly increase coverage.
This alternative is inspired by research by David Laibson and Brigitte Madrian at Harvard University along with various collaborators, which shows that such default changes in the context of so-called defined-contribution 401(k) pension plans can significantly increase participation in such plans, reducing the number of nonenrolled employees from 50 percent to 10 percent. Evidence from a broader universe of firms from money manager Fidelity Investments suggests that the effect is smaller, which could be due to a higher willingness to “opt out” at smaller firms. Fidelity finds that auto-enrollment raises participation from 53 percent to 81 percent. That is, of the 47 percent of employee that choose not to voluntarily enroll in 401(k) plans, 19 percent choose to opt out of auto-enrollment.
Applying this finding to the context of health insurance is difficult, but several considerations suggest that we would see a larger opt-out rate for health insurance than for 401(k) plans. First of all, employers actively encourage a broad cross-section of employees to participate in 401(k) plans because it is critical to meet nondiscrimination tests that allow them to tax defer the 401(k) contributions of higher-income employees. There is no such need for employers to encourage participation in health insurance plans, where nondiscrimination rules appear to be nonbinding. Indeed, employers should actively oppose auto-enrollment for health insurance: even if it encourages healthier employees to join, total employer spending rises.
Second, health insurance enrollment is a decision to which employees, particularly young employees, have already given much more consideration than to 401(k) enrollment. The largest auto-enrollment effects are found for young employees, for whom retirement is distant and so who probably weren’t considering 401(k) accounts before being auto-enrolled. These same young employees will have given much more consideration to the near-term decision about whether or not to insure.
Indeed, Fidelity data show that among 20- to 29-year-old employees, only 30 percent sign up for a 401(k) without auto-enrollment, yet only 23 percent opt out when auto-enrolled, a very large effect. Yet among workers 20 to 29 years old who are offered health insurance, 88 percent enroll today. Clearly, this is a decision that young workers are taking more seriously—and as a result the “inattention” that results in auto-enrollment increases is likely to be a much smaller consideration. Furthermore, the dollars at stake here are potentially much larger, particularly for low-income workers, so once again they will be paying much more attention, and therefore will be much likelier to opt out.
Third, a recent study from the same team of Harvard authors shows that auto-enrollment may not work well when the default option is one that is very undesirable. They study a British firm that auto-enrolled employees into a very high contribution product (12 percent of pay), and one that was dominated because employer matching only began after 12 percent of pay. They found that only 25 percent of employees remained auto-enrolled in this option after one year.
Fourth, as CBO points out in its 2008 discussion of key health issues, if you “overcontribute” to 401(k) pension plan accounts then you can get the money back with a small penalty. With health insurance, any premiums paid that you did not wish to pay are lost forever. This will further cause individuals to pay more attention to the opt-out decision and opt out more frequently if they don’t want the insurance.
Fifth, auto enrollment for health insurance raises an additional difficult issue: what to do about dependents. Pension enrollment is individual, but of the uninsured individuals offered insurance coverage today, 56 percent are dependents, not the employees themselves. Thus, auto-enrollment will not make much of a dent in the uninsured unless it extends to dependents. At the same time, 31 percent of workers who are offered employer-sponsored insurance also have a spouse that is offered insurance. This implies that inattention to auto-enrollment could lead to duplicative double coverage with inattentive individuals paying excessive bills that they may not notice until several months later.
Finally, only about one-third of the uninsured are actually offered employer-sponsored insurance in which they can be auto-enrolled. Individuals who auto-enroll outside of employer-sponsored insurance present an entirely new set of challenges. In principle, such individuals could be auto-enrolled based on past tax return information. But charging individuals premiums for insurance for which they did consciously enroll will raise a host of very difficult political and potentially constitutional issues. Moreover, given lags in tax data, many individuals would be misclassified across subsidy levels, which would either lead to difficult issues of reconciliation or higher government spending.
As a result of these limitations, I estimate that auto-enrollment will be much less effective in the health insurance context. I find that if the Affordable Care Act were stripped of the individual mandate but instead accompanied solely by auto-enrollment of individuals who are offered insurance into single coverage, then only 1.1 million uninsured would gain coverage. Auto-enrollment that included dependents (with the associated double-coverage issues) would extend the gains to another 1.7 million uninsured. If the government could successfully auto-enroll individuals into free public insurance, then that would be more effective, subject to individual concerns about having to pay for the free insurance if their income has gone up. I estimate that adding this feature would extend the gains to another 8.1 million uninsured. Finally, if the government could auto-enroll individuals in the exchanges, this would add another 3.7 million uninsured.
Thus, in the most generous case of full auto-enrollment (including those not offered employer-sponsored insurance), I estimate that:
- Twenty-four million persons would gain insurance coverage, as opposed to the 32 million that would gain coverage with the mandate. Partly this is because the erosion of employer-sponsored insurance would double if there were no individual mandate in the new health reform law; that is, twice as many individuals would lose their employer coverage as would under the mandate.
- Since young healthy individuals would opt out of coverage, premiums in the nongroup market would rise by about 11 percent.
- Strikingly, though, I estimate that net government costs would not fall at all. This is because such a larger share of enrollment under this alternative comes through government-sponsored insurance. Under the Affordable Care Act with the individual mandate, about half of the net gain in coverage is in public insurance; under auto-enrollment 80 percent of the net gain in coverage comes through public insurance. That is, under this option, 8 in every 10 newly insured are gaining coverage through government-provided insurance.
The bottom line is that losing coverage from employer sponsored insurance or the unsubsidized exchange, and making it up through fully publicly financed care, raises costs. So the government spends the same amount of money while covering two-thirds as many individuals and raising premiums in the nongroup market by more than 10 percent.
Alternative: Late enrollment penalties
An alternative to both auto-enrollment and the individual mandate is to allow voluntary opt-in to insurance under the Affordable Care Act, but then impose a penalty for late enrollment. Such an approach is followed by the Medicare Part D prescription drug plan, and enrollment in that plan was very rapid. In work with Gary Engelhardt of Syracuse University, we estimate that about half of the 28 percent of elders with no prescription drug coverage signed up for the program within one year of its introduction. Similarly, Medicare Part B, which covers outpatient costs for enrollees in Medicare, combines auto-enrollment with a late enrollment penalty (10 percent of premiums for each year of delay), and enrollment is virtually universal despite the fact that enrollees have to pay 25 percent of the premiums.
There are once again, however, a number of reasons to think that these findings overstate the impact that late enrollment penalties might have under the new health reform law without the individual mandate. Most importantly, we do not have a benchmark for what enrollment in these programs would be absent the late enrollment penalty. So while we know that enrollment under these late penalty regimes is high, we don’t know what it would be if the program were strictly voluntary. Both Medicare Part D and Part B are heavily subsidized, with the government picking up approximately 75 percent of the insurance costs. For many individuals who would be targeted for late enrollment penalties outside of employer-sponsored insurance under the Affordable Care Act, subsidy rates will be much lower, if not zero.
Moreover, the seniors being assessed these penalties particularly value the insurance they are receiving and would likely enroll at high rates even absent penalties. This will not be true for many individuals of all ages (and their dependents) who don’t voluntarily choose to enroll under the Affordable Care Act without the individual mandate. This conclusion will be further enforced by the social dynamic around universal enrollment in Medicare at age 65.
The impact of late enrollment penalties under the new health reform act would vary with the severity of the penalty, which could range from a small financial penalty to more severe penalties. Paul Starr of Princeton University suggests that if individuals do not sign up for insurance at the first opportunity under the new law, they would be barred from purchasing insurance in the new health insurance exchanges for a period of five years. This could leave individuals without access to nondiscriminatory insurance markets or insurance subsidies should they need them.
Of course, there is a tradeoff between the severity of the penalty and the realism of imposing it consistently. It seems highly unlikely that the federal government would be willing to tell a 30-year-old individual with cancer that they can’t get insurance coverage because they didn’t sign up when they were 27 years old—or that they have to pay some very large amount of money in the same situation.
Moreover, any solution such as Starr’s places the very viability of reformed nongroup insurance markets at risk. It is the young and healthy who would take their chances with a late enrollment penalty rather than sign up for insurance that they don’t fully value. As these young and healthy individuals leave the exchanges, they will raise prices for those left behind, causing even further exit—and potentially unraveling the entire market.
There is even more uncertainty, as a result, in estimating the impact of late enrollment penalties. My best estimate is under a late enrollment penalty regime:
- Twelve million individuals will gain coverage, or only about one-third of those who we would expect to see enrolled under the individual mandate.
- Premiums in the exchange would rise about 20 percent relative to the mandate case as the healthy exit the exchanges.
- Government costs would fall by only about 25 percent, however, since subsidies would still be provided to the sicker enrollees who stay behind.
- Therefore, instituting late enrollment penalties reduces coverage by 65 percent for only a 25-percent reduction in government spending, while raising premiums by 20 percent for those who do want to buy non-group insurance.
Modeling the impact of fundamental health reform is a balancing act between leaning on what is known and modeling what we need to know. In the case of the new health reform law, that balancing act was greatly assisted by the experience of Massachusetts, which provides a great case study of the world with reformed insurance markets and an individual mandate. Once we move away from the individual mandate, our estimates become considerably more uncertain.
Nevertheless, several lessons are clear from the exercises described in this paper. First, no alternative to the individual mandate can cover more than two-thirds as many uninsured as the Affordable Care Act does as passed by Congress and enacted into law. Second, no alternative to the mandate saves much money—even removing the mandate altogether, which cuts the number of uninsured covered by 50 percent to 75 percent but only reduces government spending by 25 percent to 30 percent. Strikingly, broad and aggressive auto-enrollment, which I estimate to cover two-thirds as many uninsured as the mandate, costs just as much because the coverage comes almost exclusively through auto-enrollment into public insurance. Finally, any alternative imposes much higher costs on those buying insurance in the new health insurance exchanges as the healthiest opt out and the less healthy face increased premiums.
Jonathan Gruber is a professor of economics at the Massachusetts Institute of Technology
Download this memo (pdf)