In 1997, President Bill Clinton and the Republican-controlled U.S. Congress created the State Children’s Health Insurance Program, or SCHIP, as a joint federal-state effort to provide subsidized health coverage to uninsured children with incomes too high to qualify for Medicaid. During the following year, a number of states, including North Carolina, held spirited debates about how best to proceed with creating an SCHIP program. Some opted essentially to expand their Medicaid programs. Others created stand-alone programs with benefits and eligibility rules somewhat different from the Medicaid standard.

After a contentious debate, the N.C. General Assembly created an SCHIP program called North Carolina Health Choice. It was not a straightforward Medicaid expansion, but it did offer a benefits package more similar to Medicaid than to private plans or the state employee health plan, including vision, hearing, and dental coverage.1

Perhaps the most controversial topic of debate was the issue of “crowd-out.” Simply put, this effect occurs when government programs intended to help the uninsured end up subsidizing those who would otherwise have had private health coverage. The extent to which previous increases in government health insurance, such as the Medicaid expansions of the late 1980s and early 1990s, induced crowd-out is itself a controversial issue in health policy research. Credible estimates of crowd-out rates for previous Medicaid expansions (the percentage of new enrollees who would otherwise have retained or entered the private insurance market) have varied from as little as 15 percent to 22 percent2 to as much as 50 percent.3 Graph of North Carolina children with private health insurance, by income

The difficulty in making more precise estimates of crowd-out stems from two major problems: definition and data. First, some researchers have defined crowd-out solely as families with private coverage dropping it to enroll in a government program. This is too narrow a definition. Crowd-out also occurs when uninsured families or those leaving Medicaid would have purchased private coverage but did not because of the availability of another subsidized program. Crowd-out also occurs when employers, knowing that many of their workers now have access to free or virtually free insurance, change their own dependent coverage to minimize their costs.

Not surprisingly, all three forms of crowd-out are difficult to measure. Families making a conscious decision to drop their costly private coverage and enroll in free or nearly free government programs are not necessarily likely to tell a social services employee or interviewer about their decision. As for the other two types of crowd-out, they constitute a deviation from what would have occurred in the absence of the government program. This often requires the researcher either to project based on past trends or to compare changes in the eligible population to those in an ineligible population. Both techniques are risky. The past is not always a reliable predictor of the future. And ineligible populations differ from eligible ones in such important characteristics as income, frequency and tenure of employment, and mobility.

Still, not to make a serious effort to estimate the impact of crowd-out is a mistake. Unless states take steps either to discourage crowd-out or to reduce its budgetary impact, scarce dollars intended for the uninsured will be consumed by others.4 The truly needy will thus have reduced access to the health care “safety net” while families that might otherwise have remained self-sufficient will instead be enticed on to the public dole. Neither of these outcomes is fiscally and morally justifiable.

Crowd-Out In North Carolina

The administrators and proponents of the Health Choice program do not believe that crowd-out is a significant problem. For example, in its annual report for FY 2000, Health Choice stated that “crowd-out does not appear to be enough of an issue to justify in-depth analysis.” It cited an analysis by the Cecil G. Sheps Center for Health Services Research at UNC-Chapel Hill that estimated that the crowd-out rate in Health Choice was no higher than 3.5 percent and possibly as low as 0.5 percent.5

Unfortunately, the Sheps Center analysis is not a serious attempt to estimate crowd-out effects. It relies solely on two surveys of Health Choice enrollees, first in 1999 and then again in 2000. For the latter sample of 371 new enrollees, the program’s administrators make the startling claim that “there were a very small number among whom crowd-out could even be considered” because 63 percent said their children had just left the Medicaid rolls and another 10 percent had never had any form of child health insurance.6 Assuming that these responses are truthful, they do not exempt the children from crowd-out effects. Families leaving Medicaid, for example, would in the absence of Health Choice been faced with the prospect of 1) purchasing child health insurance individually or through their employers or 2) not insuring their children. Since most families with incomes below 200 percent of poverty who were not eligible for Medicaid have historically bought private insurance for their children, it is reasonable to conclude that many or most would have done so after 1998.

The Sheps Center analysis also makes no attempt to estimate employer crowd-out effects. Two national surveys of employers in 1998 and 1999 found that between 19 percent and 24 percent would consider either dropping their dependent coverage, making it more expensive, or reducing benefits in response to the implementation of SCHIP.7 Again, such an effect could occur not only for families with preexisting dependent coverage at work but also for 1) families just leaving Medicaid, 2) families just entering the state, or 3) families with newborns. The Sheps Center ignores all of these possibilities.

In reality, there is credible evidence to conclude that there is a significant crowd-out effect in Health Choice and that thousands of North Carolina families who would otherwise have purchased their own insurance have instead enrolled in the state’s program. From FY 1997 to FY 2000, according to Health Choice’s own reports to the federal government, private health insurance declined dramatically among children with household incomes less than or equal to 200 percent of the poverty line (the income cap for Health Choice), while children with incomes just above the cap actually saw increased private coverage (see chart on the previous page). 

Graph of coverage of CHIP-eligible North Carolina children, 1997 vs. 2000In raw numbers, the number of children with household incomes at or below 200 percent of poverty who were enrolled in private health plans dropped by nearly 56,700 from 1997 to 2000, or 30 percent. Some of these children had incomes low enough for Medicaid eligibility but the available data for 2000 do not provide such a breakdown. I used 1997 data from the state’s Task Force on Child Health Insurance to estimate the Medicaid-eligibles and subtract them from both the number with private insurance and the uninsured.8 The result was an estimate of slightly more than 116,000 SCHIP-eligible children with private insurance in 1997 vs. 81,000 in 2000. That still represents a 30 percent drop in three years. During the same three-year period, the number of privately insured children with household incomes between 201 percent and 300 percent of poverty still modest but too high to qualify for Health Choice actually rose by nearly 6 percent. Children with incomes above 300 percent of poverty saw a similar 6.5 percent increase in private health plan enrollment.

It is possible that, in the absence of Health Choice, private coverage among eligible children in North Carolina might not have grown as rapidly as it did among higher-income children. It is even possible that private coverage would have declined. But a 30 percent drop in three years is highly unlikely. To put it in proportional terms, I estimate that 64 percent of N.C. children with SCHIP-level incomes were privately insured before the creation of Health Choice. By 2000, private coverage had fallen to 39 percent (see chart below). The uninsured rate fell, too, suggesting that thousands of children did indeed gain access to health insurance because of Health Choice. But many thousands of additional families simply traded private coverage they would have paid for on their own for government insurance that was free or nearly free of charge.

As we have seen, estimating the precise crowd-out rate for previous Medicaid expansions has proven difficult. The same is true for Health Choice. But a reasonable range can be constructed by assuming a best-case and worst-case scenario for the extent of private insurance coverage had there been no Health Choice. Although child health insurance data by income is not available for North Carolina before 1997, overall trends in child health coverage in the state provide a useful benchmark. The largest three-year drop since the data have been collected occurred from 1991 to 1994, when the rate of private child health insurance fell by 8.4 percent.9 If we assume that three times as large a drop in private coverage among SCHIP-eligible children would have happened from 1997 to 2000 in the absence of Health Choice a dubious assumption, at best, since no drop occurred at all for other children then an estimated 18,738 enrollees in Health Choice in 2000 would still have had private coverage without the program. This yields a crowd-out rate of 25 percent of the 74,145 Health Choice enrollees.

On the other hand, if we conservatively assume that the actual number of privately insured children would have remained constant from 1997 to 2000 for SCHIP-eligible children (even as private enrollment for other children grew by 6 percent) then nearly 35,000 enrollees would have otherwise had private coverage, a crowd-out rate of 47 percent. The resulting crowd-out range is strikingly similar to that found for previous Medicaid expansions (15 percent to 50 percent, as noted earlier).

The middle of this range is 36 percent. It is reasonable to assume, in other words, that more than a third of the $108.9 million in state and federal funds budgeted for North Carolina Health Choice in federal FY 2001 (the state share is $28.8 million), will be spent on children for whom the program was not intended.10 While not “wasted,” this money can at least be characterized as “misdirected.” It is money that cannot be used to enroll more truly uninsured children.

Policy Implications

Despite the existence of a yawning budget deficit for FY 2001-02, Gov. Mike Easley has proposed a $21 million increase in state funds for Health Choice over the next two years. This move is justified, according to the governor, because it will draw down $57.6 million in federal funds and thus allow for a 23 percent increase in Health Choice enrollment.11 But given the existence of a significant crowd-out effect for the program so far, and likelihood that it may well increase as more employers and employees become aware of the existence of free or nearly free child health insurance, such a spending increase would be hasty. Before putting additional state dollars into Health Choice, policymakers should consider changing the program to maximize its impact on the truly uninsured and minimize the cost both in taxpayer dollars and family self-sufficiency.

For example, the state might consider reinstating the original six-month waiting period for enrollment in Health Choice. In the original legislation, the waiting period fell to two months after Health Choice had existed for half a year. At last count, 17 states had waiting periods of at least six months.12 Research indicates that a longer waiting period can have a significant impact on the propensity of employers and employees to engage in crowd-out activities.13

Another idea would be to emulate 21 other states and charge monthly premiums, based on income and family size, rather than the current annual enrollment fee of $50 for one child and $100 for two or more. In all but one of the monthly-premium states, the annualized out-of-pocket cost for SCHIP families is higher than in North Carolina, thus providing more of a disincentive to choose government coverage rather than private coverage.14 Moreover, the payment of monthly premiums will make Health Choice recipients more like the rest of North Carolina families, easing their transition to self-sufficiency.

Finally, policymakers should seriously considering expanding North Carolina’s refundable tax credit for private child insurance. Currently, families purchasing health plans for their children with after-tax dollars can take a $300 credit (if their income is less than 225 percent of poverty) or a $100 credit if their income is less than $100,000. While some have advocated eliminating the $13 million tax credit in order to expand Health Choice15, the better policy would be the opposite. Doubling the credit to $600 and $200, respectively, would provide an additional incentive for families to opt for private coverage, while also providing tax fairness to those SCHIP-eligible families who have remained off the dole. The state should also take steps to publicize the tax credit, similar to what has already occurred with Health Choice.

John Hood, President 


  1. John Wagner, “Deal reached to insure children,” The News & Observer (Raleigh), April 29, 1998.
  2. Harriette B. Fox and Margaret A. McManus, “The Potential for Crowd Out Due to CHIP: Results from a Survey of 450 Employers,” Fact Sheet No. 3, Maternal & Child Health Policy Research Center, Washington, D.C., March 1998.
  3. David M. Cutler and Jonathan Gruber, “Does Public Insurance Crowd Out Private Insurance?” Quarterly Journal of Economics, Vol. CXI, Issue 2, May 1996, pp. 431-466.
  4. Eugene Lewit and Linda Schuurmann Baker, “Child Indicators: Health Insurance Coverage,” The Future of Children, Center for the Future of Children, New York, N.Y., Vol. 5, No. 3, Winter 1995.
  5. Annual Report 2000, N.C. Health Choice, N.C. Department of Health and Human Services, January 4, 2001, p. 12.
  6. Ibid.
  7. See Fox and McManus; also Jack A. Meyer, et. al., “Crowd-Out Under CHIP,” Economic and Social Research Institute, Washington, D.C., April 1999.
  8. Final Report, Task Force on Child Health Insurance, N.C. Department of HHS, November 1997, pp. 2-3.
  9. Health Insurance Coverage Status and Type of Coverage by State Children Under 18: 1987 to 1999,” U.S. Census Bureau.
  10. Spending data obtained from Annual Report 2000, p. 22.
  11. North Carolina State Budget, 2001-2003, Summary of Recommendations, March 12, 2001, p. 108.
  12. “SCHIP Crowd Out Provisions,” National Conference of State Legislatures, April 2000,
  13. See, for example, Fox and McManus, p. 1.
  14. SCHIP Cost Sharing,” NCSL web site, April 2000.
  15. Barbara Solow, “Frozen Out,” The Independent Weekly, February 28, 2001, p. 18.