Estimating Federal Payments and Eligibility for Basic Health Programs: An Illustrative Example

The Characteristics of BHP Eligibles by State

The federal BHP payment formula depends on applicable benchmark premiums and on four characteristics of BHP enrollees: age (within ranges specified by the BHP federal payment methodology), income (within FPL ranges specified by the BHP federal payment methodology), number of persons in the tax unit (the household unit, as defined for purposes of determining eligibility both for BHP and QHP subsidies), and number of BHP-eligible persons in the tax unit who receive coverage through BHP. In order to compute payments, the joint distribution of these four characteristics—in other words, the number of enrollees at each benchmark premium level who possess every possible combination of the above four characteristics—must be known. For each state, we estimated the number of the joint distribution of these characteristics among people who would be eligible for BHP in 2016.¹

We did not model how many of those eligible for BHP would actually enroll in the program. This depends to a large extent on the BHP premiums and beneficiary cost sharing, and states have a lot of flexibility in setting these elements of BHP policy.

Methods

To produce these estimates, we began with the Urban Institute’s Health Insurance Policy Simulation Model-American Community Survey (HIPSM-ACS). To obtain a large, representative sample population for each state, we pooled together the observations on the 2009, 2010, and 2011 American Community Surveys (ACS). Among national surveys conducted by the U.S. Census Bureau, the American Community Survey (ACS) has the largest state-specific samples and so is likely to provide the most reliable estimates. However, a limitation of both this data set and the other data set frequently used (the Current Population Survey-Annual Social and Economic Supplement) is that they do not include information about offers of employer-sponsored insurance (ESI), which almost always preclude subsidy eligibility.² States that fail to take such offers into account will overestimate the prevalence of relatively high-income BHP-eligible consumers, since ESI offers grow increasingly common as income rises.³ As a result, such states will underestimate federal BHP funding per BHP enrollee, since QHP subsidies, hence BHP funding levels, decline as income rises. The estimates presented here do not share this problem, since HIPSM incorporates, via statistical matches with other data sources, information about unaccepted ESI offers.

Immigration Status. We impute documentation status for non-citizens in each year of survey data separately based on a year-specific model used in the CPS. Documentation status is imputed to immigrants in two stages, using individual and family characteristics, based on an imputation methodology that was originally developed by Passel, the most-used source of estimates of immigrants not lawfully present.⁴ Undocumented immigrants and lawfully present non-citizens, including immigrant adults who have been U.S. residents for less than five years, are generally ineligible for Medicaid.

Tax units and filing. To model tax units and filing behavior, we use 2011 tax rules (including thresholds for tax filing requirements), Earned Income Tax Credit (EITC) eligibility guidelines, and poverty guidelines as defined by the U.S. Department of Health and Human Services. Baseline coverage and post-ACA eligibility are based on estimates from HIPSM-ACS.

Tax units and filing status are determined based on the IRS guidelines set forth by the 2011 1040 Instructions and the 2011 EITC eligibility guidelines. The primary tax filing unit for each family is defined as the head of the family, the spouse, and any qualifying children or qualifying relatives (as defined by the IRS). In multi-generational households, nuclear subfamilies are tested for their filing status. If they are not found to file as a unit themselves, they are tested to qualify as dependents of the head of the household.

Tax filing status is determined based on characteristics of the head of the tax unit and pooled income within the tax unit. Married couples are assumed to be filing jointly to qualify for tax credits. As support within the household is not captured by the ACS, any unmarried tax unit head with dependents is considered filing as a head of household. Any other unmarried person without dependents is tested as single. To determine requirement to file, individual Adjusted Gross Income (AGI) is pooled for each person within the tax unit and compared to the 2011 minimum mandatory filing threshold.

Due to limitations of the income that is captured by the ACS, some taxable income categories could not be included in total income. Capital gains are not reported as investment income in the ACS, so it was not counted. Paid alimony was also excluded; however, internal analysis based on CPS alimony data suggests this exclusion would not affect our results. The ACS does not collect data on unemployment compensation, but because this was likely an important form of income for people at the margin of the Medicaid and subsidy eligibility thresholds, it was imputed based on reported unemployment compensation from the 2008 CPS.

None of the adjustments needed to calculate AGI are reported by the ACS, so we therefore take total income as a proxy for AGI. Total income is calculated as the sum of wages, business income, farm income, rents, most forms of positive investment income, retirement income, unemployment compensation, and the taxable portion of social security income.

EITC eligibility is calculated in a slightly different way. AGI is pooled only among the head of the tax unit, the spouse (if filing as a married couple), and qualifying children. Qualifying dependents are not tested to file for EITC individually because they are either childless dependents (ineligible for EITC) or are found not to file in subfamily analysis. However, because they are claimed on the tax unit head’s return, they take on the EITC eligibility status of their tax unit.

Once it was determined which tax units were required to file and which were eligible for EITC, units were assigned filing decisions. A 2005 Treasury Report estimated that about 7.4 million taxpayers who were required to file did not in Tax Year 2003.⁵ That year, approximately 131 million individual tax returns were filed,⁶ meaning the filing rate among those required to file was about 95%. A study by the IRS of Tax Year 2005 filings estimated the following EITC participation rates, by number of qualifying children: 55.6% among those without qualifying children, 73.6% among those with one qualifying child, and 85.9% among those with two or more qualifying children.⁷ Based on these rates, tax units were randomly assigned their decision to file or not file.

Eligibility for Medicaid/CHIP, QHP subsidies, and BHP. Medicaid and subsidy eligibility are determined using MAGI, which adds nontaxable social security income to AGI. Unit-level MAGI is pooled among the unit head, the spouse (if married), and any qualifying children with an individual AGI above the single tax filing threshold. The income of other qualifying children and qualifying relatives is not included. This is then used to calculate a ratio of MAGI to the applicable federal poverty level (FPL) of the unit. Special prorating of units that include undocumented parent(s) or childless spouses is used to scale the total AGI (including that of the undocumented family members) by a ratio of the FPLs including and excluding the undocumented family members.

Medicaid eligibility for some groups, particularly the blind and disabled, does not change under the ACA. We model their eligibility using pre-ACA rules. To determine Medicaid and CHIP eligibility for other groups, tax unit-level MAGI-as-a-percentage-of-FPL is assigned to the tax unit head, the spouse (if married), and qualifying children with individual AGI above the single tax filing threshold. Excluded qualifying children and qualifying relatives are automatically eligible for Medicaid under CMS regulations. Under the ACA, the children of non-filing qualifying dependents also automatically qualify for Medicaid. The remaining parents, childless adults, and children are then tested for Medicaid eligibility based on the corresponding eligibility threshold in their state of residence. Children who are found ineligible for Medicaid are tested for CHIP eligibility.

QHP subsidy eligibility is determined slightly differently. To be eligible for subsidies, one must have a MAGI-as-a-percentage-of-FPL between 100 and 400%. Eligibility for any public coverage precludes eligibility for subsidies, so subsidy-eligible consumers cannot be eligible for Medicaid or CHIP under the ACA, as determined above, nor can they currently be eligible for Medicare. Finally, no unit member can have an offer of single coverage that costs less than 9.5% of family MAGI. For this determination, we use the HIPSM-ACS imputation of employer offers and the affordability of those offers.

Those eligible for BHP are by definition those eligible for QHP subsidies who have incomes below 200% FPL.

Single Distributions of Each Characteristic. The resulting data allowed us to produce reliable estimates of the single distributions of BHP eligibles by state of age group, FPL income group, number of people in the tax unit, and number of BHP eligibles within the tax unit. These are Tables A1, A2, and A3.

Joint Distributions for Each State. As noted earlier, estimating federal BHP payments requires the joint distribution of all four characteristics by state. That is, one must know how many BHP-eligible residents of a state share a particular combination of age, FPL level, household size, and number of BHP-eligible household members. This would mean separating the BHP-eligible population for each state into 240 different groups.⁸ To get reliable estimates for so many small groups of people would require a sample size for each state far larger than what our data provide. We overcame this difficulty using a standard small area estimation technique that relies on our data having a large enough sample size to estimate this four-trait joint distribution among BHP-eligibles nationally. For each state, we reweighted the national joint distribution to match the individual state’s single distribution of age group, FPL income group, household size, and the number of BHP-eligible individuals per household.⁹ Thus, we used estimates in which we had confidence—state-level single distributions of characteristics and the national joint distribution—to estimate the state-level joint distribution, which could not itself be tabulated directly from the data. The single distributions for each state are shown in tables A1-A3 and the final joint distribution estimates are shown in Table A4. One additional single distribution, involving household size, is not included here, but is available upon request from the authors.

Results

The following tables present the data on the characteristics of the BHP-eligible population by state. Tables A1-A3 provide summary-level statistics on age, income range, and the number of BHP-eligible people in the household unit for all 50 states and the District of Columbia. Table A4 provides detailed estimates of the joint distribution of BHP-eligible consumers by the four characteristics listed above. These detailed estimated are not provided for several states (Alaska, Delaware, the District of Columbia, North Dakota, South Dakota, and Wyoming) due to small sample sizes in those states. Detailed estimates are also not provided for New York because more comprehensive Urban Institute estimates have already been incorporated into state budget projections. Because of sample size considerations, we did not distinguish between FPL income ranges below 138% FPL. The number of BHP-eligible persons in the household unit represents the maximum number of people in the household who can enroll in BHP. Because very few BHP-eligible people are in households with more than five members or in households with more than three BHP-eligible members, our largest listed categories included households with five or more members and with three or more BHP-eligible members. In Table A4, we present data for households with one to four members. You can access the complete data in a downloadable Excel file of Appendix Table A4 (.xls).

Table A1: BHP Eligibles by Age
State	19-20		21-34		35-44		45-54		55-64		Total
	N	%	N	%	N	%	N	%	N	%	N
Alabama	4,042	5%	30,794	35%	16,405	19%	13,343	15%	22,587	26%	87,172
Alaska	730	4%	8,080	47%	2,040	12%	2,765	16%	3,744	22%	17,358
Arizona	4,614	4%	41,738	36%	20,834	18%	19,598	17%	29,125	25%	115,909
Arkansas	2,606	5%	19,441	35%	10,394	19%	9,470	17%	13,810	25%	55,720
California	46,615	6%	335,180	40%	154,246	19%	149,334	18%	147,330	18%	832,704
Colorado	4,900	5%	37,949	39%	16,602	17%	17,882	18%	20,136	21%	97,469
Connecticut	3,444	8%	17,814	41%	5,359	12%	6,128	14%	10,775	25%	43,520
Delaware	736	6%	4,909	39%	2,178	17%	1,800	14%	2,901	23%	12,523
DC	1,253	15%	3,065	38%	727	9%	843	10%	2,216	27%	8,103
Florida	23,137	5%	176,938	35%	98,005	20%	93,656	19%	107,119	21%	498,855
Georgia	10,465	5%	80,941	38%	41,128	20%	36,648	17%	41,607	20%	210,789
Hawaii	891	3%	8,720	34%	4,539	18%	5,365	21%	6,085	24%	25,600
Idaho	1,593	4%	15,628	41%	6,612	18%	5,537	15%	8,331	22%	37,701
Illinois	11,913	6%	81,309	38%	36,543	17%	38,332	18%	44,418	21%	212,515
Indiana	7,554	6%	50,822	38%	22,726	17%	21,858	16%	29,945	23%	132,905
Iowa	2,875	6%	18,301	41%	7,201	16%	7,370	17%	8,516	19%	44,263
Kansas	3,100	6%	19,360	39%	8,417	17%	9,056	18%	10,271	20%	50,203
Kentucky	2,982	4%	29,472	36%	13,878	17%	13,433	16%	22,069	27%	81,834
Louisiana	4,522	5%	36,219	39%	16,402	18%	14,606	16%	20,969	23%	92,717
Maine	945	4%	7,718	30%	3,491	14%	5,078	20%	8,189	32%	25,421
Maryland	4,455	5%	32,278	37%	16,674	19%	16,270	19%	17,541	20%	87,218
Massachusetts	5,941	8%	32,600	43%	11,939	16%	11,577	15%	13,413	18%	75,470
Michigan	8,396	4%	62,469	33%	29,357	16%	34,450	18%	52,527	28%	187,199
Minnesota	3,984	6%	25,776	37%	6,623	10%	10,723	15%	22,360	32%	69,466
Mississippi	2,189	4%	18,976	35%	10,368	19%	9,038	17%	13,971	26%	54,541
Missouri	5,343	4%	45,599	38%	22,000	18%	19,555	16%	26,792	22%	119,289
Montana	1,248	4%	11,455	39%	4,924	17%	5,102	18%	6,347	22%	29,075
Nebraska	1,232	4%	12,311	40%	5,552	18%	6,157	20%	5,243	17%	30,495
Nevada	2,224	4%	23,549	38%	11,811	19%	11,254	18%	13,012	21%	61,850
New Hampshire	1,193	5%	8,822	37%	3,779	16%	5,237	22%	4,715	20%	23,747
New Jersey	7,215	4%	61,796	38%	33,973	21%	28,459	18%	30,972	19%	162,416
New Mexico	2,239	5%	17,579	37%	8,649	18%	7,955	17%	10,740	23%	47,161
New York	23,288	6%	148,887	41%	67,099	18%	58,707	16%	66,749	18%	364,729
North Carolina	8,706	5%	65,002	35%	36,562	19%	32,422	17%	44,836	24%	187,528
North Dakota	575	4%	6,090	45%	1,910	14%	1,858	14%	2,967	22%	13,400
Ohio	8,202	4%	70,131	35%	35,944	18%	34,827	17%	51,463	26%	200,567
Oklahoma	3,498	5%	29,213	38%	14,672	19%	14,111	18%	16,101	21%	77,596
Oregon	3,959	5%	34,061	39%	15,765	18%	14,239	16%	19,600	22%	87,625
Pennsylvania	11,531	5%	77,880	34%	40,083	17%	42,014	18%	57,625	25%	229,132
Rhode Island	1,460	7%	8,172	40%	3,298	16%	3,407	17%	3,842	19%	20,179
South Carolina	5,488	6%	34,154	35%	16,509	17%	18,123	18%	23,826	24%	98,101
South Dakota	1,142	8%	5,731	39%	2,655	18%	1,980	14%	3,081	21%	14,588
Tennessee	5,369	4%	42,740	35%	21,458	18%	22,255	18%	29,572	24%	121,394
Texas	31,271	5%	231,706	41%	112,162	20%	94,753	17%	100,362	18%	570,254
Utah	3,547	6%	26,562	47%	9,865	18%	8,000	14%	8,142	15%	56,116
Vermont	788	6%	4,149	33%	2,245	18%	2,025	16%	3,402	27%	12,608
Virginia	7,742	6%	48,259	37%	24,876	19%	21,629	16%	28,898	22%	131,403
Washington	6,677	5%	53,526	41%	22,020	17%	23,129	18%	26,174	20%	131,526
West Virginia	899	3%	11,874	34%	5,037	14%	6,873	20%	10,174	29%	34,855
Wisconsin	5,119	6%	31,933	36%	15,401	17%	14,814	17%	22,402	25%	89,667
Wyoming	564	5%	3,593	35%	1,390	13%	1,672	16%	3,098	30%	10,318
* Data suppressed due to low sample size ** See the detailed estimates of BHP costs and savings in state budget projections, based on Urban Institute modeling Source: Health Insurance Policy Simulation Model-American Community Survey, 2014