SHA DAC S TATE H EALTH A CCESS February 2002/Issue 4 D ATA A SSISTANCE C ENTER A Health Data Resource for States What is Behind the 8 Percent Drop in Uninsurance: Changes in CPS Health Insurance Measurement and the Effect on State Policy In March 2000, the Census Bureau added a • SCHIP verification component to the Current • State specific health insurance Population Survey (CPS) health insurance programs module. For the first time, respondents were • CHAMPUS/VA/Military Health Care asked directly whether they were uninsured. • Indian Health Service The Census Bureau found that using the veri- • Private Insurance fication items results in a more accurate This is the "residual" approach to measuring estimate of the rate of uninsurance. Without health insurance coverage. Respondents were the verification question, the 1999 estimate classified as being uninsured if they did not of the number of uninsured was 42.6 million; answer "yes" when asked if they had any of the with the verification question, the 1999 esti- various types of insurance. (Respondents with mate of the uninsured population was 39.3 only Indian Health Service were not consid- million, a decrease of 7.7 percent.1 The ered insured.) change is attributable to a modification in how coverage is measured and does not PROBLEMS WITH THE RESIDUAL represent an actual reduction in the rate of APPROACH uninsurance. This issue brief describes the change in the Census Bureau's approach to Research conducted by the Urban Institute estimating the number of uninsured, and the and others (the Center for Studying Health effect of the change on state policy. System Change, for one) found that the residual approach was problematic.2 The THE RESIDUAL APPROACH Urban Institute's National Survey of Prior to March 2000, the CPS March America’s Families began asking respondents Supplement health insurance module did not directly whether they were uninsured. They directly ask survey respondents whether they found that some of the people who answered were uninsured. Respondents were asked if "no" to each type of health insurance were, in they had any of the types of health insurance fact, insured. This direct uninsurance "veri- University of Minnesota listed below during the past year. fication" question was followed by another School of Public Health • Medicare opportunity for the respondent to declare • Employer-based what type of insurance she or he had. Sponsored by a grant • Medicaid from The Robert Wood Johnson Foundation. DIRECT VERIFICATION QUESTION EFFECT ON STATE HEALTH POLICY In an attempt to replicate the Urban Institute's study The changes to the CPS will affect the implementation for use in the CPS, the Census Bureau, for the first of state health policy in at least three ways. First, states time, added a direct verification question to the March that use the CPS data to budget and forecast public pro- 2000 demographic supplement. The Census Bureau gram participation will have to adjust their forecasting wanted to evaluate this question before adopting it for models. Second, the new CPS numbers will be used to use in making health insurance coverage estimates. determine the state's federal allocation for the State The Census Bureau's research found that when the Children's Health Insurance Program (SCHIP). direct verification questions were used, about 7.7 per- Finally, the changes in the CPS will affect state policy cent of those previously classified as not having health by putting another number into the mix for estimating insurance reported that they were, in fact, insured. the number of uninsured. The Census Bureau used the lower figure in its 2000 Adjustment of forecasting models: CPS data are often estimates of the number of uninsured. used for estimating the number of people who are Had the verification question methodology been used eligible for a public program. The changes in the to produce the health insurance report covering calen- measurement will necessitate changes in how states use dar year 1999, the estimated number of people without the CPS for budgeting and forecasting. For example, health insurance in 1999 would have been reported as if a state had estimated 100,000 eligible people for a 39.3 million rather than 42.6 million. The difference program, the new estimate, on average, would decrease does not represent an actual decline in the number of that estimate to 92,300 eligible people (7.7 percent). people without health insurance, but rather means that If the state had a 50 percent take-up rate, with 50,000 past estimates of the uninsurance rate were biased people enrolled, the take-up rate for forecasting should upward (more people were considered uninsured than be adjusted to 54 percent. actually were uninsured). The verification question Federal funding of SCHIP: The State Children's Health corrects for the bias. Insurance program uses the Current Population Survey's health insurance data to allocate federal funds STATE COVERAGE ESTIMATES to states. The higher the number of uninsured, low- The Census Bureau will be adjusting its state estimates income children in a state, the more money the state of health insurance coverage based on the results from receives for its SCHIP program. The decreasing num- the verification question. The two-year average ber of uninsured will give more weight to other decrease in each state's uninsurance rate is included in components of the allocation formula (namely, the Table 1. The uninsurance rate for all states declined. number of low income children), and could cause The largest decline was 14.5 percent in Rhode Island states that experience larger relative declines in the and the smallest decline was 2.9 percent in Wyoming. number of uninsured children to receive less SCHIP Nineteen of the fifty-one states (including the District money. of Columbia) experienced a statistically significant per- centage decline greater than the national two-year aver- age of 8.1 percent. This is strong evidence for a state effect in the decline of uninsurance due to the addition of a verification item. The states with statistically sig- nificant differences are denoted in Table 1. Table 1. Two-Year Average State Reduction in the Uninsurance Rate With Verification Compared to US Average Reduction: 1999-2000 States Two-Year Average Difference Between Uninsurance Rate Two-Year Average State Average With Verification Reduction Reduction and US (1999-2000) (1999-2000) Average Reduction United States 14.15% 8.07% Alabama 13.35% 7.27% -0.80% Alaska 18.80% 3.61% -4.46%** Arizona 18.05% 6.88% -1.19% Arkansas 14.20% 5.95% -2.12%* California 18.55% 7.30% -0.77% Colorado 14.25% 9.21% 1.14% Connecticut 8.45% 12.82% 4.75%* District of Columbia 14.05% 10.78% 2.71% Delaware 10.15% 10.51% 2.44% Florida 17.65% 6.39% -1.68%** Georgia 14.80% 6.36% -1.71% Hawaii 10.20% 11.80% 3.73% Idaho 16.85% 8.58% 0.51% Illinois 13.25% 9.53% 1.46% Indiana 10.70% 11.41% 3.34% Iowa 8.15% 11.91% 3.84% Kansas 11.60% 4.09% -3.98%** Kentucky 13.05% 6.33% -1.74% Louisiana 20.40% 5.21% -2.86%** Maine 11.20% 8.11% 0.04% Maryland 10.35% 11.51% 3.44% Massachusetts 9.30% 12.61% 4.54%** Michigan 10.00% 9.54% 1.47% Minnesota 8.15% 10.26% 2.19% Mississippi 14.35% 8.22% 0.15% Missouri 8.75% 12.93% 4.86%** Montana 18.15% 5.21% -2.86%** Nebraska 10.00% 7.37% -0.70% Nevada 16.95% 12.90% 4.83%** New Hampshire 8.05% 12.44% 4.37% New Jersey 12.25% 11.91% 3.84%** New Mexico 23.95% 5.88% -2.19%** New York 15.15% 8.75% 0.68% North Carolina 13.65% 7.17% -0.90% *p<.05 **p<.01 Table continues on next page SHA DAC S TATE H EALTH A CCESS D ATA A SSISTANCE C ENTER Table 1. Two-Year Average State Reduction in the Uninsurance Rate With Verification Compared to US Average Reduction: 1999-2000 (continued from Page 3) States Two-Year Average Difference Between Uninsurance Rate Two-Year Average State Average With Verification Reduction Reduction and US (1999-2000) (1999-2000) Average Reduction North Dakota 11.40% 6.08% -1.99% Ohio 10.55% 7.48% -0.59% Oklahoma 17.85% 5.61% -2.46%* Oregon 13.80% 5.46% -2.61%* Pennsylvania 7.95% 12.25% 4.18%** Rhode Island 6.00% 14.46% 6.39%* South Carolina 13.75% 10.57% 2.50% South Dakota 11.30% 7.44% -0.63% Tennessee 10.30% 9.24% 1.17% Texas 21.75% 5.23% -2.84%** Utah 13.30% 6.67% -1.40% Vermont 10.85% 8.29% 0.22% Virginia 12.85% 8.22% 0.15% Washington 13.55% 11.56% 3.49%* West Virginia 14.95% 9.10% 1.03% Wisconsin 8.55% 6.47% -1.60% Wyoming 14.70% 2.94% -5.13%** Source: 2001 and 2002 Current Population Survey *p<.05 **p<.01 Notes 1 Rajan, Shruti, Stephen Zuckerman and Niall Brennan, "Confirming Insurance Coverage in a Telephone Survey: Evidence from the National Survey of America's Families," Inquiry, Fall 2000 University of Minnesota (Vol. 37, No. 3), 317-327. Division of Health Services Research and Policy 2 Nelson, Charles T., and Robert Mills. 2001. The March CPS Health Insurance Verification and Its Effect on Estimates of the Uninsured. US Census Bureau: Washington DC. http://www.cen- 2221 University Avenue sus.gov/hhes/hlthins/verif.html. Suite 345 Minneapolis, MN 55414 Phone 612-624-4802 Fax 612-624-1493 www.shadac.org IB-04-0202