Expansions in public health insurance programs are designed to offer a safety net to vulnerable Americans unable to obtain basic health insurance and regular access to medical care. In recent years, expansions of state Medicaid programs under SCHIP have reduced the number of uninsured children in this country.
But such efforts come with a price. For every expansion that covers those previously uninsured, some people may drop coverage, or not seek coverage they are eligible for to participate in the public program. This phenomenon is known as "crowd-out" — the public assistance program crowding private insurance out of the market.
Policy-makers are concerned about crowd-out because it limits the impact of public coverage expansions. When crowd-out occurs, some of the scarce resources are used to cover people who would have purchased private insurance anyway.
The Robert Wood Johnson Foundation (RWJF) recently asked researchers from the State Health Access Data Assistance Center (SHADAC) at the University of Minnesota in Minneapolis to study the incidence of crowd-out.
The problem, says Gestur Davidson, PhD, senior research associate at SHADAC and the lead author of the RWJF’s Synthesis Report on crowd-out, is that no one really knows the best way to measure public program crowd-out, and many efforts at doing so have used different definitions and methods, thus making comparisons difficult. "Quite a few conceptualizations of crowd-out are used in the literature, and then, within individual ways of defining the concept, there can be some differences in the way it gets measured," he notes.
There are three ways that crowd-out can occur:
• People drop private coverage for public. A person or family drops private insurance, either employment-based or individually purchased, to enroll.
• A public program enrollee refuses an offer of private coverage. Someone with public coverage refuses an employer’s offer of insurance, which that person would have accepted in the absence of the public program. This phenomenon is known as "within enrollment" crowd-out.
• An employer changes coverage offerings in response to the existence of a public program. An employer changes elements of its insurance offerings, for instance, dropping dependent coverage or increasing employee premiums, resulting in an employee losing or deciding to drop private coverage and enroll in a public health insurance program.
Researchers measure crowd-out by examining changes in public and private coverage after creation or expansion of public programs. It is difficult, however, to determine whether changes in private coverage are directly related to public-program expansions (i.e., wouldn’t have occurred if public program expansion did not exist). Estimates are imprecise and vary greatly depending on type of coverage expansion and assumptions, methods, and data used, as well as time period covered by the study.
Researchers often measure crowd-out by observing total change in health insurance coverage occurring over a period of time due to all possible causes including those independent of the expansion of a public program and those related to it. Then, using sophisticated statistical models, they construct likely scenarios to estimate how much private and public insurance coverage would have changed in the absence of the expansion. Their estimate of crowd-out, presented as a range rather than a single number, is the difference between these model-based estimates of the changes that would have occurred without the program and those that did occur when the program was introduced.
However, researchers often use different definitions of crowd-out, which contributes to confusion when differing estimates of crowd-out are compared. The most common definition compares the reduction in the share of the population with private coverage to the increase in the share of the population with public coverage due to the expansion. A less restrictive definition focuses on the amount of crowd-out that occurs throughout the public program following an expansion — not just among the newly eligible population. This definition usually produces lower estimates of crowd-out than the previous one. Still other studies compare the decline in private coverage associated with program expansions to the overall decline in private coverage during the period, rather than to the increase in public coverage. This approach tends to also produce lower estimates of crowd-out.
Some estimates focus on the extent to which program expansions reduce the number of uninsured, but this broad definition can, and often does, produce a larger crowd-out estimate than the narrow definition, which focuses on how much private coverage fell as a result of the expansions. Although the variety of definitions and methodologies can be confusing, it’s important to understand that different definitions are used to answer different questions, Mr. Davidson says. "The different definitions/conceptualizations, in fact, serve different purposes. And one could say they yield different perspectives on how important crowd-out might actually be."
For example, using one definition, a researcher might determine that a certain number of people enrolled in a newly expanded public assistance program would have access to private insurance. How-ever, that number might include both people enrolling in the new program with less strict requirements and people dropping coverage who are only eligible for the older portion of the plan with stricter enrollment requirements.
Another strategy, he adds, is to express how much of the total public program enrollment growth over the period might have been the result of people who otherwise would have had some private insurance. This method is likely to show less crowd-out than the previous one, but researchers may have specific reasons for using this method. "Policy-makers examining different studies need to be aware of the different ways that crowd is measured and defined so that they don’t make inadequate comparisons," Mr. Davidson says.
According to the report, the potential for crowd-out is greater among families with income above the federal poverty level that are more likely than poor families to have private insurance coverage. Crowd-out rates also may be higher if whole families can enroll together in public coverage.
Crowd-out rates likely will change over time, influenced by the economy, labor market conditions, characteristics of private coverage, and attitudes toward public coverage. Examining the raw data also does not tell the entire story behind public program crowd-out, Mr. Davidson adds. "In some cases, people who drop their private insurance to enroll in the public program can substantially lower out-of-pocket costs since their premium contributions — if they are participating in an employer-sponsored program or, if they are directly purchasing private, nongroup insurance coverage — can actually be quite high, relative to their available income. Moreover, many who drop private insurance might gain in the services and benefits covered as well as access to care that they now have with the public program."
This is an important point for policy-makers worried about crowd-out to consider, he notes. Crowd-out can mean important benefits to some with very low incomes. "I’m sure policy-makers do not like to think that those who could afford private insurance are taking some of the available public monies. But it is not clear how much of the crowd-out that public programs are experiencing comes from those who could easily afford good private coverage and don’t buy it, but enroll in the public program instead."
Unfortunately, there’s no good way to measure the last phenomenon. "Even if we had a single definition of crowd-out and method for measuring, it is technically so very difficult to estimate the amount of crowd-out that might be present," he says. "You cannot look at an individual case that enrolls — knowing what insurance they had just before enrolling — and confidently predict that they would have had private insurance for all or even good parts of their enrollment in the public program if the public program had not existed. People drop private insurance coverage all the time, for all kinds of reasons; enrolling in a less expensive public program is just one of them."
Just because someone drops private insurance and later enrolls in a public program does not mean the program was the motivating factor. "There is a similar problem with identifying cases of no crowd-out from those who are uninsured and then enroll in the public program," he says. "Some of them might have enrolled in private insurance without the public program’s availability. In other words, knowing where someone came from, doesn’t tell you where they would have been."
States most commonly have used waiting periods and, more recently, cost-sharing as tools to limit crowd-out in SCHIP, the authors stated. While, no real evidence exists on the effectiveness of waiting periods, logically they are likely to reduce some forms of crowd-out.
To ease problems for families facing serious hardships, some states exempt families from waiting periods if they have high medical expenses, experience involuntary loss of coverage, or purchase coverage in the individual market. Cost sharing also may limit crowd-out by reducing the difference in out-of-pocket costs between public and private coverage, but may discourage the uninsured from enrolling in or using health benefits offered by public programs. Measures to control crowd-out, though hard to evaluate, likely will result in some reductions but may discourage the uninsured, those the program expansions are designed to help, from participating.
"There is an inherent trade-off between targeting efficiency [keeping crowd-out at low levels] and making significant inroads in reducing the number of uninsured," Mr. Davidson says. "You could define a program that could probably achieve very low levels of crowd-out — say, restricting it to only those who are currently unemployed or who have very low cash reserves and no possibility of using a COBRA program. That would yield a very low crowd-out, but also have very low numbers enrolling compared to the total number of uninsured."
Policy-makers should consider the trade-offs between limiting crowd-out and covering the uninsured. Crowd-out limits the impact of public coverage efforts, but lower-income families enrolling in public programs may gain a more stable source of insurance. While anti-crowd-out measures will probably reduce the substitution of public for private coverage, they also may lower participation in public programs and raise equity concerns. They also can be costly and require substantial effort to implement. To achieve meaningful reductions in the number of uninsured, some amount of crowd-out seems inevitable.
"This trade-off will exist, to some extent, no matter what the specific policy approach is to reducing the number of uninsured. It is not restricted to the direct expansions of public programs," Mr. Davidson says. "Programs that would provide refundable tax credits to individuals to purchase private, nongroup coverage would also entail public dollars [in the form of taxes not collected], [thus] displacing private dollars, which is the essence of crowd-out. There is no getting around it," he notes.
(To see the RWJF report, go to www.policysynthesis.org.)