REFERENCES
3 HEALTH INSURANCE COVERAGE OF CHILDREN UNDER 19: FY93 AND FY94, SELECTED MONTHS A-18
7 PATTERNS OF HEALTH INSURANCE COVERAGE AMONG CHILDREN UNDER 19: FY93 AND FY94 A-25
8 FREQUENCY OF UNINSURANCE AMONG CHILDREN WITH ANY COVERAGE, BY SOURCE: FY93 AND FY94 A-27
9 CHILDREN UNDER 19 SIMULATED AS ELIGIBLE FOR MEDICAID COVERAGE A-33
13 INSURANCE COVERAGE OF CHILDREN UNDER 19 BY SIMULATED MEDICAID ELIGIBILITY: SEPTEMBER 1994 A-44
14 NUMBER OF UNINSURED CHILDREN BY SIMULATED MEDICAID ELIGIBILITY A-46
Using data from the Survey of Income and Program Participation (SIPP), this report examines issues related to the measurement of health insurance coverage and Medicaid participation and the simulation of Medicaid eligibility and draws some observations about the sources of health insurance coverage among children, the frequency with which children lack health insurance coverage, and the frequency with which these same children are eligible but apparently not enrolled in Medicaid. Subsequent reports will investigate the dynamics of health insurance coverage and Medicaid eligibility and participation among children and the characteristics of children by their health insurance coverage.
This report is organized as follows. Section A provides an overview of the SIPP, including its strengths and limitations, its general representativeness, and its representation of the population of children over time. Section B discusses our use of the SIPP to measure children’s health insurance coverage. Section C presents estimates of health insurance coverage among children. Section D details our methodology for simulating Medicaid eligibility and presents estimates of Medicaid- eligible children by their basis of eligibility. Section E presents estimates of Medicaid participation and other insurance coverage among Medicaid-eligible children, and Section F discusses strategies for evaluating the Medicaid eligibility simulation and presents some comparisons with program administrative statistics as well as estimates of the impact of one feature of our simulation--the definition of the family unit.
The data on which this report is based are from the 1992 panel of the SIPP. The SIPP is a longitudinal survey whose respondents are interviewed every four months about their activity during the preceding four months. The questions include a lengthy series of “core” items, included in every interview, and periodic “topical modules” that collect data more infrequently on specialized areas. One quarter of the sample, constituting a “rotation group,” is interviewed in every month, so that the data for a given calendar month are based on a roughly equal distribution of respondents answering questions about activities one month ago, two months ago, three months ago, and four months ago. The staggered interviewing is intended to ensure that no calendar month of data is affected unduly by recall bias or other error associated with distance from the interview.
The Census Bureau collected nine waves of data--that is, nine interviews--from the entire 1992 SIPP panel sample and a tenth wave of data from three of the four rotation groups (that is, three- quarters of the sample). These data provide a common reference period covering three full calendar years--1992, 1993, and 1994--although, as we will explain, the Census Bureau is not releasing all of the data collected for the final three months of 1994.
Several features of the SIPP make these data especially appealing for the analysis of children’s health insurance coverage. The SIPP provides a detailed measure of health insurance coverage for every month of the two to three year duration of a panel. Because of the SIPP rotation group design, the estimates for a given calendar month are based on a median two-and-a-half-month recall, with one quarter being only one month and one quarter being four months. Measures of the duration of new spells of particular types of coverage or lack of coverage can be constructed by aggregating the reports from successive interview waves, so that no matter how long the measured duration of a particular spell, no part of a reported spell relies on respondent recall beyond four months. In addition to providing measures of health insurance coverage, SIPP also provides very detailed measures of demographic and economic characteristics--again on a monthly basis. This affords us the opportunity to construct contemporaneous measures of circumstances that may affect eligibility for and enrollment in particular types of coverage.
Despite the strengths of the SIPP design, however, there are some notable limitations. Monthly reports of a number of characteristics--including Medicaid participation and uninsurance--show evidence of a pronounced “seam effect.” That is, monthly transitions (for example, in reported Medicaid coverage or health insurance coverage generally) are reported as occurring disproportionately between the four-month reference periods of interview waves rather than within these reference periods. Whereas we would expect only one quarter of such transitions to occur between reference periods, we find evidence that 75 to 90 percent of the transitions in certain statuses occur between interview waves--as if respondents were reporting their coverage (or interviewers recording them) in four-month chunks rather than month-by-month. As we demonstrate in another report [no citation yet], this has a profound impact on the reported distribution of spells of Medicaid coverage and uninsurance, and it may affect the point-in-time estimates as well (unless respondents are equally likely to “round up” as “round down” their reported months of coverage in a reference period). Like other surveys, SIPP shows evidence of underreporting of program participation and many sources of income, although there is evidence to suggest that SIPP does better in this regard than surveys with annual reference periods and less detailed measurement of these characteristics.
It is important to recognize the implications of the SIPP’s longitudinal design for the representativeness of the information that it collects. The SIPP sample is selected from the resident population of the United States, excluding persons living in military barracks or institutions. The sample is designed to be representative of this population at the time that it is drawn, and the initial respondents are weighted to Census Bureau estimates of the size of this population by age, sex, race, and Hispanic origin. The SIPP sample is dynamic, however. Over the life of a SIPP panel some respondents leave the sample and others are added. As a result, the sample size and its representativeness change over time.
Attrition. Respondents who refuse to continue participating in the survey, move to an unknown address, or move more than 100 miles from a SIPP primary sampling unit and cannot be interviewed by telephone are lost from the panel. Because they continue to belong to the population that the SIPP panel was selected to represent, the sample weights of other sample members must be adjusted to compensate for their loss. The 1992 panel had a 9.3 percent nonresponse rate to the initial interview and a cumulative nonresponse rate of 26.2 percent through the ninth interview.
Exits from the Population. Persons who die, move outside the country, enter institutions, or move into military barracks leave the population as well as the sample. Because these losses affect the population as well as the sample, they are not treated as attrition. There is no adjustment to the weights of other panel members to compensate for their loss.
Additions to the Sample. Persons who move into the households of panel members (including those who are born to panel members) become sample members and remain so for as long as they continue to reside with original panel members. Likewise, persons into whose household an adult SIPP panel member moves become sample members as well--again, for as along as the panel member continues to reside with them.
Births clearly represent additions to the population as well as the sample. Other persons added to the sample after the initial interview may or may not represent additions to the population that the SIPP sample represents. Persons who move into or return to the country, leave institutions, or move out of military barracks constitute additions to the population. If they move into SIPP households they become SIPP sample members, and through their addition to the sample the SIPP can be said to represent all persons who joined the population and moved into households that were included in the initial population. Persons who move into the population but form their own households cannot join the SIPP sample. Strictly speaking, then, the SIPP sample does not represent these additions to the population over time. But the SIPP sample weights, as we shall explain, take account of these additions, and so they are represented in number if not actual sample members.
SIPP Weights. To enable inferences from the SIPP sample to the total population the Census Bureau constructs both cross-sectional and longitudinal weights. The cross-sectional weights are created for each calendar month. Weights for a given month are assigned to all persons for whom data were collected in that month, and they are constructed so that they sum to an estimate of the total population by age, sex, race, and Hispanic origin in that month. These weights account for sample attrition (see below) as well as net additions to the population.
The Census Bureau assigns longitudinal weights to all initial sample members who remain through the final interview or leave the survey universe, providing that they miss no more than one consecutive interview.(1) For the 1992 panel, these persons constituted about 74 percent of the initial sample (where the latter includes first wave nonrespondents). These longitudinal weights are adjusted to compensate for persons who attrited through nonresponse (or were never interviewed), and at the outset they sum to the Census Bureau’s estimate of the SIPP population in March 1992. Because of panel members who exit the population as well as the sample, the weighted sample total declines over time. At any point after the first interview, the longitudinally-weighted SIPP panel represents the survivors of the population that the panel represented fully at the start.
The Census Bureau does not assign longitudinal weights to children born after the first interview. These children cannot be weighted with the same scheme that is used for sample members who were actually present for the initial interview, and the Bureau has elected not to apply an alternative weighting scheme.(2) After the first year, then, the weighted longitudinal sample contains no infants. A year later it contains no children under age two, and a year after that it contains no children under age three. For many research purposes--including ours-- this is not acceptable. Therefore, we have followed what has become a commonly used practice of assigning newborns the weights of their mothers.(3)
Adjustments for Nonresponse. Both the cross-sectional calendar month weights and the longitudinal weights take into account characteristics of the panel members who attrited, but the limitations of this adjustment must be recognized. The nonresponse adjustments cannot fully account for the ways in which the attriters may differ from panel members who remain in the sample because some of these differences cannot be known. For example, some of the attrition may be influenced by important changes in circumstances--loss of employment, divorce, birth of a child-- that occurred after the attriter’s last interview. In addition, attriters may simply be different in ways that are not observed but which affect their behavior post-attrition.
Table 1 presents comparative estimates of the population of children represented by the SIPP panel sample and the population of children represented by the individual calendar month samples. The latter estimates were obtained by summing the calendar month weights by single year of age for selected months. These population totals represent, approximately, the Census Bureau’s estimates of the population that would have been eligible for selection into the SIPP sample in each of the individual months. Estimates of the populations represented by the panel and cross-section samples are compared at four points in time: January 1992, October 1992, September 1993, and September 1994. January 1992 is the common reference month for the four rotation groups in the first wave of the 1992 panel. The next three months represent the beginning, middle, and end of the two-year period defined by FY93 and FY94, or the period on which our analysis is focused.
The first thing to note in this table is that the population of children to which the SIPP panel “weights up” in January 1992 actually exceeds the size of the population implied by the calendar month weights--by about 1.1 million children. We have no explanation for a difference of this magnitude. While the SIPP panel is weighted to estimates of the relevant population in March rather than January and, therefore, would not be expected to reproduce the January 1992 population totals, neither would we expect it to exceed the January 1992 population counts, much less by such a large margin. With this discrepancy in 1992, and the opposite trends in the two series, the two estimates of children under 19 cross between January and October, 1992.
From January 1992 through September 1994, the SIPP panel estimate of the population of children declines by about 2.2 million while the population implied by the calendar month weights rises by 2.9 million. The decline in the SIPP estimates can be attributed to the SIPP panel’s underrepresentation of births, which propagates through the younger ages. In January 1992 the SIPP panel sample represents an estimated 4.5 million infants. By October of that year the number of infants has dropped by nearly 1.4 million to 3.1 million. The number rises some by September 1993 but then drops by half a million by September 1994. Because the children born into the panel in 1992 become the panel’s one-year-olds in 1993 and two-year-olds in 1994, the effect of the underrepresentation of births is compounded. By September 1994 the SIPP panel represents 3.5 million fewer children under the age of three than it does in January 1992. The Census Bureau’s population estimates reflected in the calendar month weights indicate that the size of this population did decline over this period, but by only 150,000.
The net difference of 3.35 million between the two samples accounts for most of the 4.0 million children that would be eligible for SIPP in September 1994 but are not represented by the 1992 panel. The remaining .65 million is spread over the ages 3 through 18. The population estimates show this population growing by 3.1 million between January 1992 and September 1994 whereas the SIPP panel shows growth of 1.3 million. The population of children 3 through 18 grows in SIPP because the number of children who move into this age group from younger ages exceeds the number who “age out” at the upper end or leave the population through death, migration, or institutionalization. The SIPP panel estimates of children 3 to 5 grow by more than the Census Bureau’s population estimates because the SIPP panel overrepresents infants in January 1992. In the 6 to 10 age group, the SIPP panel declines slightly over time while the Census Bureau’s population estimates grow by nearly 650,000. In the 11 to 15 age group, the SIPP panel estimate increases by 376,000 while the population estimate rises by 1.1 million. Finally, in the 16 to 18 age group, the population represented by the SIPP panel grows by 156,000 compared to 797,000 for the cross-section sample..
That the differences between the growth trajectories of the SIPP panel and the total population (of SIPP-eligible children) are relatively small over most of the age range suggests that we can generalize from the SIPP panel to the full population fairly readily. It is only at the lower end of the age distribution that we need to be conscious of major differences between the sample and the population. Indeed, with the SIPP panel representing just over three-quarters of the estimated number of children under three in the population, we should be aware of the potential impact on the distribution of characteristics that differ substantially between very young children and older children.
The SIPP ascertains health insurance coverage by means of a series of questions that ask about specific types of coverage or about coverage in general. These questions are placed relatively early in the SIPP interview, when respondents are likely to be more attentive. By contrast, the health insurance questions in the March supplement to the Current Population Survey (CPS) are placed near the end of the interview.
The applicable questions are reproduced below.
20a.(If ... is 65 years of age or older or ... has a work disability) Medicare is a health insurance program for disabled persons and persons 65 or older. People covered by Medicare have a card that looks like this (SHOW FLASHCARD L). Was ... covered by Medicare?
23a.(If ... is 18 or older or the designated parent or guardian of children under 18 years old who live in the household) During the 4-month period, was ... covered by (use local name for Medicaid) or another public assistance program that pays for medical care?
b.May I see ...’s (use local name for Medicaid) card to record the claim number?
23c.(If ... is the designated parent or guardian of children under 18 years old who live in this household) Were any of ...’s children (under 18) covered by (use local name for Medicaid)?
d.(If yes) Which children were covered?
23e.(If 23a or 23c is marked yes) Was (.../(and)...’s children) covered during the entire 4- month period?
f.(If no) In which months was (.../(and)...’s children) covered?
24a.Was ... covered by a health insurance plan at any time during the past 4 months? (Include CHAMPUS, CHAMPVA, and military coverage. Exclude Medicaid, Medicare, and plans paying benefits only for accidents or specific diseases.) (If no, skip to 24k.)
b.Was ... covered by a health insurance plan during the entire 4-month period?
c.(If no) In which months was ... covered?
d.Was ...’s health insurance coverage from a plan in ...’s own name (primary policy holder) or was ... covered as a family member on someone else’s plan? (If own name, skip to 24f.)
24e.Whose plan covered ...? (Skip to 24k.)
f.Was ...’s policy obtained through ...’s current employer or union, through a former employer, through the CHAMPUS or CHAMPVA programs, or in some other way?
g.(If current employer or union or former employer) Did ...’s employer or union (former employer) pay all, part or none of the premium (cost) of this plan?
h.Was ...’s plan an individual plan or a family plan? (If individual, skip to 24k.)
i.Other than ..., which persons in this household were covered by ...’s plan? (Include children as well as adults.)
j.Did ...’s plan cover anyone who did not live in this household during the past 4 months? (If yes) Who did the plan cover?
24k.(If ... is the designated parent or guardian of children under 15 years old who live in the household) Were all of ...’s children under 15 years old covered by a health insurance plan? (Include CHAMPUS, CHAMPVA, and military plans.) (Exclude Medicare, Medicaid, and plans paying benefits only for accidents or specific diseases.)
l.(If no) Which children were covered by a health insurance plan?
24m.(If 24k is yes or one or more children is listed in response to 24l) Were any of these children covered by the plan of someone who did not live in the household during the past 4 months? (If yes) Which children?
These questions were asked of all household members 15 and older, with some additional qualifying restrictions as noted. Coverage of children under 15 was ascertained from those questions that asked explicitly about the coverage of respondents’ children or other household members--that is, question 23d for Medicaid and questions 24i, 24k and 24l for other types of coverage. Note that question 24i could be the source of reported coverage for adults as well as children.
From the fields provided on the SIPP files we assigned each sample person to one of 13 categories of insurance coverage in each calendar month from January 1992 through September 1994. In making assignments to children, and presumably many spouses as well, we often had to refer to the record of a parent or other adult in the household to determine the source of coverage. This was not true for Medicaid or Medicare, but it was true for every other source. The SIPP data file identifies the household member providing the coverage when the plan is in another member’s name, and we used this information to link to that person’s record and access variables describing the source of coverage. When the coverage was provided by someone outside of the household, no source could be identified. We relegated to a separate category those children whose only coverage was provided by someone outside the household. Similarly, if a child’s coverage was reported under question 24k or 24l, then no information on the source was available.(4) We assigned such children to a residual coverage category indicating that we had no information on the type of coverage.(5)
Finally, a person could have been reported as covered by more than one type of plan during a given month. Clearly, Medicaid could be reported in combination with any other type of plan. Married persons could be reported as having their own employer-sponsored or other plan and being covered by their spouse’s plan. Similarly, children could be reported as covered by both parents’ plans, although the Census Bureau’s coding of the responses does not seem to allow for recording coverage under more than one parent’s plan (there is only one variable pointing to another household member as the source of coverage). Where two or more sources of coverage are reported, they may have been overlapping, or one source may have terminated while another began. Rather than trying to capture and display multiple sources of coverage, we elected to assign a single source of coverage to each person, following a priority scheme. In view of the focus of our research, we assigned Medicaid the highest priority. That is, any child who was reported to be covered by Medicaid, regardless of whatever other coverage may have been reported in that month, was coded as a Medicaid enrollee.(6) Priority was accorded to other coverages in the following order:
Beyond Medicaid, then, this scheme gives priority to assigning the type of coverage for which we have the most information.
Table 2 reports estimates of the health insurance coverage of children under 19 in each of the first three waves of the 1992 SIPP panel. These results are based on the full panel sample--that is, sample members who were present for the duration of the panel or until such time as they left the SIPP universe. Results are presented for the one month that was common to the reference periods of all four rotation groups in each interview wave--that is, January 1992 for wave 1, May 1992 for wave 2, and September 1992 for wave 3. Presenting Table 2 allows us to introduce the 13 categories of health insurance coverage that we identified with the SIPP variables and to show what change in the distribution of coverage may have occurred between the beginning of the 1992 and the eve of FY93, which along with FY94 is the focal period for the rest of this report.
The upper panel of Table 2 presents estimates of the number of children in each of the 13 coverage categories while the lower panel gives the percentage distribution for all children in each year. Given that the number of children represented by the panel declines over time while the total population of children rises, the percentage distributions provide greater comparability over time and may apply nearly as well to the full population as to the population that the panel actually represents in each month.
The first two sources of coverage, Medicaid and Medicare, require no explanation. Categories 3 through 6 refer to the current or former employer of a child’s parent or guardian, generally, although an older teen could have reported coverage by his or her own employer. Category 3, which accounts for 55 percent of children in each wave, refers to an employer- or union-sponsored plan with the employer or union paying all or part of the cost of the premiums. Category 4, which accounts for 2 percent or less of children, includes plans that the employer or union offers with no subsidization. Category 5 involves coverage that a former employer continues to subsidize while category 6, representing no more than half a percent of all children, is coverage associated with a former employer but with no (further) employer subsidy. Coverage obtained under COBRA would appear to fall in category 6, although the questions on which this category is based could easily lead COBRA participants to report their coverage elsewhere. Categories 7 through 9 refer to coverage provided by CHAMPUS, CHAMPVA, or the military. Category 10 refers to a source of coverage other than 1 through 9. This may include a state-only plan or coverage purchased in the private insurance market. Category 11 is coverage provided by someone living outside the child’s household--typically a divorced or absent parent. SIPP tells us nothing about the source of the coverage because the person in whose name the coverage is held is not interviewed, but we can infer that this coverage would be assigned to one of categories 3 through 10.
Category 12 consists of coverage that could not be classified under one of the preceding 11 categories. This category accounts for only .3 percent of all children in the first wave but then grows to 1 percent in the second wave and remains close to that level in wave 3. We suspected that most of the children classified under category 12 were placed in that category by their parents’ responses to question 24k. That is the children were reported as covered but were not identified with the plan of any adult in the household. If all of the children assigned to category 12 were allotted to the category for that reason, then we would expect to find that all of them were under 15 years of age, based on the final screen for question 24k. While disproportionate numbers of the children in category 12 were indeed under age 15, there remained enough who were over that limit to persuade us that some other explanation was operative. Because the health insurance variables on the SIPP file are not identified with specific questions, it is difficult to determine how a child could end up in category 12 other than by question 24k. We are left to infer that some of the information on children’s coverage must have been incomplete. For example, the Census Bureau may have coded the child as covered under another household member’s plan but either failed to identify or misidentified the household member responsible for the coverage. With further review of individual records it may be possible to resolve why these code 12 assignments were made, but it may not be possible to determine the correct coverage in each instance. Given the low frequency of this category, further investigation may have little merit.
The final category consists of children with no reported coverage. The relative frequency of this category remained at 13.7 percent between waves 1 and 3. During this same period, the proportion of children reporting Medicaid coverage rose from 16.1 percent to 17.3 percent. This rise was offset by a decline in other types of coverage sufficient to leave unchanged the proportion of children reported as uninsured.
Table 3 presents estimates of children’s health insurance coverage for selected months of FY93 and FY94. The first page provides estimates of numbers of children while the second page gives percentages of all children. Reported Medicaid coverage continues the rise observed in Table 2., growing to 19.0 percent of all children by September 1994. The percentage of children who are reported to be without health insurance declines by a full percentage point from September 1992 (in Table 2). The only other notable changes are a decline in the proportion reporting coverage under an employer plan to which the employer makes no contribution and a rise in the proportion of children who are reported to be insured but with no details provided. The former drops from 1.7 percent of all children to 1.2 percent while the latter increases from .8 percent to 1.1 percent.
TABLE 3
HEALTH INSURANCE COVERAGE OF CHILDREN UNDER 19: FY93 AND FY94, SELECTED MONTHS
Number of Children
Source of Coverage Oct-92 Dec-92 Jun-93 Sep-93 Dec-93 Jun-94 Sep-94
TOTAL 71,549,000 71,323,000 71,111,000 70,868,000 70,795,000 70,197,000 69,935,000
Medicaid 12,614,000 12,829,000 13,328,000 13,369,000 13,233,000 13,264,000 13,253,000
Medicare 0 0 0 0 0 6,000 6,000
Current Employer
Paying All or Part of 39,174,000 39,341,000 38,872,000 38,350,000 38,925,000 37,929,000 38,039,000
Paying None of Cost 1,228,000 1,074,000 1,107,000 968,000 844,000 991,000 837,000
Former Employer
Paying All or Part 781,000 639,000 727,000 685,000 790,000 1,031,000 506,000
Paying None of Cost 276,000 253,000 214,000 246,000 275,000 272,000 309,000
CHAMPUS 539,000 705,000 631,000 567,000 502,000 573,000 465,000
CHAMPVA 89,000 51,000 63,000 79,000 55,000 39,000 88,000
Military 954,000 867,000 864,000 891,000 876,000 829,000 793,000
Other 3,007,000 3,050,000 2,823,000 2,831,000 2,727,000 2,911,000 2,907,000
Provided by Someone
outside the Household 2,841,000 2,852,000 2,948,000 3,072,000 2,845,000 2,962,000 3,035,000
Details Unknown 559,000 522,000 587,000 539,000 702,000 670,000 786,000
Uninsured 9,489,000 9,141,000 8,947,000 9,271,000 9,021,000 8,719,000 8,911,000
Percent of Total Children
Source of Coverage Oct-92 Dec-92 Jun-93 Sep-93 Dec-93 Jun-94 Sep-94
TOTAL 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Medicaid 17.6 18.0 18.7 18.9 18.7 18.9 19.0
Medicare 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Current Employer
Paying All or Part of 54.8 55.2 54.7 54.1 55.0 54.0 54.4
Paying None of Cost 1.7 1.5 1.6 1.4 1.2 1.4 1.2
Former Employer
Paying All or Part 1.1 0.9 1.0 1.0 1.1 1.5 0.7
Paying None of Cost 0.4 0.4 0.3 0.3 0.4 0.4 0.4
CHAMPUS 0.8 1.0 0.9 0.8 0.7 0.8 0.7
CHAMPVA 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Military 1.3 1.2 1.2 1.3 1.2 1.2 1.1
Other 4.2 4.3 4.0 4.0 3.9 4.1 4.2
Provided by Someone
outside the Household 4.0 4.0 4.1 4.3 4.0 4.2 4.3
Details Unknown 0.8 0.7 0.8 0.8 1.0 1.0 1.1
Uninsured 13.3 12.8 12.6 13.1 12.7 12.4 12.7
SOURCE: Survey of Income and Program Participation, 1992 Panel.
TABLE 4
CHILDREN UNDER 19 ENROLLED IN MEDICAID OR WITHOUT
HEALTH INSURANCE AT END OF FISCAL YEAR OR EVER IN YEAR
Number of
Children Total Percent
Under 19 Children of Total
by Health Under 19 Children
Health Insurance by Period Insurance in Period Under 19
Medicaid
Enrolled in September 1993 13,369,000 70,868,000 18.9%
Ever enrolled in FY93 17,800,000 74,691,000 23.8%
Enrolled in September 1994 13,259,000 69,935,000 19.0%
Ever enrolled in FY94 17,795,000 73,619,000 24.2%
Ever enrolled in FY93 or FY94 21,043,000 77,697,000 27.1%
Uninsured
Uninsured in September 1993 9,271,000 70,868,000 13.1%
Ever uninsured in FY93 16,089,000 74,691,000 21.5%
Uninsured in September 1994 8,911,000 69,935,000 12.7%
Ever uninsured in FY94 15,936,000 73,619,000 21.6%
Ever uninsured in FY93 or FY94 21,074,000 77,697,000 27.1%
SOURCE: Survey of Income and Program Participation, 1992 Panel.
The upper panel of Table 4 compares enrollment in Medicaid at the end of each fiscal year to enrollment ever during the year. The lower panel makes the same comparison for uninsurance. For Medicaid the number ever enrolled in each fiscal year is one-third higher than the number enrolled at the end of the year. The number of children ever enrolled in Medicaid during the full two-year period is 58 percent higher than the number enrolled at the end of either year.(7) When compared to the one-year ever-enrollment, the two-year figure indicates very modest turnover from one year to the next, as the number of children ever enrolled over two years is less than one-fifth higher than the number ever enrolled in the first or second year.
Turnover among the uninsured is considerably higher than it is among Medicaid enrollees. The number of children ever uninsured during a fiscal year is about 75 percent higher than the number who are uninsured at the end of the year. The number ever uninsured during a two-year period is about 30 percent higher than the number ever uninsured in either year alone and about 130 percent higher than the number uninsured at the end of either year (or at any one time). For the two-year period, the percentage of children ever uninsured matches the fraction who were ever on Medicaid: 27.1 percent.
Table 5 presents estimates of the number and percentage of children who were without health insurance coverage by month for the calendar years 1992 through 1994. The percentage uninsured declines over the course of the first year but shows no trend after that. There is no evidence of seasonality in these numbers. We wanted to select two or three months to present statistics in this and the other technical appendices, and this table suggests that the choice is not particularly important. We elected to focus on the final two months of FY93 and FY94 in most of our cross- sectional tables, supplementing these with the first month of FY93 when appropriate. September 1993 happens to have a relatively high estimate of the uninsured, at 13.1 percent, compared to other months whereas September 1994, at 12.7 percent, is close to the average. It is important to keep in mind that the month-by-month results suggest that the difference between these two estimates does not reflect any trend.
TABLE 4
CHILDREN UNDER 19 ENROLLED IN MEDICAID OR WITHOUT
HEALTH INSURANCE AT END OF FISCAL YEAR OR EVER IN YEAR
Number of
Children Total Percent
Under 19 Children of Total
by Health Under 19 Children
Health Insurance by Period Insurance in Period Under 19
Medicaid
Enrolled in September 1993 13,369,000 70,868,000 18.9%
Ever enrolled in FY93 17,800,000 74,691,000 23.8%
Enrolled in September 1994 13,259,000 69,935,000 19.0%
Ever enrolled in FY94 17,795,000 73,619,000 24.2%
Ever enrolled in FY93 or FY94 21,043,000 77,697,000 27.1%
Uninsured
Uninsured in September 1993 9,271,000 70,868,000 13.1%
Ever uninsured in FY93 16,089,000 74,691,000 21.5%
Uninsured in September 1994 8,911,000 69,935,000 12.7%
Ever uninsured in FY94 15,936,000 73,619,000 21.6%
Ever uninsured in FY93 or FY94 21,074,000 77,697,000 27.1%
SOURCE: Survey of Income and Program Participation, 1992 Panel.
Table 6 presents similar figures for reported Medicaid coverage. Here it is notable that the reported coverage rises from 16.1 percent in January 1992 to 18.7 percent in June 1993, then levels off (with modest swings). It is notable that the percentage point rise in reported Medicaid enrollment is nearly three times the decline in the percentage of children reported to be without health insurance coverage. This suggests that two-thirds of the enrollment increase is due to net movement from other sources of coverage rather than net movement out of the uninsured. We address this issue more directly in Technical Appendix B.
TABLE 6
NUMBER AND PERCENTAGE OF CHILDREN UNDER 19 WHO WERE REPORTED
TO BE COVERED BY MEDICAID, BY MONTH: 1992 THROUGH 1994
Thousands of Children Percent of All Children
Month 1992 1993 1994 1992 1993 1994
January 11,618 12,842 13,231 16.1 18 18.7
February 11,711 12,901 13,228 16.3 18.1 18.7
March 11,981 13,187 13,079 16.7 18.5 18.5
April 12,319 13,264 13,185 17.2 18.6 18.7
May 12,231 12,983 13,109 17.1 18.2 18.6
June 12,373 13,328 13,270 17.3 18.7 18.9
July 12,593 13,294 13,372 17.6 18.7 19.1
August 12,518 13,233 13,418 17.5 18.6 19.2
September 12,390 13,369 13,259 17.3 18.9 19
October 12,614 13,236 12,946 17.6 18.7 18.5
November 12,800 13,263 12,830 17.9 18.7 18.4
December 12,829 13,233 12,815 18 18.7 18.4
SOURCE: Survey of Income and Program Participation, 1992 Panel.
Table 7 provides information on sources of health insurance coverage in FY93, FY94, and the combined, two-year period. Figures for FY93 refer to coverage at any time during that year, and likewise for the FY94 figures. Figures for “FY93 and FY94" refer to the two year period. These estimates are based on our monthly measure of insurance coverage. While a child may in fact have been covered by more than one source in a month (or even at the same time), and SIPP can tell us about multiple sources of coverage during a month, we measured only one source per month, as explained above. Persons with multiple sources in this table, therefore, were covered by those sources at different times during the year (or two-year period for the last two columns of the table).
PATTERNS OF HEALTH INSURANCE COVERAGE AMONG CHILDREN UNDER 19: FY93 AND FY94
FY93 FY94 FY93 and FY94
Coverage Number Percent Number Percent Number Percent
All Children under 19 74,691,000 100.0 73,619,000 100.0 77,697,000 100.0
Children with No Coverage during the Period 5,001,000 6.7 4,731,000 6.4 3,569,000 4.6
Children with Any Coverage during the Period 69,690,000 93.3 68,888,000 93.6 74,128,000 95.4
Children with only one source of coverage 62,995,000 84.3 62,442,000 84.8 62,554,000 80.5
Employer-sponsored 47,187,000 63.2 46,366,000 63.0 46,921,000 60.4
Medicaid 13,918,000 18.6 14,124,000 19.2 14,075,000 18.1
Other 1,889,000 2.5 1,952,000 2.7 1,558,000 2.0
Children with multiple sources of coverage 6,695,000 9.0 6,446,000 8.8 11,574,000 14.9
Employer-sponsored and Medicaid 3,550,000 4.8 3,373,000 4.6 6,242,000 8.0
Employer-sponsored and other 2,813,000 3.8 2,775,000 3.8 4,606,000 5.9
Medicaid and other 225,000 0.3 171,000 0.2 349,000 0.4
All three sources 107,000 0.1 127,000 0.2 376,000 0.5
SOURCE: Survey of Income and Program Participation, 1992 Panel.
NOTE: All children under 19 includes children who were under 19 at the
start of the indicated period as well as children born during the period.
We note that 6.7 percent of children reported no coverage of any kind in FY93, 6.4 percent reported no coverage in FY94, and 4.6 percent reported no coverage in either year. The remainder of the children in each of these periods reported coverage for at least part of the period. Between 84 and 85 percent reported only one source of coverage in each of the two years while 80.5 percent reported only one source over the two-year period. Within each of the two years, 63 percent reported only employer-provided coverage while about 19 percent reported only Medicaid coverage, with about 2.5 percent reporting coverage from another source. For the two-year period, these figures changed little: 60 percent reported only employer-provided insurance, 18 percent reported only Medicaid, and 2 percent reported only another source of coverage. We know from Table 3 that the proportion of children ever covered by Medicaid in the two-year period was higher than the proportion covered in either year alone. Table 7 shows that some of this increase is due to more children reporting Medicaid along with another source over the two-year period relative to one year. Specifically, 9 percent of children report Medicaid in combination with one or two other sources over the two-year period compared to 5 percent for either year alone. (The rest of the increase in Medicaid enrollment must be among children who were otherwise uninsured.)
In Table 8 we look at the incidence of uninsurance among children who reported any coverage during either or both fiscal years. Among children with any coverage, about 16 percent reported one or months of uninsurance in either fiscal year, and 24 percent reported a spell of uninsurance over the two-year period. The probability that a child with any coverage during a period was ever uninsured during that period varies substantially by type of coverage. Among children who were ever covered by Medicaid in either year, 28 percent were uninsured for part of the year. Among those who were ever covered by Medicaid over the two year period, the percentage who were ever uninsured during that period was 41 percent. For those with employer-provided insurance, the percentages uninsured were less than half these figures. Among those with other coverage, the proportions with any months of uninsurance were about midway between those for children ever covered by Medicaid versus employer-provided insurance.
FREQUENCY OF UNINSURANCE AMONG CHILDREN WITH ANY COVERAGE, BY SOURCE: FY93 AND FY94
FY93
Source of Coverage FY93 FY93 and FY94
Children with any coverage 69,690,000 68,888,000 74,128,000
Number ever uninsured 11,088,000 11,205,000 17,504,000
Percent ever uninsured 15.9% 16.3% 23.6%
Medicaid 17,800,000 17,795,000 21,043,000
Number ever uninsured 5,077,000 5,002,000 8,569,000
Percent ever uninsured 28.5% 28.1% 40.7%
Employer-provided 53,657,000 52,641,000 58,146,000
Number ever uninsured 6,750,000 6,852,000 11,627,000
Percent ever uninsured 12.6% 13.0% 20.0%
Other coverage 5,034,000 5,025,000 6,890,000
Number ever uninsured 975,000 1,032,000 2,031,000
Percent ever uninsured 19.4% 20.5% 29.5%
Children with one source of coverage 62,995,000 62,442,000 62,554,000
Number ever uninsured 9,387,000 9,543,000 12,934,000
Percent ever uninsured 14.9% 15.3% 20.7%
Medicaid 13,918,000 14,124,000 14,075,000
Number ever uninsured 3,808,000 3,825,000 5,163,000
Percent ever uninsured 27.4% 27.1% 36.7%
Employer-provided 47,187,000 46,366,000 46,921,000
Number ever uninsured 5,150,000 5,300,000 7,276,000
Percent ever uninsured 10.9% 11.4% 15.5%
Other coverage 1,889,000 1,952,000 1,558,000
Number ever uninsured 429,000 418,000 495,000
Percent ever uninsured 22.7% 21.4% 31.8%
SOURCE: Survey of Income and Program Participation, 1992 Panel.
If we look at just those children who report one source of coverage, so that we can isolate the “impact” of the source of coverage, we find even more pronounced differences by type of coverage. While the incidence of uninsurance is generally lower among children with only one source versus two or more, children whose only source was Medicaid during a year had a 27 percent incidence of uninsurance while those whose only source was employer-provided insurance had only an 11 percent incidence of uninsurance. Among those with coverage from another source during a year, between 21 and 23 percent were uninsured for at least one month in the year.
These figures are particularly interesting in light of the policy focus on the CPS, which counts as uninsured only those persons who reported having had no coverage during the year and provides no estimate of persons who were uninsured for only part of the year. Estimates of the kind reported in Table 8 cannot be constructed with CPS data
The regulations governing eligibility for the Medicaid program are exceedingly complex. There are numerous routes by which a child may qualify for enrollment, and many of the eligibility provisions and parameters vary by state. Relative to other means-tested programs, Medicaid presents a far greater challenge for simulation of program-eligibility--both in terms of the complexity of the rules and the data requirements that they generate. Even the most sophisticated simulation models of Medicaid eligibility employ many simplifications (see, for example, Giannarelli 1992). More typically, estimates of Medicaid eligibility are highly simplified. For example, the General Accounting Office has reported findings based on simulating only the federally mandated poverty- related expansions (U.S. GAO 1995). While this captures the majority of Medicaid-eligible children born after September 30, 1983 (because eligibility via cash assistance programs has lower income thresholds), it attributes no eligibility at all to older children.
The data requirements for a Medicaid eligibility simulation are substantial. Like most means- tested entitlement programs, eligibility determinations are based on monthly income, and countable income includes a number of potential disregards for which the source of income and various kinds of monthly expenditures may be relevant. Participation in certain programs makes families or children eligible for Medicaid, so data on program participation are needed. Because persons who were eligible for AFDC were often eligible for Medicaid even if they did not participate in AFDC, the Medicaid eligibility determination incorporated AFDC eligibility determination, and this has been extended in some form into the post-welfare reform era. This introduces the need to construct several alternative family income measures, which must be compared to sets of state-specific parameters, which vary by family size. AFDC and some of the other Medicaid provisions are limited to particular types of families, creating a need for family demographic and economic data. In addition, the AFDC unit may be a subset of the entire co-resident family, and other aspects of the Medicaid eligibility determination may exclude some family members as well, so there is a need for additional family demographic data as well as the economic characteristics of family members. Furthermore, AFDC imposed a resources test, and other components of the Medicaid program have resource limits as well, so a simulation must include measures of not only financial resources but vehicles as well. Finally, expenditures on health care are instrumental to the determination of eligibility under the medically needy provisions, so health care expenditure data are needed to fully simulate this component of the Medicaid program.
Clearly the monthly data collected by SIPP make it an exceptional data source for simulating Medicaid eligibility, although, as we shall see, even SIPP does not provide all of the data needed for a full Medicaid eligibility simulation. The more sophisticated simulations using the CPS address the monthly data problem to some degree by simulating monthly income streams, but this is no substitute for what SIPP can provide in this regard. Another area where SIPP surpasses the CPS is in its collection of asset data, including vehicles. While this is done only twice in the life of a panel, and we will show that these data have a number of problems, researchers using the CPS to simulate Medicaid must contend with having no direct measures of financial assets--only income generated from some of such assets--and no measures of vehicle assets whatsoever.
In our own simulation of Medicaid eligibility, we replicated the rules for determining eligibility via nine different sets of provisions, assigning the child to the first eligibility category for which the child was found to qualify. In Section 1 we describe these nine eligibility categories, and in Section 2 we present estimates of children simulated to be eligible for each of these categories. Finally, in Section 3 we discuss findings on the impact of resources in excluding children from eligibility.
The first three categories of the nine that we simulate are based on reported participation in programs that provide more or less automatic Medicaid eligibility: Supplemental Security Income (SSI), the foster child “program,” and Aid to Families with Dependent Children (AFDC). The fourth category consists of children whose families were simulated to be eligible for AFDC but did not report participation, and category five includes children who were not AFDC-eligible but would have qualified for Medicaid under a more generous need standard in the states that offered such eligibility. Category six consists of the poverty-related expansions that extend eligibility at higher income levels to children born after September 30, 1983. Category seven extends eligibility to older children in a few states and, in fewer states, to younger children at higher poverty levels than does category six.
Category eight encompasses the so-called “Ribicoff children.” The Ribicoff provisions extended Medicaid eligibility to children in low income families that could not qualify for AFDC because both parents were present and the principal wage earner was neither disabled nor unemployed. For children born after September 30, 1983, eligibility under the Ribicoff provisions has been superseded by the poverty-related expansions, with their higher income thresholds and broader coverage. (In 1993, 22 states limited their Ribicoff coverage to special groups of children.) Ribicoff children do include one small group who were born after September 30, 1983. In states that offer a medically needy program (see below), the generally higher resource limits specified for that program apply to Ribicoff children as well. Ultimately (by September 30, 2002), the poverty-related provisions will extend coverage to all poor children under 19. The higher resource limits that may apply to Ribicoff children will continue to extend eligibility to children in families with very low income but assets above the AFDC limits, providing that the Ribicoff provisions are not rescinded.
Finally, category nine represents a partial simulation of the medically needy provisions. The medically needy program applies to families that do meet the AFDC dependent child definition, so it would not include any Ribicoff children. For families in states that provide medically needy coverage, this program allows generally higher income and asset levels, to which families can “spend down” their income or resources on medical care over a specified period and thereby qualify to have any additional expenditures during the period covered by Medicaid. Our simulation does not reflect the spenddown feature because SIPP does not provide data on health care expenditures. Instead, the simulation takes account of the higher income and resource limits in the states that offer them.
Our final comment on the Medicaid simulation concerns limitations in the SIPP’s identification of states. Nine states are not identified individually on the SIPP file. These small states are grouped with other small states from the same region and reported as three groups:
To carry out our simulations, which employ a number of state-specific parameters, we had to assign respondents in these state groups to individual member states. We did this randomly, consistent with other MPR simulation models that use SIPP data. Arguably, with Vermont having much higher Medicaid eligibility limits than Maine, and Alaska having higher poverty guidelines than the other states with which it is grouped, the random assignment will result in less accurate eligibility simulations for sample members in the first and third state groups. Theoretically, we could have used the results of the simulations to improve upon the random assignment. For example, true Vermont residents, if incorrectly assigned to Maine, would be more likely to appear ineligible for the Medicaid coverage that they report. Seemingly ineligible participants in Maine who would be eligible under the Vermont income thresholds could be reassigned to Vermont, and Vermont nonparticipants who were eligible based on Vermont criteria but not Maine criteria could be reassigned to Maine. The number of sample cases that are likely to be affected is small, however, and it is doubtful that we could identify more than a handful of cases that clearly merited reassignment. This implies, of course, that the impact of the incorrect assignments is likely to be minimal.
2. Estimates of Children Eligible for Medicaid
Table 9 presents estimates of children simulated to be eligible for Medicaid in October 1992, September 1993, and September 1994. Table 9 shows that the total number of children simulated to be eligible for Medicaid lies very close to 17 million in each of the three months. AFDC participants are the single largest group of eligibles, followed by children made eligible by the poverty-related expansions. Children simulated to be eligible for AFDC but who do not report participation are the next largest group, at less than half the size of the children eligible under the poverty-related expansions. The medically needy are the next largest group and would be significantly larger if we could simulate the spenddown provisions. SSI recipients represent a steady group of 400 thousand eligibles. The qualified child provisions make about 150,000 children eligible in October 1992. This rises to nearly 600,000 in 1994 as a direct result of seven additional states joining the four that offered this form of eligibility in 1993. In reviewing this table, it is helpful to keep in mind that the average SIPP respondent represents about 7,000 persons. Estimates as large as 70,000 may be based on sample sizes of only 10 persons. The small numbers in categories 2 and 5 in particular support our decision not to attempt to simulate additional provisions that would affect similarly few children.
TABLE 9
CHILDREN UNDER 19 SIMULATED AS ELIGIBLE FOR MEDICAID COVERAGE
October September September
Basis of Eligibility 1992 1993 1994
Total children eligible 16,986,000 16,835,000 17,024,00
1. SSI recipient 442,000 406,000 402,000
2. Foster child 86,000 112,000 85,000
3. AFDC participant 7,076,000 7,273,000 6,776,000
4. AFDC-eligible nonparticipant 2,722,000 2,454,000 2,857,000
5. Need standard eligible 27,000 29,000 13,000
6. Poverty-related expansions 5,516,000 5,510,000 5,492,000
7. Qualified child 151,000 128,000 596,000
8. Ribicoff child 222,000 278,000 146,000
9. Medically needy without spenddo 744,000 645,000 657,000
SOURCE: Survey of Income and Program Participation, 1992 panel.
Readers familiar with some of the programs that qualify children for Medicaid will recognize that some of these estimates of eligibles appear quite low. After we examine reported participation, below, we will compare our estimates of participants (and eligibles) to program statistics and discuss possible reasons for the shortfall.
Table 10 presents estimates, by age, of the number of children simulated to be eligible for Medicaid. Here we combine related categories of eligibility to eliminate the smallest categories. Of the estimated 17 million children who were eligible for Medicaid at specific points in FY93 and FY94, between 1.2 and 1.6 million were infants, 6.5 to 7.5 million were age 1 to 5, and 4.1 to 4.9 million were 6 to 10. Children 11 to 15 accounted for 2.5 to 3.0 million while children 16 to 18 represented 1.3 to 1.5 million.
We have divided the 6 to 10 age group into children 6 to 8 and 9 to 10 so that we can see the effect of poverty-related eligibility being extended to children who turned 9 and 10 in FY93 and FY94. These children would have been ineligible prior to FY93 whereas they were fully eligible by the end of FY94. Indeed, we see no eligible children in this age group in October 1992, but 353 thousand in September 1993 and 722 thousand in September 1994.
TABLE 10 CHILDREN UNDER 19 SIMULATED AS ELIGIBLE FOR MEDICAID COVERAGE, BY BASIS OF ELIGIBILITY AND AGE Basis of Eligibility Infant 1 to 5 6 to 8 9 to 10 11 to 15 16 to 18 Total Total children eligible, October 1991,386,000 7,464,000 2,993,000 1,122,000 2,496,000 1,524,00016,985,000 SSI recipient, foster child, or AFDC participant 400,000 2,888,000 1,294,000 710,000 1,566,000 746,000 7,604,000 AFDC-eligible nonparticipant or need standard eligible 215,000 925,000 420,000 227,000 518,000 443,000 2,748,000 Povery-related expansions 748,000 3,576,000 1,191,000 0 0 0 5,515,000 Qualified child, Ribicoff child, or medically needy without spendd 23,000 75,000 88,000 185,000 412,000 335,000 1,118,000 Total children eligible, September 11,603,000 6,891,000 2,805,000 1,486,000 2,736,000 1,315,00016,836,000 SSI recipient, foster child, or AFDC participant 571,000 2,653,000 1,338,000 737,000 1,756,000 736,000 7,791,000 AFDC-eligible nonparticipant or need standard eligible 228,000 835,000 320,000 265,000 513,000 323,000 2,484,000 Povery-related expansions 794,000 3,286,000 1,077,000 353,000 0 0 5,510,000 Qualified child, Ribicoff child, or medically needy without spendd 10,000 117,000 70,000 131,000 467,000 256,000 1,051,000 Total children eligible, September 11,168,000 6,580,000 2,982,000 1,862,000 2,958,000 1,475,00017,025,000 SSI recipient, foster child, or AFDC participant 240,000 2,619,000 1,318,000 760,000 1,582,000 744,000 7,263,000 AFDC-eligible nonparticipant or need standard eligible 260,000 842,000 405,000 307,000 674,000 383,000 2,871,000 Povery-related expansions 661,000 3,001,000 1,108,000 722,000 0 0 5,492,000 Qualified child, Ribicoff child, or medically needy without spendd 7,000 118,000 151,000 73,000 702,000 348,000 1,399,000 SOURCE: Survey of Income and Program Participation, 1992 panel.
Despite this sizable increase in the number of children who were Medicaid-eligible due to the continuing phase-in of the poverty-related criteria over this period, our estimates of total children eligible for Medicaid do not change. There is a decline in Medicaid eligibility among children 1 to 5 that offsets the rise in eligibility not only among children 9 to 10 but also among children 11 to 15. We infer that this decline is due to the SIPP’s underrepresentation of births and its creeping effect on estimates of the number of children up to age 3 by the end of the panel.
3. The Impact of Resources on Medicaid Eligibility
To qualify for Medicaid under many of the provisions requires that a child’s family pass not only an income test but a resources test as well. For example, to be eligibile for AFDC a family’s countable resources could not exceed $1,000. Countable resources included financial assets and non-commercial vehicles, with a family being able to exclude up to $1,500 from the value of one vehicle. The $1,000 limit did not allow families to hold much of a cash reserve against future needs or own very recent automobiles, so it is conceivable that the resource limit may have disqualified non-negligible numbers of families who would otherwise have qualified for AFDC. With Medicaid, however, some of the other eligibility provisions in many states have higher resource limits or none at all. It is plausible, then, that the ultimate net impact of resource limits is to affect how children qualify for Medicaid and not whether they qualify.
The measurement of family resource holdings in surveys is weak at best. The CPS includes no direct questions on resources. Reseearchers have had to rely on reported income from assets to estimate financial assets by dividing the annual income streams by assumed annual rates of return. Most low income families report minimal if any income from financial assets, so the lack of better data was not a serious problem for part of the population of interest. But for families closer to the margin or with fluctuating income flows that may have made their children periodically eligible for Medicaid, the absence of better data on resources--and particularly on vehicles--had a potentially serious impact on the simulation of Medicaid eligibility.
The SIPP 1992 panel provides measures of financial assets and vehicles at two points in time: specifically waves 4 and 7, which are centered around January 1993 and January 1994. We used these data to estimate monthly resources for the prior, intervening, and subsequent months. For financial assets we found so little relationship between the reports of holdings at the two points in time that rather than estimating monthly financial assets by applying a growth model to the two data points we simply averaged the two values and assigned this mean value to every month.(8) Given the low magnitudes of financial assets that were reported for lower income families generally, we felt that this approach was likely to yield lower error on average. For vehicles we also found considerable change between waves, and we could see that this was due to the frequency with which families acquired and disposed of their vehicles rather than how they valued them. In terms of their contribution to total resources, vehicle assets clearly dominated financial assets among low income families, and the quality of the data reported appeared to be much higher. Because of the frequency with which vehicles changed from one wave to the next, and the difficulty of matching vehicles between the two waves, we ruled out a strategy of estimating the monthly equity of each vehicle and opted, instead, to perform a straight line interpolation between countable equity in the two waves to assign vehicle assets to the intervening months and to extend this line at either end to extrapolate pre-wave 4 and post-wave 7 values.
To determine the impact of resources on simulated eligibility, we did the following. At each stage of our simulation we identified children who were ineligible solely because of excess resources. That is, these children were categorically eligible and met the income criteria under a particular basis of eligibility, but their families’ countable resources exceeded the applicable limit. Any of these children could become eligible at a later stage in the simulation, under a different basis of eligibility, where resource limits were higher or nonexistent. Table 11 reports the number of children who were found to be ineligible for Medicaid solely because of resources, by the stage at which this initially occurred, and breaks them down by their final simulated eligibility status. In
October 1992 we found 1.4 million children who were ineligible at any stage of the simulation solely because of their family resources. More than half of these (806 thousand) would have been eligible for AFDC except for excess resources, and most of the remainder (519 thousand) would have been eligible under the poverty-related expansions. Of the total, 58.8 percent were ultimately simulated to be eligible, with 20.4 percent eligible under the poverty-related expansions, 1.2 percent eligible under the qualified child provisions, 6.3 percent eligible as Ribicoff children, and 30.9 percent eligible under the medically needy provisions (without spenddown). Not surprisingly, the distribution by the final eligibility simulation varies by the stage at which the child was first simulated to be Medicaid eligible except for their resources. Of the 806 thousand children who were otherwise eligible for AFDC, nearly 86 percent were ultimately simulated as Medicaid-eligible. But of those who were first simulated to be Medicaid-eligible (except for resources) under the poverty- related criteria, only 26 percent were found to be Medicaid-eligible under less restrictive criteria. Because the Ribicoff child and medically needy provisions represent the final stage of our simulation, none of the 84,000 children who first met the categorical and income criteria at this point but exceeded the resource limits could become eligible at a later stage.(9) We find similar results in September 1993 and 1994. The number of children affected by the resource tests is somewhat smaller in 1993 (less than 1.3 million) and somewhat larger (1.6 million) in 1994, and the percentage of these who are ultimately simulated to be eligible for Medicaid is somewhat higher than in October 1992.
TABLE 11
CHILDREN SIMULATED TO BE INELIGIBLE FOR MEDICAID SOLELY BECAUSE OF EXCESS RESOURCES,
BY INITIAL ELIGIBILITY SIMULATION AND FINAL ELIGIBILITY SIMULATION: SELECTED MONTHS
Final Eligibility Simulation
Initial Eligibility Simulation
Basis of Eligibility
Stage at Which Child Was Ineligible Poverty-
First Simulated to be Eligib Total for Eligible Related Qualified Ribicoff Medically
Except for Excess Resources Number Medicaid Subtotal Expansions Child Child Needy
Percentage of Total
October 1992
Total 1,409,000 41.2 58.8 20.4 1.2 6.3 30.9
AFDC-eligible 806,000 14.3 85.7 35.6 2.1 0 48
Need Standard Eligible 0
Poverty-related Expansions 519,000 73.7 26.3 0 0 17.1 9.2
Ribicoff Child 45,000 100 0 0 0 0 0
Medically Needy w/out Spendo 39,000 100 0 0 0 0 0
September 1993
Total 1,267,000 34.5 65.5 24.1 1.4 6.5 33.5
AFDC-eligible 750,000 10.1 89.9 40.7 2.3 0 47
Need Standard Eligible 0
Poverty-related Expansions 483,000 67.9 32.1 0 0 17.2 15
Ribicoff Child 6,000 100 0 0 0 0 0
Medically Needy w/out Spendo 28,000 100 0 0 0 0 0
September 1994
Total 1,605,000 38.3 61.7 25.9 5 4.2 26.7
AFDC-eligible 953,000 7.6 92.4 43.6 8.4 0 40.4
Need Standard Eligible 6,000 100 0 0 0 0 0
Poverty-related Expansions 602,000 81.8 18.2 0 0 11.1 7.1
Ribicoff Child 13,000 100 0 0 0 0 0
Medically Needy w/out Spendo 31,000 100 0 0 0 0 0
SOURCE: Survey of Income and Program Participation, 1992 Panel.
Compared to the 17 million children who are simulated to be eligible for Medicaid, the additional number who would be eligible except for their resources is small: about 600 thousand (41 percent of 1.4 million in October 1992 and 38 percent of 1.6 million in September 1994). This relatively modest impact of resource limits can be attributed in no small part to the higher resource limits that prevail in many states for the poverty-related expansions and medically needy programs. Over 80 percent of the children excluded from AFDC because of their resources achieve eligibility under these other programs. The number who would be excluded from eligibility altogether because of their resources would be much higher if the stringent AFDC resource test were applied at all stages of the Medicaid eligibility determination. Nevertheless, it is also clear that the resources held by families whose children would qualify for Medicaid on the basis of their family income and demographic characteristics are generally quite low.
Table 12 presents Medicaid participation rates for simulated Medicaid eligibles by basis of eligibility. The table also includes estimates of Medicaid participation rates for children who were simulated to be ineligible due to their resources and for all other children. This latter group will include children who actually were eligible for Medicaid but not identified as eligible in our simulation. This will include children who were eligible for components of Medicaid that we did not simulate as well as children who were erroneously simulated as ineligible. The biggest portion of this group, we suspect, is children who were eligible for the medically needy program after spending down their income or resources. We recognized that this is a large group, but the SIPP does not provide sufficient data on health care expenses to support simulation of this group.
TABLE 12
MEDICAID PARTICIPATION RATES BY SIMULATED MEDICAID ELIGIBILITY:
CHILDREN UNDER 19, SELECTED PERIODS
October September September
Basis of Eligibility 1992 1993 1994
All Eligible Children 62.9 66 64.7
1. SSI recipient 78.6 70.7 74.6
2. Foster child 51.9 72.3 76.7
3. AFDC participant 100 100 100
4. AFDC-eligible nonparticipant 47.3 47.8 53.8
5. Need standard eligible 19.6 25.9 0
6. Poverty-related expansions 32.1 38.3 36.7
7. Qualified child 31.3 40.4 22.7
8. Ribicoff child 6.5 22.7 4.8
9. Medically needy without spenddo 13.8 8.9 26.5
Children ineligible due to resource 19.5 26 17.1
All other ineligible children 3.3 4 4.1
Total children 17.6 18.9 19
SOURCE: Survey of Income and Program Participation, 1992 Panel.
Over all children simulated to be eligible for Medicaid, we estimated participation rates between 63 percent and 66 percent. This reflects a 100 percent participation rate for the single largest eligibility group--AFDC participants. Recall that this participation is not actually an observed outcome but the result of Census Bureau editing of Medicaid coverage. All persons reporting AFDC coverage are edited to have Medicaid coverage as well. SSI recipients have the next highest participation rates at between 71 and 79 percent. That these rates are not much higher is almost certainly due to our inability to reliably identify SSI recipients in the SIPP--a matter that we discuss further below. Foster children have participation rates between 52 percent and 77 percent, but this is a small group among our simulated eligibles, and these rates have very high sampling error. The same can be said about the need standard eligibles, qualified children, and Ribicoff children.
AFDC-eligible nonparticipants have participation rates between 47 and 54 percent. We suspect that many of these children actually were AFDC participants but that this coverage was not reported to the SIPP interviewers. We theorize, in addition, that the Medicaid participation rate among the actual AFDC participants is higher than the participation rate among those who truly did not receive AFDC benefits. Finally, while the medically needy without spenddown are not as small a group as these other components, the estimated participation rates show excessive variability over the three months.
Children who were simulated to be ineligible due solely to their resources show participation rates between 17 and 26 percent, or not unlike some of the groups of simulated eligibles. This is perhaps not too surprising. The simulation of resources is probably the weakest part of the overall eligibility simulation, so this group may include a number of children who were in fact eligible or would have been simulated as eligible in an earlier month.
Table 13 provides some evidence as to why the Medicaid participation rates may not be higher among segments of the population of eligible children. This table reports all sources of insurance coverage as well as the residual uninsured group for September 1994. While the Medicaid participation rate among all children simulated to be eligible is 65 percent, fewer than half of the remainder are uninsured: 17 percent. The remaining 18 percent have either employer-provided insurance (16 percent) or some other form of coverage. Among children eligible under the poverty- related expansions, 33 percent report employer-sponsored coverage, as do those eligible for the qualified child, Ribicoff child, or medically needy programs without spenddown. Among those children who were simulated to be ineligible solely because of their resources, 48 percent report employer-sponsored coverage. Nevertheless, 33 percent of this group and comparable percentages of all of the Medicaid eligibility groups that are not eligible through participation in SSI, foster care, or AFDC remain uninsured. The uninsurance rate among all other Medicaid-ineligible children (which includes some who were in fact eligible but whose eligibility could not be simulated) is 11.1 percent, compared to 12.7 percent for all children.
If significant numbers of the Medicaid-eligible nonparticipants are in fact covered by some other form of insurance, then it may be informative to calculate an alternative Medicaid participation rate that excludes children with other coverage. If we calculated such a rate from the estimates reported in Table 13, by dividing the Medicaid-eligible children who report participation by those who are either participants or uninsured, we would obtain an overall participation rate of 79 percent. Participation among the children eligible as qualified children, Ribicoff children, or medically needy without spenddown would be raised to 37 percent. Participation among children eligible under the poverty-related expansions would be 58 percent, and participation among the AFDC-eligible nonparticipants and need standard eligible would be 63 percent.
TABLE 13
INSURANCE COVERAGE OF CHILDREN UNDER 19
BY SIMULATED MEDICAID ELIGIBILITY: SEPTEMBER 1994
Employer-
Basis of Eligibility Medicaid Provided Other Uninsured Total
Number of Children
Total children eligible 11,012,000 2,748,000 330,000 2,933,00017,024,000
SSI recipient or foster child 365,000 67,000 5,000 50,000 487,000
AFDC participant 6,776,000 0 0 0 6,776,000
AFDC-eligible nonparticipant
or need standard eligible 1,536,000 417,000 12,000 905,000 2,870,000
Poverty-related expansions 2,018,000 1,809,000 227,000 1,438,000 5,492,000
Qualified child, Ribicoff child,
or medically needy without spen 317,000 455,000 86,000 541,000 1,399,000
Children ineligible due to resourc 105,000 294,000 17,000 199,000 615,000
All other ineligible children 2,142,00041,815,000 2,560,000 5,779,00052,296,000
Total children 13,259,00044,858,000 2,907,000 8,911,00069,935,000
Percent of Children by Eligibility
Total children eligible 64.7 16.1 1.9 17.2 100.0
SSI recipient or foster child 74.9 13.8 1.0 10.3 100.0
AFDC participant 100.0 0.0 0.0 0.0 100.0
AFDC-eligible nonparticipant
or need standard eligible 53.5 14.5 0.4 31.5 100.0
Poverty-related expansions 36.7 32.9 4.1 26.2 100.0
Qualified child, Ribicoff child,
or medically needy without spen 22.7 32.5 6.1 38.7 100.0
Children ineligible due to resourc 17.1 47.8 2.8 32.4 100.0
All other ineligible children 4.1 80.0 4.9 11.1 100.0
Total children 19.0 64.1 4.2 12.7 100.0
SOURCE: Survey of Income and Program Participation, 1992 panel.
Table 14 examines the distribution of all uninsured children by simulated Medicaid eligibility status and provides estimates of the contribution of each group to the total uninsured in October 1992, September 1993, and September 1994. Simulated Medicaid-eligibles make up 30 to 33 percent of all the uninsured. We speculate that a more complete simulation would add at least a few percent to these figures. More than half of the Medicaid-eligible uninsured are eligible under the poverty-related provisions while most of the rest are AFDC-eligible nonparticipants. Children simulated to be ineligible for Medicaid solely because of their resources represent a consistent 2 percent of the uninsured. Finally, about two-thirds of all uninsured children are simulated to be ineligible for Medicaid.
In viewing these results, we must be conscious that surveys underreport program participation, and the SIPP is certainly no exception to this rule. For programs such as AFDC and food stamps, researchers have taken this into account most readily by calculating participation rates with administrative counts rather than reported participation in the numerator. With Medicaid we cannot do so as readily. Administrative statistics do not provide counts of Medicaid enrollees for exactly or even approximately the same universe that we are examining here. Furthermore, it is likely that we are missing as many as a few million eligibles from our simulation. We can make a rough calculation, however, with some assumptions about Medicaid underreporting and about the number of eligible children who are missing from our simulation. In Technical Appendix D we estimate the undercount of children ever enrolled in Medicaid to be about 15 percent. Czajka et al. (1998) estimated the underreporting of AFDC and food stamps in the 1992 SIPP panel to be 23 percent, so the Medicaid reporting appears to be considerably better.(10) Inflating the reported number of children enrolled in Medicaid in September 1994 to compensate for an assumed underreporting of 15 percent would raise the estimated number of enrollees to 15.6 million. Adding four million additional eligibles to our simulated number would increase the estimated number of eligible children to 22 million. These two figures would imply an overall participation rate approaching.75 percent.
TABLE 14
NUMBER OF UNINSURED CHILDREN BY SIMULATED MEDICAID ELIGIBILITY
October September September
Basis of Eligibility 1992 1993 1994
Number of Uninsured Children
All eligible children 2,859,000 2,958,000 2,933,000
1. SSI recipient 41,000 64,000 50,000
2. Foster child 23,000 0 0
3. AFDC participant 0 0 0
4. AFDC-eligible nonparticipant 848,000 892,000 905,000
5. Need standard eligible 22,000 8,000 0
6. Poverty-related expansions 1,505,000 1,522,000 1,438,000
7. Qualified child 64,000 37,000 196,000
8. Ribicoff child 97,000 146,000 36,000
9. Medically needy without spenddown 259,000 291,000 309,000
Children ineligible due to resource limit 202,000 146,000 199,000
All other ineligible children 6,427,000 6,166,000 5,779,000
Total uninsured children 9,489,000 9,271,000 8,911,000
Percent of Total Uninsured Children
All eligible children 30.1 31.9 32.9
1. SSI recipient 0.4 0.7 0.6
2. Foster child 0.2 0 0
3. AFDC participant 0 0 0
4. AFDC-eligible nonparticipant 8.9 9.6 10.2
5. Need standard eligible 0.2 0.1 0
6. Poverty-related expansions 15.9 16.4 16.1
7. Qualified child 0.7 0.4 2.2
8. Ribicoff child 1 1.6 0.4
9. Medically needy without spenddown 2.7 3.1 3.5
Children ineligible due to resource limit 2.1 1.6 2.2
All other ineligible children 67.7 66.5 64.9
Total uninsured children 100 100 100
SOURCE: Survey of Income and Program Participation, 1992 Panel.
Whatever the actual participation rate, it is clear that the components of Medicaid that extend eligibility to children who are not eligible for cash assistance programs achieve markedly lower participation rates than those components that do serve the cash assistance population. This has important implications for outreach among not only existing components of the eligible population but particularly among future expansion populations. This would extend, as well, to the Children’s Health Insurance Program (CHIP), which must contend with a more difficult outreach task than making new eligibles familiar with a program with a long history and a record of enrolling more than a fifth of the child population at any point in time and well over a quarter in a given year.
Clearly, it is difficult to infer a great deal about the quality of a Medicaid eligibility simulation by considering the plausibility of implied participation rates. We employed two other strategies. Program administrative statistics generated by the Health Care Financing Administration(now known as Centers for Medicare and Medicaid Services(CMS)) (HCFA(now known as CMS)) give us information not only on the total size of the Medicaid population but aspects of its composition. Unfortunately, it is difficult to match up the published data with our survey-based estimates, but we can still learn something by comparing the two sets of statistics. For subsets of the Medicaid population, data from other programs are relevant as well and may be more amenable to direct comparison. Another approach to evaluating a simulation model is to examine its sensitivity to key assumptions. We took this approach with one aspect of our simulation: the exclusion of selected family members from the program unit when calculating countable income and resources and unit size. Below we present our findings.
1. Comparison with Program Administrative Statistics
Table 15 reports HCFA(now known as CMS) administrative estimates of children ever enrolled in the Medicaid eligibility category “children under 21" by maintenance assistance status in FY 93 and FY94. This group of children is in some respects more inclusive and in other respects less inclusive than the
child population that we have studied with our Medicaid simulation. The HCFA(now known as CMS) administrative category excludes persons under 19 who may be enrolled in Medicaid as adults, and it excludes children eligible because of disability. On the other hand, it includes children living in institutions and children above the age of 19--although neither is a very large group. On balance, it is likely to understate the total number of persons under 19 who are enrolled in Medicaid and would be eligible for the SIPP.
The total figures in Table 15 can be compared to the estimates of ever enrollment reported in Table 4. These comparisons suggest Medicaid coverage rates of 92 and 88 percent respectively, which support our contention that the administrative totals are less inclusive than our sample estimates of Medicaid participants. Perhaps more importantly, however, Table 15 gives us an idea of the relative size of five groups of enrollees--three of which can be compared to our eligibility categories. Keeping in mind that Table 15 includes ever enrollees whereas the eligibles reported in Table 9 represent specific points in time, what we can infer from Table 15 are the relative sizes of the different maintenance assistance categories. We also need to keep in mind that these different segments of the Medicaid population appear to be characterized by different participation rates. We would not expect the simulated eligibles to match their distribution exactly. Nevertheless, comparing Tables 9 and 15 suggests that the medically needy who are not included in our simulation may number about 2.5 million if the medically needy participation rate is 50 percent and about 6 million with a 25 percent participation rate. The size of the population of categorically needy receiving cash is consistent with our simulated AFDC participants once we allow for the impact of AFDC underreporting on the figures in Table 9 and the point-in-time versus ever-in-year reference period. We would require further clarification of the non-cash categorically needy and the pre- and post-1988 legislation categories to draw further inferences. It is likely that these cut across our simulated eligibility categories in ways that prevent comparisons, but there may be additional inferences that we can draw.
TABLE 15
PROGRAM ADMINISTRATIVE ESTIMATES OF CHILDREN UNDER 21
ENROLLED IN MEDICAID, BY MAINTENANCE ASSISTANCE STATUS
Maintenance Assistance Status FY93 FY94
TOTAL 19,427,214 20,326,634
Categorically Needy
Receiving Cash 11,622,810 11,645,544
Not Receiving Cash 2,403,110 2,663,401
Medically Needy 1,698,721 1,664,811
Pre-1988 Legislation 1,568,067 1,755,649
1988 and Later Legislation 2,134,506 2,597,229
NOTE: "Children under 21" is a basis of eligibility. It does not include
all enrollees under 21 years of age.
SOURCE: Health Care Financing Administration(now known as Centers for Medicare and Medicaid Services(CMS)).
Because they are based on reported program participation, our first three categories of simulated Medicaid eligibles can be compared to national estimates of participants to evaluate how fully our simulated eligibles match the actual numbers of children who would have been eligible for Medicaid as SSI recipients, foster children, or AFDC participants.
SSI Recipients. While the number of children covered by federal SSI increased from 624 thousand in December 1992 to 893 thousand in December 1994, the SIPP estimates of SSI children show no growth at all over this period. In October 1992, the SIPP estimate of 442 thousand is 71 percent of the administrative estimate of SSI child recipients three months later, but by September 1994 the SIPP estimate is only 45 percent of the December administrative total. Seeing the lack of growth in the SIPP estimate, we re-examined our coding of SSI participation among children to determine if we were inadvertantly missing new entrants. SIPP panels originating before 1996 did not identify who was actually covered by SSI in a household that reported income from this source. Instead, a single monthly benefit amount was reported on the record of the person who received the check, and an SSI coverage code identified whether the benefit was awarded to (1) adult(s) only, (2) child(ren) only, or (3) both adult(s) and child(ren). Complicating matters further, this code was reported only in the interview wave in which a household first reported receipt of SSI. When codes of 2 or 3 were reported, we imputed SSI coverage to one or more children, based on their disability status, which was measured in wave 6. A careful review of our procedures led to coding changes that actually lowered our estimates from somewhat higher numbers. Clearly, the SSI component in our simulation is problematic. We would hope that the 1996 SIPP panel will provide better identification and coverage of SSI children.
Foster Children. This group is prone to being underreported in the SIPP or any other longitudinal survey because of the changes in household membership that foster children experience. The foster child population is a very dynamic one. Relative to the total children who are identified as foster children at a point in time, the numbers moving into our out of that status over the course of a year is high. The SIPP is likely to miss many if not most of the changes to foster child status because they are accompanied by changes in household membership. In the SIPP, children who become foster children and leave their original sample households are not followed and, therefore, are treated as attriters. They are not assigned longitudinal weights and cannot be counted in population estimates. Children who join new households as foster children are not assigned weights, either. They appear in the sample and, presumably, are identified as foster children, but they are excluded from population estimates. Researchers wanting to focus on foster children can develop alternative weights for such children, but these children cannot be combined with the rest of the SIPP sample of children. Because foster children represent such a small part of the Medicaid population and make little if any contribution to the uninsured, we did not pursue such a strategy.
Our SIPP estimates of 85,000 to 112,000 foster care children compares to administrative estimates of between 224,000 and 244,000 children in federally assisted foster care over this same period and 427,000 to 468,000 children in the entire foster care population. If SIPP is misidentifying one-half to two-thirds of the children in federally assisted foster care, then estimates of the uninsured may be overstated by close to this amount (and Medicaid enrollment by a corresponding amount). If, on the other hand, SIPP is in fact missing rather than misidentifying these children, there is no impact on the uninsured population. The effect on estimated Medicaid enrollment is only modest in either case, but it is likely that those children who are represented in SIPP may not be very typical of the rest of the foster care population.
AFDC Children. Prior to welfare reform, children covered by AFDC were by far the largest component of the Medicaid population. Children receiving AFDC were automatically eligible for Medicaid and, in theory, automatically enrolled. In the SIPP, children reported as covered by AFDC are classified in the Census Bureau’s editing routines as covered by Medicaid, regardless of whether such coverage is reported, so we cannot use these data to estimate how many AFDC children were not reported as covered by Medicaid.(11) Program administrative estimates of the average monthly number of children enrolled in Medicaid grew from 9.1 million in 1992 to 9.5 million in 1994 whereas we estimated the number of child participants reported in the SIPP to be about 7 million over this period. These figures are consistent with the 77 percent coverage that Czajka et al. (1998) estimated for the entire AFDC population in the 1992 SIPP panel. We speculate that many of the unreported AFDC children may appear among the estimated 2.5 to 2.9 million AFDC-eligible nonparticipants that we reported in Table 9 and that this may contribute to an overestimate of the true rate of Medicaid participation among children who actually were nonparticipants in AFDC.
Seeing the percentage of eligible children reported to be covered by employer-provided or other insurance, we were concerned that our simulation model was overly generous in attributing eligibility under the poverty-related expansions. We wondered, in particular, if too many children were simulated to be eligible as subfamily members in larger families that would not otherwise be eligible. A child in a subfamily with sufficiently low income can qualify for Medicaid even if the total family income relative to the poverty line is well above the poverty level that the subfamily must not exceed to qualify. Eligibility determination becomes complex in these cases.
To investigate this aspect of our eligibility simulation, we calculated the percentages of total eligibles that could be attributed to different types of families at different poverty levels. For this purpose, we used the Census Bureau’s identification of subfamilies rather than the sometimes smaller units that may have been the basis for our eligibility simulation.(12) We did this separately for each of three groupings of the eligibility categories. The results are presented in Table 16, which supports our simulation. We combined the SSI, foster child, and AFDC participants into a single category because the eligibility unit in each of these cases was pre-defined. We combined the AFDC or need standard eligible nonparticipants with the medically needy because all three of these categories make use of AFDC-type family units. The final group consists of children eligible under the poverty-related expansions or the qualified child or Ribicoff provisions. It is here that our use of subfamilies--and, specifically, “related” subfamilies--could have had the most impact on simulated eligibility.
TABLE 16
DISTRIBUTION OF MEDICAID-ELIGIBLE CHILDREN BY SUBFAMILY MEMBERSHIP
AND POVERTY LEVEL: MEDICAID ELIGIBILITY GROUPS, MARCH 1993
Federal Poverty Level of Child's Family
Medicaid Eligibility Group and Less than 100% to 200% or
Subfamily Membership 100% < 200% Greater Total
SSI recipient, foster child,
or AFDC participant 78.0 16.5 5.6 100.0
Related subfamily 7.6 5.7 2.0 15.3
Unrelated subfamily 4.1 0.4 0.0 4.4
Not in subfamily 66.4 10.3 3.6 80.3
AFDC or need standard eligible nonparticipant
or medically needy without spenddown 79.8 12.0 8.2 100.0
Related subfamily 8.9 5.2 5.7 19.8
Unrelated subfamily 7.0 0.0 0.0 7.0
Not in subfamily 63.9 6.7 2.6 73.2
Poverty-related, qualified child,
or Ribicoff child 55.4 38.8 5.8 100.0
Related subfamily 1.3 3.8 2.5 7.6
Unrelated subfamily 2.3 1.4 0.0 3.7
Not in subfamily 51.7 33.6 3.3 88.7
SOURCE: Survey of Income and Program Participation, 1992 panel.
As it turns out, children in related subfamilies in families with incomes above the poverty line account for only 6.3 percent of the simulated eligibles in this group. This fraction is actually no bigger than the proportion of eligibles in each of the other two groupings that are attributed to related subfamilies in families above the poverty line. While a larger fraction of children simulated to be eligible under the poverty-related provisions are in families above the poverty line, this is to be expected. The poverty-related criteria provide eligibility at higher income levels across the board than do the other provisions.
Czajka, John L., Scott Cody, and Larry Radbill. “Analysis of Whether Poverty Estimates Vary by the Month of Measurement.” Draft Report. Washington, DC: Mathematica Policy Research, Inc., 1998.
General Accounting Office. “Health Insurance for Children: Many Remain Uninsured Despite Medicaid Expansion.” (GAO/HEHS-95-175) Washington, DC: 1995.
Giannarelli, Linda. An Analyst’s Guide to TRIM2--The Transfer Income Model, Version 2. Washington, DC: The Urban Institute Press, 1992.
(1) The Census Bureau also constructs calendar year longitudinal weights, using the sample of persons with completed interviews covering a given calendar year.
(2) Because they were born after the first interview, these children do not represent the entire universe of newborns who would be eligible for SIPP sample selection, so the population controls that are used to adjust the SIPP sample weights are not exactly appropriate for weighting these children. It would seem, however, that adequate alternative controls could be developed.
(3) We have also backfilled data for infants whose first appearance with sample data occurs one or more months after their births. In doing so, we have assumed that in most cases the lag between birth and the first sample appearance is nothing more than reporting error rather than an accurate statement of when the child arrived in the household. Adopted infants, for example, might not appear for some months after their births. Advancing the month of appearance for some infants has the effect of raising the total infant population in those months.
(4) It appears from the questionnaire that specific months of coverage were not identified for children whose coverage was reported solely in questions 24k or 24l. If so, our supposition is that these children were assigned health insurance coverage in all four months of the reference period.
(5) We can be virtually certain that such coverage did not include Medicaid or Medicare, however, as these were captured separately from the question 24 sequence. But in the absence of revisions to the questionnaire, these generic children’s coverage questions would be the most likely place for respondents to report children covered by state Children’s Health Insurance Plans.
(6) An alternative strategy that should be considered in future research is to take account of simulated Medicaid eligibility in making this determination. A child simulated to be eligible for Medicaid would be assigned to Medicaid over other reported sources of coverage while a child simulated to be ineligible for Medicaid would be assigned to another reported source of coverage. This approach would have the advantage of increasing the consistency between estimated Medicaid coverage and simulated Medicaid eligibility, and it might also reduce the number of erroneous reports of Medicaid coverage--particularly those associated with the seam effect. For the present research, however, we thought it important to base our analysis on Medicaid coverage as reported.
(7) The percentages of children ever enrolled in Medicaid show smaller increments than the absolute numbers because they reflect the increased number of children who were at risk of Medicaid enrollment over a longer period, as shown in the second column of Table 4. That is, the number of persons who were ever under 19 during a year is greater than the number who are under 19 at any one time during the year, and the number who were ever under 19 over a two-year period is greater than the number who were ever under 19 during the first or second year alone. Compared to the first year population, the two-year population includes children born in the second year-- between 3 and 4 million. Compared to the second-year population, the two-year population includes children who turned 19 during the first year--again, between 3 and 4 million.
(8) We inferred that the reporting error in the wave 4 and wave 7 values was simply too high to allow us to estimate reasonably accurate growth rates.
(9) Children who would qualify under the Ribicoff provisions except for their resources cannot qualify as medically needy because the categorical eligibility criteria for these programs are mutually exclusive.
(10) This may reflect the fact that the Census Bureau uses reports of AFDC participation to edit reported Medicaid participation (if anyone reports AFDC participation without Medicaid enrollment, the data are edited to indicate Medicaid enrollment) but cannot and does not do the reverse, as Medicaid enrollment does not necessarily imply AFDC participation, even when a family is eligible for AFDC.
(11) Prior to the 1996 panel, these so called “logical imputations” were not identified in the imputation flags. With the 1992 panel, therefore, we cannot estimate what proportion of the children reported as covered by AFDC and Medicaid had their Medicaid status imputed in this way.
(12) We might have excluded from the unit a child over 18, for example, or a child receiving SSI or foster child payments, whereas these children might be included in the census subfamily or family. Our simplification reflects what data on family unit were readily available for tabulation.
Home Pages:
Assistant Secretary for Planning and Evaluation
ASPE)
U.S. Department of Health and Human Services
(HHS)