COVERAGE BEFORE AND AFTER SPELLS OF UNINSURANCE
TRANSITIONS B-24
E CHURNING VERSUS TURNOVER B-43
REFERENCE B-55
NOTE: All of the following tables are in the Adobe Acrobat Portable Document Format (pdf). You need to have Acrobat Reader (a free plug-in) installed on your system in order to view these tables.
This report uses data from the Survey of Income Program Participation (SIPP) to investigate dynamic aspects of the lack of health insurance, eligibility for Medicaid, and Medicaid participation among children in 1993 and 1994. After describing our data source we present findings on the duration of spells of uninsurance based on an analysis of new spells as well as spells active at a point in time. We explain why these alternative approaches to measuring the duration of uninsurance yield such different distributions of spell length. We also present estimates of the amount of time that uninsured children are eligible for Medicaid. From there we turn to an examination of the patterns of health insurance coverage preceding and following completed spells of uninsurance, and we compare spells of different lengths. We do this for spells of Medicaid-eligible uninsurance as well. Next we examine the frequency of transitions in coverage, beginning with changes in coverage between one year and the next and then turning our attention to monthly transitions in uninsurance and Medicaid enrollment, culminating in the presentation of estimates of cumulative monthly exit rates from uninsurance, Medicaid-eligible uninsurance, non-Medicaid-eligible uninsurance, and Medicaid enrollment. In the closing sections we discuss churning versus turnover as phenomena in the dynamics of uninsurance and Medicaid enrollment and then summarize our principal findings.
The findings presented herein are based on an analysis of longitudinal data from the 1992 panel of the SIPP. Starting with an initial sample of about 20,000 households representing the resident, noninstitutionalized population of the United States, except for those living in military barracks, the Census Bureau reinterviewed respondents every four months over a period of three calendar years. Respondents were asked a lengthy set of questions about their labor force participation, sources and amounts of income, family and household composition, and participation in government programs in each of the preceding four months. The interviews were staggered, with one quarter of the sample, constituting a “rotation group,” being interviewed each month. The purpose of this design was, in part, to minimize the impact of recall error and other reporting problems on the quality of the data compiled for calendar months. The estimates for a given calendar month are based on roughly equal proportions of the respondents describing circumstances one month ago, two months ago, three months ago, and four months ago.
Several features of the SIPP make these data especially appealing for the analysis of dynamic behavior in 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. 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 interviews, 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. For example, we can construct a reasonably good simulation of Medicaid eligibility on a monthly basis and compare these monthly estimates to reports of Medicaid participation and insurance coverage--or the lack thereof--on a monthly basis as well. Arguably, there are no better data for investigating the dynamics of children’s health insurance coverage in the context of changing circumstances, including Medicaid eligibility.
Despite these strengths, however, the SIPP does possess some notable limitations that will become evident as we proceed through our findings. These include some well-documented but not well understood problems with the quality of retrospective monthly reports of program participation and health insurance coverage.
The length of time that children spend without health insurance coverage is exceedingly important to policymakers. Brief spells of uninsurance carry very different policy implications than long spells. While transitional spells of even one month in duration carry risks to the temporarily uninsured, the policy responses that they invite are clearly different than those demanded to address chronic lack of insurance. The advent of the Children’s Health Insurance Program (CHIP) gives heightened importance to the duration of uninsurance as a policy variable. Federal law allows states considerable freedom in determining how much of the uninsured population to cover and how to define eligibility. A number of states are specifying minimum periods of uninsurance before a child can qualify for the new coverage.
How long do spells of uninsurance last? Here we present findings with respect to the duration of spells without insurance. We contrast two different ways of identifying spells: new spells that began within a particular 12-month period and active spells that were in progress at the end of that period. In addition to looking at spells of uninsurance, we examine spells in which the uninsured were eligible for Medicaid and, theoretically, could have availed themselves of such coverage.
Table 1 describes the duration of spells of uninsurance that began during FY93. For all of these spells, the first set of columns in Table 1 gives the distribution of completed duration in single months, up to a maximum of 13 or more. For most of this last category we did not observe the spell completion; we know only that these spells lasted at least 13 months. We combine all spells of 13
TABLE 1 (PDF document.)
months or greater because 13 is the longest duration that we can observe with the SIPP panel for a spell starting in the final month of FY93. To produce representative estimates of longer durations, we must limit our sample to spells starting earlier in the year. For spells starting in the first six months of FY93, therefore, the second set of columns in Table 1 reports completed durations up to 19 months or more.(1)
Perhaps the most striking feature of these distributions is the heaping of reported durations at four months. Of the 9.2 million spells that began in FY93, more than one-third were completed in exactly four months. This is not a true reflection of spell durations but a phenomenon of the measurement process in the SIPP. This heaping of responses at four months, and to a lesser degree at multiples of four months (there are local peaks at 8, 12, and 16 months), is the result of a “seam effect.” As we mentioned above, SIPP respondents are interviewed every four months and asked extensive questions about their circumstances in each of the four preceding months. For reasons that are not fully understood, transitions in many characteristics are reported as occurring dispro- portionately between the four-month reference periods of interviews--that is, at the seams--rather than within these reference periods. It happens that this seam effect is particularly strong for transitions into and out of insurance coverage. For example, while we would expect only one quarter of new spells of uninsurance to begin in the first month of a reference period or end in the last month, 75 to 90 percent of new spells are reported as beginning in the first month or ending in the fourth month. The transitions, in other words, are reported to have occurred between the reference periods of the interviews. Explanations for this phenomenon include genuine recall error as well as various kinds of behaviors that respondents or interviewers may engage in to speed up or simplify the interview. Clearly, treating the previous four months as a single reporting period--for at least some kinds of responses--simplifies the interview, but if that is what is happening in the reporting or recording of health insurance coverage, then we have to interpret the data accordingly.
Because of the staggered interviewing, described earlier, the seam effect does not show up in the estimates reported for calendar months. But this is not to say that the seam effect does not influence the overall magnitude of these calendar month or point-in-time estimates. Unless respondents (or interviewers) are as likely to “round up” or “round down” their reported months of participation in a reference period, the monthly estimates of aggregate participation will be affected. In a later report, we consider how the seam effect might affect estimates of the number of children who are uninsured.
The seam effect is not unique to SIPP, by any means, and the research that preceded the initiation of the first SIPP panel in 1984 included an examination of this phenomenon. Panel surveys with annual interviews find the same problem when they attempt to measure monthly behavior. To the extent that recall error contributes to the seam effect in SIPP, the impact of such error will be magnified in surveys that attempt to measure behavior over longer periods of time--such as an entire year, which is the reference period for much of the data collected in the March supplement to the Current Population Survey (CPS). Indeed, the fact that CPS estimates of the number of people who had no health insurance coverage during the previous calendar year resemble point-in-time estimates more closely than they resemble the calendar year estimates produced by panel surveys is indicative of a serious response problem that may share some underlying causes with the SIPP seam effect.
Nevertheless, it is clear that we have to take account of the seam effect when we interpret data on spell durations, such as those reported in Table 1. Estimates of the duration of completed spells reflect the maximum impact of the seam effect because each reported spell is subject to this influence at both ends. Of the nearly 3.2 million spells of uninsurance reported as being completed in exactly four months, how many were really completed in four months, and how many were completed in three or five months, two or six months, or even one month or seven months? And how many spells of one, two, or even three months duration were not reported at all?
While there may be ways to tease out more information from the SIPP data, we have not attempted to do so here.(2) Instead we simply recognize the limitations of these data. Specifically, despite what the SIPP data appear to tell us about the length of short spells of uninsurance, we would suggest that these data should not be used to draw inferences about the relative frequency of spell lengths shorter than six months or about longer spell lengths in groupings of fewer than four months. In other words, it may be reasonable to infer from Table 1 that about 60 percent of new spells of uninsurance in FY93 were completed in six months or less, but the data will not support inferences about how many of these spells may have been completed in two months or less, or four months or less. Similarly, it may be correct to infer that about 8 percent of spells lasted 9 to 12 months or that 3.5 to 4 percent lasted 15 to 18 months but certainly not that only .1 percent were completed in exactly 13 months. In general, we need to be conscious that the numbers reported at 4 months, 8 months, 12 months, or 16 months almost certainly include more spells of shorter or longer lengths than they do spells of the nominal length.
Spell lengths of six months or longer or 12 months or longer are becoming important in defining eligibility for CHIP coverage in a number of states. What can we infer about the relative frequency of such spells from the data presented in Table 1? Read literally, the data in Table 1 tell us that about 43 percent of the spells that began in FY93 lasted six months or longer (100 minus the 57.3 percent with lengths of five months or less), and 26.4 percent of the new spells in that year lasted 12 months or longer. Recognizing that most of the spells reported as having durations of exactly four or 12 months were actually longer or shorter than that, we would suggest that the true proportion of new spells with durations of six months or longer could be as much as 5 percentage points higher--up to 48 percent--and that the true proportion of spells with durations of 12 months or longer could be 2 to 3 percentage points lower than the observed figure, or closer to 24 percent.
Where Table 1 presented distributions of new spells of uninsurance that began in FY93, Table 2 presents distributions of all spells that were active at the end of FY93. For all children who were uninsured in September 1993, Table 2 tells us how long they had been uninsured. The distribution of these “current” durations can be compared to the distribution of completed durations of new spells, which is reproduced from Table 1 in the right hand columns of Table 2.
It is important to understand how the durations of active spells reported in the left side of Table 2 differ from the durations of new spells reported in the right side (and in Table 1). First, the durations of active spells are incomplete. More specifically, they are “right censored.” If we think of a duration as spanning from left to right on a time line, in other words, what we do not observe for these active spells is how far their durations extend to the right. All of the spells terminate at the same point in time--September 1993. We do not know how long these durations will be when they are completed. The second respect in which the active spells portrayed in Table 2 differ from the completed spells presented earlier is that they share a common month. All of the active spells
TABLE 2 (PDF document.)
include September 1993. While there is a lot of overlap in the spells reported in Table 1, these spells do not share a common month.
These two differences have important implications for how the distributions of durations differ. First, let us consider what the durations reported on the left side of Table 2 represent. Durations of one month represent spells that started in September. Ultimately, very few of these spells will end at one month. Instead, if we could follow them through to completion, we would find that they have essentially the same distribution as the spells reported in Table 1. Durations of two months refer to all spells that started in August 1993, minus those that ended in that same month--that is, spells with completed durations of one month. If we were to follow these spells to completion, they too would have the same distribution as the spells reported in Table 1 except for the complete absence of spells of one month duration. Similarly, the 12-month durations include all spells that started in October 1992 minus those that ended in any of the months from October 1992 through August 1993.(3)
What we notice first in the distribution of active spells is the absence of a seam effect. There are no peaks whatsoever at multiples of four months. This absence of a seam effect is due entirely to the right censoring. Most of these spells are not complete, so the seam effect is hidden. One quarter of the spells that will eventually end in four months shows up as one month in length, one quarter shows up as two months in length, one quarter appears as three, and one quarter as four months in length. Far from producing peaks at four months, the seam effect actually contributes to the relative evenness of the distribution of spells across durations of one to four months.
The second thing we notice is that apart from the higher proportion of spells with durations of three months or less, the spells that are active at the end of the year (and this would be true of spells that were active during any given month of the year) are longer, on average, than the spells that started during the year--even though most of the active spells are not yet complete. The median length of the active spells lies between 9 and 10 months whereas the median completed length of the new spells lies close to 4 months. Nearly 46 percent of the active spells have durations in excess of 12 months compared to just 20 percent of the spells that started during the year.
If we followed the active spells to their completion, we would obtain a distribution that is heavily skewed toward long durations. Why is this? It turns out that the representation of spells among active spells is directly proportional to their completed duration. Spells of one month duration are represented solely by spells that began in a single month: September 1993. Spells of two months duration are represented solely by spells that began in either of two months: August or September. Spells of three months duration are represented by spells that began in any of three months, and so on. Thus spells of 12 months duration are represented by spells that began in any of 12 months--October 1992 through September 1993--while spells of 36 months duration are represented by spells that began in any of 36 months--from October 1990 through September 1993. Compared to a distribution of spells starting in the same month or the same year, therefore, spells of exactly 36 months in length are represented at 36 times their relative frequency.
We could estimate the distribution of completed durations among spells that were active in September 1993 by taking the completed distribution of new spells reported in Table 1 and multiplying the number of one-month spells by 1, the number of two-month spells by 2, number of three-month spells by 3, and so on. Obviously, we cannot complete this exercise with the data presented in Table 1 because we do not know the completed duration of spells that exceeded 18 months in length. But if we make an assumption about the average completed duration of spells that were completed in more than 18 months, then we can estimate the relative frequency of such spells among active spells.
Table 3 displays this projected distribution of the completed duration of spells that were active in September 1993. These calculations assume, of course, that the distribution of completed durations did not change over time, so that the relative frequency of, say, an 18-month spell is the same for spells starting at any time between October 1991 and September 1993. Whether or not this assumption is satisfied, the projected distribution illustrates our point. In Table 3 we see that only 22 percent of these spells are completed in six months or less compared to about 60 percent of the new spells in Table 1. At the other end of the distribution, 49 percent of the spells reported in Table 3 extend beyond 18 months compared to only 15.5 percent of the spells in Table 1.
If these different ways of measuring the duration of uninsurance yield such different distributions, can they all be useful? The answer is yes, but the different distributions address different questions. Table 1 reflects a representative sampling of spells. It tells us that most spells of uninsurance among children are relatively short in length, 60 percent being six months or less. The distribution has a long tail, however, with at least 20 percent of spells (we suggested somewhat higher) exceeding 12 months in length and about three-fourths of these surpassing 18 months. It is this long tail that makes possible the skewed distribution that we see when we look at the spells of a representative sample of people with spells active in a give month. People who experience long spells are more likely to be uninsured in a given month than people who experience short spells, so children who are uninsured at a point in time exhibit disproportionately many long spells. Table 3 suggests that about half of the 9 million children who were uninsured in any given month of FY93 will have had spells of more than 18 months in length by the time they become insured (again), and most of these spells will extend well beyond 18 months. Finally, Table 2 addresses a question that
TABLE 3 (PDF document.)
is becoming very pertinent with the introduction of CHIP--namely, if eligibility is limited to children who have been uninsured for a specified number of months, how many children will satisfy these criteria? Table 2 indicates that with a 12 month minimum, about 47 percent of all uninsured children would meet this requirement. The percentage among children who meet the income and other requirements is likely to be higher. Lowering the duration requirement to 6 months would add another 16 percent of children--a relatively modest increase given that the minimum duration would be cut in half.
The large number of uninsured children who appear to be eligible for Medicaid has captured the attention of policymakers and advocacy groups alike. In Technical Appendix A we present estimates that 30 to 33 percent of the children who were uninsured at a point in time in FY93 or FY94 were eligible for Medicaid, based on an eligibility simulation that accounted for a substantial majority but certainly not all Medicaid-eligible children. Here we ask how long children who are uninsured and eligible for Medicaid tend to remain both uninsured and Medicaid-eligible.
Table 4 provides information on the duration of spells of Medicaid-eligible uninsurance. A spell of Medicaid-eligible uninsurance begins in the first month that a child is both uninsured and Medicaid-eligible. A child may have been uninsured for one or more months before becoming Medicaid-eligible or may have been Medicaid-eligible at the start of a spell of uninsurance. Likewise, a child may have lost Medicaid eligibility before becoming insured or may have remained eligible through the end of a spell of uninsurance.(4) Because patterns may vary, we cannot infer the full length of a spell of uninsurance from the number of months that a child was both Medicaid eligible and uninsured.
TABLE 4 (PDF document.)
We began our Medicaid-eligibility simulation in October 1992, so we cannot identify new spells of Medicaid-eligible uninsurance starting before November 1992. (We need to observe the previous month to determine if a new spell has started.) The spells reported in Table 4, therefore, began in November 1992 or any of the subsequent months through September 1993.
Spells of Medicaid-eligible uninsurance tend to be much shorter than spells of uninsurance generally. More than one-half of all spells started in FY93 were completed in three months or less, and only 6 percent extended beyond one year. Among all spells of uninsurance started during the same period, 20 percent reached beyond one year, and more than 15 percent extended at least another six months beyond that.
Finally, the SIPP seam effect is much less pronounced for spells of Medicaid-eligible uninsurance than for all spells of uninsurance. We attribute this to the fact that simulated Medicaid- eligibility is not a characteristic that is reported by survey respondents or recorded by survey interviewers but a complex construct of a great many variables that change at different rates and times. Nevertheless, there are local peaks at four, eight, and 12 months. Taking this into account suggests that the proportion of spells of Medicaid-eligible uninsurance completed in three months or less could be as high as 60 percent.
C. COVERAGE BEFORE AND AFTER SPELLS OF UNINSURANCE
Other than children who are uninsured at birth, children who begin spells of uninsurance do so by losing previous coverage. What is the source of this coverage? Does it mirror the coverage of all insured children, or do the uninsured come disproportionately from former Medicaid enrollees? Similarly, when children regain coverage do they resemble the rest of the insured population with respect to the sources of this coverage or do they rely on a different mix of sources? And do children tend to return to the same type of coverage that they had prior to their spell of insurance, or do they often end up with another source of coverage? We address these questions by comparing the source of coverage before and after completed spells of uninsurance. We also examine how these answers differ by the duration of spells of uninsurance. After addressing these question with respect to spells of uninsurance, we examine spells of Medicaid-eligible uninsurance to determine what precedes and follows these spells. Here, however, a spell need not be preceded by coverage; instead a child may enter a spell of Medicaid-eligible uninsurance from the state of having been uninsured but not eligible for Medicaid, and may leave a spell of Medicaid-eligible uninsurance by losing Medicaid eligibility without regaining any form of coverage.
Table 5 presents for spells of uninsurance of various durations a cross-tabulation of insurance coverage in the month before the spell began by insurance coverage in the month after the spell ended (that is, the first month of coverage). Separate tabulations are reported, first, for all spells completed in 1 to 12 months, followed by tabulations of spells completed in 1 to 4 months, 5 to 8 months, and 9 to 12 months. A tabulation of spells completed in 13 to 18 months, derived from spells that started in just the first half of FY93, is reported at the bottom of the table. We did not pool this subset of spells with those completed in 12 months or less because the spells of longer duration would be underrepresented in the pooled tabulation.(5)
TABLE 5 (PDF document.)
The estimates reported in each subtable are percentages of the total number of spells in each duration group. Thus the percentages in the total column at the right give the distribution of insurance coverage prior to the spell of uninsurance; the percentages in the total row beneath each subtable give the distribution of insurance coverage after the spell of uninsurance; and the percentages inside the row and column margins give the distribution of spells by combinations of prior and subsequent insurance coverage.
For all spells of uninsurance completed in 12 months or less, 54 percent were preceded by employer-sponsored coverage, 39 percent were preceded by Medicaid, 4.5 percent were preceded by another form of coverage, and 2.5 percent were experienced by children who were not in the survey universe prior to the spell. These last would be newborns who were without health insurance at birth. The distribution of sources of coverage following a spell of uninsurance is very similar to the distribution prior to the spell, with 53 percent being covered by an employer-sponsored plan, 40 percent by Medicaid, 5 percent by another type of plan, and 1.6 percent leaving the survey universe.(6)
In October 1992, based on these same data, about 19 percent of insured children were covered by Medicaid, 76 percent by an employer-sponsored plan, and 5 percent by another type of plan. Thus the frequency of Medicaid as the source of coverage before and after a spell of uninsurance lasting 12 months or less is more than double the frequency of Medicaid among all insured children. Coverage other than Medicaid or an employer-sponsored plan occurs with about the same frequency among all insured children as among those who later or previously experienced spells of uninsurance, while employer-sponsored coverage occurs with lower frequency before or after a spell of uninsurance than it does among all insured children.
The similarity of the distributions of sources of coverage before and after spells of uninsurance does not necessarily mean that children end their spells of uninsurance by returning to the same coverage that they lost to begin their spells, but we see from the cross-tabulation of coverage before and after uninsurance that the two sources are clearly not independent. For all spells of uninsurance completed in 12 months or less, 28 percent were preceded and followed by Medicaid enrollment, and 42 percent were preceded and followed by employer-sponsored coverage. About 9 percent of children changed from employer-sponsored coverage to Medicaid, and somewhat fewer children made the reverse transition.
Among spells completed in 12 months or less, we see little variation in these basic patterns by duration. But spells completed in 13 to 18 months present a very different picture, showing an even greater prominence of Medicaid. Prior to a spell lasting 13 to 18 months, 48 percent of the children were insured by Medicaid and 52 percent by an employer-sponsored plan. We found no incidence of other sources of coverage and no children who remained uninsured for 13 months past their births. Following a spell of 13 to 18 months, 57 percent of children enrolled in Medicaid compared to 32 percent becoming covered by an employer-sponsored plan. Medicaid was the source of coverage both before and after a long spell of uninsurance among 38 percent of the children who experienced such spells. Employer-sponsored coverage preceded and followed 22 percent of these spells--barely half the percentage among shorter spells. Children who left a spell of uninsurance that was preceded by employer-sponsored coverage were almost as likely to end their spells by acquiring Medicaid coverage as employer-sponsored coverage, and other coverage was nearly half as common as Medicaid. A possible explanation for the differential importance of Medicaid by duration is that parents who lose employer-sponsored coverage tend to defer pursuing other coverage for their children because they expect to regain employer-sponsored coverage. If they are unable to regain employer-sponsored coverage within a year or so, they try to obtain Medicaid. It could be, as well, that many of these children did not qualify for Medicaid in the months after employer-sponsored coverage was lost and became eligible only after their parents spent down savings and other resources.
2. Spells of Medicaid-Eligible Uninsurance
A spell of uninsurance in which a child is eligible for Medicaid can begin or end with the child losing Medicaid eligibility without regaining insurance coverage. Table 6 presents cross-tabulations of coverage before and after spells of Medicaid-eligible uninsurance, but it differs from Table 5 in that it includes uninsurance without Medicaid eligibility as a status before and after a spell.
Nearly one-half of children who experienced spells of Medicaid-eligible uninsurance lasting 12 months or less were uninsured but ineligible for Medicaid before beginning their spells, and 31 percent left spells of Medicaid-eligible uninsurance by losing their Medicaid eligibility. Medicaid enrollment preceded 27 percent of the spells of Medicaid-eligible uninsurance and followed 31 percent of them. Only 19 percent of children who experienced spells of Medicaid-eligible uninsurance were covered by employer-sponsored insurance immediately prior to their spells, and just under 15 percent gained employer-sponsored coverage to end their spells. Uninsurance without Medicaid-eligibility bracketed 34 percent of the 1 to 12 month spells of Medicaid-eligible uninsurance while actual Medicaid enrollment bracketed 16 percent of the spells. Employer- sponsored coverage preceded and followed only 6 percent of spells.
As we have seen, spells of Medicaid-eligible uninsurance tend to be quite short, so the spells completed in 12 months or less are dominated by spells completed in 1 to 4 months. The pattern of
TABLE 6 (PDF document.)
coverage before and after spells of Medicaid-eligible uninsurance changes dramatically as the duration of these spells increases. Until durations of 13 months or longer the changes are largely confined to the distribution of coverage prior to the spell of Medicaid-eligible uninsurance. Medicaid enrollment becomes more important than Medicaid ineligibility, accounting for more and more of the coverage prior to such spells. Among spells completed in 13 to 18 months, 43 percent began with children leaving Medicaid enrollment compared to 31 percent that started with uninsured children becoming Medicaid-eligible. The reverse trend occurred for the source of coverage following a spell of Medicaid-eligible uninsurance, although there are minimal changes between spells of 1 to 4 month duration and spells of 9 to 12 month duration. Among the longer spells, 65 percent ended with children losing Medicaid eligibility but remaining uninsured. Only 6 percent of the children gained employer-sponsored coverage while 29 percent enrolled in Medicaid.
In looking at the combinations of coverage that precede and follow 13 to 18 month spells of Medicaid-eligible uninsurance, we find that 29 percent of these spells are bracketed by Medicaid enrollment and 31 percent are bracketed by uninsurance without Medicaid eligibility. We found no instances of children having employer-sponsored coverage both before and after a spell of Medicaid- eligible uninsurance. All of the children who lost employer-sponsored coverage to become uninsured but eligible for Medicaid ended their spells by losing their Medicaid eligibility rather than enrolling in Medicaid. This is quite different from what we observed with all spells of uninsurance, but we cannot tell from Table 6 what may have happened to these children after losing Medicaid eligibility. Many of them may have gained employer-sponsored coverage, and some could have regained Medicaid eligibility and become enrolled. Finally, in viewing the bottom panel of Table 6 we need to keep in mind that, according to Table 4, these spells account for less than four percent of all spells of Medicaid-eligible uninsurance that began in FY93.
In the previous section we examined transitions into and out of uninsurance but restricted our attention to completed spells of uninsurance and, even more narrowly, completed spells of Medicaid- eligible uninsurance.. In this section we address the issue of transitions more broadly, looking first at changes in health insurance coverage at roughly one- and two-year intervals, then month-to-month transitions and, finally, cumulative monthly exits from uninsurance and Medicaid enrollment.
Table 7 provides a cross-tabulation of health insurance coverage in October 1992 by health insurance coverage in September 1993 and in September 1994. The top row reports the number of children who were in each coverage group in October 1992. Below that is reported, for each October 1992 coverage group, the percentage distribution of children in September 1993 and, in the lower panel, September 1994.
Of the 12.6 million children who were reported as enrolled in Medicaid in October 1992, about 80 percent were still enrolled in September 1993 and 74 percent in September 1994. These figures do not reflect continuous enrollment, necessarily. A child who left Medicaid during 1993 but returned by September would be included in the 80 percent. Nevertheless, they indicate a very high degree of continuity in the reliance on Medicaid. Most of those who were not enrolled in Medicaid one or two years later were uninsured: 11 percent in September 1993 and nearly 14 percent in September 1994. Most of the remainder--7 percent in September 1993 and close to 10 percent in September 1994--acquired employer-sponsored coverage.
Of the 9.5 million children who were uninsured in October 1992, 62 percent were still uninsured in September 1993 and 52 percent in September 1994. Those who became insured were more likely to find such coverage in employer-sponsored plans than in Medicaid but at nowhere near the four
TABLE 7 (PDF document.)
to one rate that characterizes the dominance of employer-sponsored coverage over Medicaid among all insured children. In September 1993, 20 percent of the children who were uninsured 11 months earlier had employer-sponsored coverage while 14 percent were enrolled in Medicaid. One year later, 25 percent had employer-sponsored coverage, and 17 percent were covered by Medicaid. Other coverage accounted for 2.5 percent of these children in September 1993 and 3.7 percent in September 1994 while 2.1 percent had left the SIPP population by the first date and 3.2 percent by the second. In September 1993 we also find that 4.4 percent of those with employer-sponsored coverage in October 1992 had become uninsured compared to 8.6 percent of those with other coverage. By September 1994 the percentage uninsured among those with employer-sponsored coverage in October 1992 had risen modestly to 6.1 percent while the percentage uninsured among those with other coverage initially had actually fallen slightly.
Changes in the health insurance coverage of individuals between points in time one and two years apart give us a very good idea of the magnitudes of gross flows of children into and out of particular types of coverage over the span of one and two years. But they do not show all movement. In particular, they do not capture multiple changes by the same individuals. To gain a different perspective on changes in health insurance coverage over time, we examine changes between one month and the next. More specifically, we examine month-to-month transitions into and out of uninsurance and into and out of Medicaid.
How many children leave uninsurance each month and how many enter uninsurance each month? Tables 8 and 9 present estimates of monthly transitions out of and into uninsurance, respectively. Each of these tables is based on 23 successive cross-tabulations of insurance status between one month and the next.
TABLE 8 (PDF document.)
TABLE 9 (PDF document.)
For each month from October 1992 (9210) through August 1994 (9408), Table 8 reports the total number of children without health insurance, followed by a percentage distribution of their coverage in the next month, the number who became insured, and a percentage distribution of the type of insurance coverage that these latter acquired. Average monthly estimates of each of these quantities are presented at the bottom of the table, and we consider these first. On average, 9 percent of the roughly 9 million uninsured children left that state each month, with 8.3 percent becoming insured and the remainder leaving the survey universe or aging out of the child population.(7) Of the 752,000 who became insured each month, 41 percent enrolled in Medicaid, 52 percent gained coverage under an employer-sponsored plan, and 7 percent obtained coverage by another type of insurance.(8)
The 752,000 children, on average, who became insured between one month and the next plus the additional 64,000 children who left the survey universe or turned 19 years old each month represent a sizable number. Outflows of this magnitude would have depleted the stock of uninsured children in a little over a year had there been no flows in the opposite direction. Of course, there were flows in the opposite direction, and they have to compare in magnitude to the flows out of uninsurance as the net change in the number of uninsured children over the 23 months is less than the outflow in a single month. Moreover, we know from Table 7 that 62 percent of the children who were uninsured in October 1992 were still uninsured in September 1993, and nearly 52 percent were still uninsured a year later. From the relative frequency of short durations among new spells of uninsurance, reported earlier, we can infer that most of the children who exited uninsurance during FY93 and FY94 did not become uninsured until after the start of the period. It is because of the number of children flowing into uninsurance, as we shall see below, that 9 percent of uninsured children could leave that population month after month with little effect on the total number of children who were uninsured at a point in time.
The individual monthly results that make up the bulk of Table 8 are of interest for what information they may provide on seasonal patterns and trends in exits from uninsurance. In research conducted for the Food and Nutrition Service, using the 1992 and 1993 SIPP panels, Czajka et al. (1998) found clear evidence of seasonality in the total number of children in poverty. In the research presented in Technical Appendix A, however, we found no evidence of seasonality in the total number of uninsured children. If there are any seasonal patterns in the movement of children into and out of uninsurance, of course, they would show up most clearly in the numbers of children entering and leaving uninsurance each month and in the distribution of their sources of coverage before and after spells of uninsurance.
We see no evidence of seasonality in any aspect of the monthly transitions reported in Table 8. Nor do we see clear evidence of a trend in the number becoming uninsured, even though the number without insurance declines by more than 700,000 over the 23 months. While the monthly estimates of children leaving uninsurance bounce around quite a lot, varying from a low of 564,000 to a high of 884,000, this variation appears to be due entirely to sampling error.
Table 9 reports estimates of the number of children who were without health insurance in November 1992 through September 1994, the percentage distribution of their coverage in the preceding month, the number who became uninsured in the current month, and the percentage distribution of insurance status in the preceding month among those who became uninsured. On average, 791,000 children became uninsured each month. Of this number, 52 percent had employer coverage in the preceding month, 39 percent had Medicaid coverage, about 6 percent had other coverage, and less than 3 percent were newborns or children returning to the survey universe. The distribution of prior month coverage among those becoming uninsured compares fairly closely to the distribution of prior month coverage among children who began spells of uninsurance in FY93 that lasted one to 12 months (see Table 5). As we explained above, with reference to the distribution of coverage after completion of a spell of uninsurance, this similarity of the two distributions is not at all surprising, given that Table 8 refers to all new spells that started in FY93 or FY94 (exclusive of October 1992) and Table 5 refers to a large subset of the spells that began in FY93.
In the monthly results the numbers who became uninsured in September 1993 and 1994 stand out because they are the only estimates in excess of one million. Nevertheless, there is no other evidence of seasonal change in the monthly estimates, and so we infer that there is no seasonality in the movement of children into uninsurance. Similarly, we see no seasonal swings in any of the components of change. These vary a lot from month to month but with no clear pattern, and so we conclude that their variation is due to sampling error, essentially. Nor do there appear to be any long-term trends in either the number becoming uninsured or the composition of coverage in the preceding month. The final months show a rise in the number becoming uninsured and, with it, the proportion of uninsured children who were insured the preceding month, but these are strictly short- term changes that in reality may represent nothing more than sampling error.
Table 10 shows for each month from October 1992 through August 1994 the number of children under 19 who were enrolled in Medicaid, a percentage breakdown of the types of coverage they had in the next month, the number of children leaving Medicaid in the next month, and a percentage distribution of the type of coverage that followed Medicaid. Average monthly transitions are reported at the bottom of the table. On average about 4 percent of Medicaid enrollees under 19 years of age or more than half a million children (555,000) left Medicaid each month. Of these, 56 percent became uninsured while about 37 percent acquired another form of insurance and 7 percent left the survey universe or aged out of the population of children under 19.
Table 11 reports monthly transitions into Medicaid. Of the 583,000 entering Medicaid each month, on average, just over half or 53 percent were previously uninsured, 34 percent had employer- sponsored or other insurance coverage, and 14 percent were either new to the survey universe or returning to the survey universe. In fact, virtually all if not all of those entering the survey universe were newborn children who were enrolled in Medicaid at birth (an extension of their mothers’ coverage in pregnancy, no doubt). The size of this group reflects both the high rate of Medicaid coverage among infants and the disproportionately large share of the Medicaid population that infants represent.
TABLE 10 (PDF document.)
TABLE 11 (PDF document.)
Neither Table 10 nor Table 11 shows any evidence of seasonality or trend in the exits from or entrances into Medicaid enrollment. That Medicaid enrollment among the SIPP population of children under 19 grows by nearly 500,000 over the period can be attributed to the fact that new enrollments tended to outnumber exits from Medicaid in each month rather than one series growing or declining more rapidly than the other.
The monthly transitions reported in Table 8 show that about 8 percent of the children who were uninsured at any point in time in FY93 or FY94 became insured in the next month while Table 7 indicates that 36 percent of the children who were uninsured in October 1992 were insured 11 months later and 46 percent were insured 23 months later. Similarly, the monthly transitions reported in Table 10 show that, on average, about 4 percent of the children enrolled in Medicaid at any time in the two-year period left the program the next month while Table 7 shows that about 20 percent exited over the course of 11 months and 26 percent over 23 months. In each case the change over 11 or 23 months is considerably less than we would see if each cohort of children who were uninsured or enrolled in Medicaid in a given month experienced the same exit rates month after month.(9) Here we examine the cumulative effect of monthly transitions on children who started out in a particular state of health insurance coverage in October 1992. For each of three initial statuses-- uninsurance, Medicaid-eligible uninsurance, and Medicaid enrollment--we present estimates of the distribution of children by coverage status in each of the next 23 months. Over time, then, we see what fraction of the original population of, say, uninsured children is still uninsured in each of the next 23 months and what fraction has moved to either Medicaid, employer-sponsored insurance, or other insurance.
Table 12 presents the distribution of children who were uninsured in October 1992 by their health insurance coverage in each of the next 23 months.(10) As we saw in Table 8, about 9 percent leave the uninsured population in the first month. This exit rate begins declining immediately, dropping to between 2 and 3 percent within four months and to about 1 percent within another four months. By the final months of FY94, the percentage remaining uninsured is declining by less than one percentage point per month (or less than 2 percent of those remaining uninsured). Initially, children who leave uninsurance are about equally likely to enroll in Medicaid and obtain employer- sponsored coverage, but after two months those children who have left uninsurance are increasingly more likely to be covered by employer-sponsored coverage than by Medicaid. For some of the children who end up with employer-sponsored coverage, the path from uninsurance to employer- sponsored coverage may pass through Medicaid. That is, children may leave uninsurance by obtaining Medicaid coverage but then move from Medicaid to employer-sponsored coverage at a later date. We cannot tell from this table the paths that children take. On the other hand, it is quite clear from these data that as long as 23 months after children have left uninsurance, Medicaid remains a much more important source of coverage than it is among all insured children. In September 1994, children who were uninsured in October 1992 are less than one-and-a-half times
TABLE 12 (PDF document.)
as likely to have employer-sponsored coverage as Medicaid (24.8 percent versus 16.8 percent of all children who were uninsured in October 1992). Among all insured children in September 1994 who were under 19 and part of the SIPP universe in October 1992, the number covered by employer- sponsored plans is nearly four times the number covered by Medicaid (45.5 million versus 12.3 million).
Table 13 reports the distribution of children who were uninsured and eligible for Medicaid in October 1992 by their health insurance coverage in each of the next 23 months. Within one month, more than 20 percent of the children who were Medicaid-eligible and uninsured had changed to another coverage status. About 7 percent remained uninsured but were no longer eligible for Medicaid while a comparable percentage had enrolled in Medicaid. Just over 5 percent had employer-sponsored coverage while only .2 percent had other coverage. About .6 percent had left the population. After two months, the percentage remaining Medicaid-eligible and uninsured had dropped to under 65 percent, and within seven or eight months it had reached 40 percent. After 23 months, only 26 percent of the original population of children who were Medicaid-eligible and uninsured at the outset remained part of that population. Note, however, that 21 percent of the original population were still uninsured. Comparing Tables 13 and 12, we see that the Medicaid- eligible uninsured were only slightly more likely to have left the state of uninsurance than all uninsured: 52.5 percent versus 48.5 percent. Even so, Medicaid remained the dominant source of coverage among those who left uninsurance, accounting for 27 percent of the original population compared to 18 percent with employer-sponsored coverage. This represents a reversal of the relative importance of these two sources of coverage among all children who were uninsured in October 1992.
TABLE 13 (PDF document.)
To highlight the impact of Medicaid eligibility, Table 14 presents the distribution of children who were uninsured but not eligible for Medicaid in October 1992 by their health insurance coverage in each of the next 23 months. These children left uninsurance somewhat more slowly than those who were initially eligible for Medicaid. After one month, 86 percent remained uninsured and not eligible for Medicaid while another 6.5 percent had gained Medicaid eligibility but remained uninsured. About 3 percent had enrolled in Medicaid and another 3 percent had obtained employer- sponsored coverage. These figures imply a Medicaid participation rate of about one third for children who are uninsured and newly eligible for Medicaid. Within a few months the Medicaid eligible are evenly split between those participating and those not participating in Medicaid. Toward the end of the 23-month period, the Medicaid participants are between one-and-a-half and two times the number of Medicaid-eligible nonparticipants, implying a Medicaid participation rate in excess of 60 percent.
Within four months of October 1992, children who have left the uninsured population are twice as likely to be covered by an employer-sponsored plan as Medicaid. This relative importance of the two sources of coverage among children who started out as uninsured and not eligible for Medicaid persists through September 1994.
Table 15 reports the distribution of children enrolled in Medicaid in October 1992 by their health insurance coverage in each of the next 23 months. Children without health insurance coverage are apportioned between those who are eligible for Medicaid and those who are not. As reported in Table 10, about 3 to 4 percent of the children enrolled in Medicaid at any point in time leave the program within the next month. Table 15 shows that this rate of departures from Medicaid persists over as many as four months before slowing down. By the fourth month, 14 percent of the
TABLE 14 (PDF document.)
TABLE 15 (PDF document.)
original Medicaid population has exited, with 5.5 percent picking up employer-sponsored coverage, .2 percent acquiring other coverage and an equal number leaving the survey population. Altogether 8 percent have become uninsured, with nearly 5 percent remaining Medicaid eligible. This latter percentage changes only marginally thereafter. The fraction of all October 1992 Medicaid participants classified as uninsured but still eligible for Medicaid never rises above 6.0 percent whereas the proportion who are uninsured but not eligible for Medicaid continues to ascend, gradually, reaching 8 percent by September 1994.
After four months, the proportion of October 1992 Medicaid children who are covered by employer-sponsored insurance stands at 5.5 percent. This proportion rises only another four percentage points over the next 19 months, reaching 9.6 percent by September 1994. In sum, the picture presented by Table 15 is one of very slow change in the population of children who start FY93 enrolled in Medicaid. Two years later nearly three-quarters of these children are still enrolled in Medicaid, with only 11 percent having moved to some other type of coverage and nearly 14 percent having become uninsured. More than a third of the uninsured appear to have retained their Medicaid eligibility. While this is puzzling, it is important to keep in perspective the small size of this group. At 5.5 percent of those who were enrolled in Medicaid in October 1992 they number 692,000, which makes them barely more than one quarter of the 2.6 million children who were uninsured and Medicaid-eligible in September 1994.
The results presented in the preceding section indicate that about 600,000 children, on average, enrolled in Medicaid each month of FY93 and FY94 while nearly 800,000 became uninsured. Comparable numbers of children left Medicaid or became insured, so that the net change in the Medicaid child caseload or the population of uninsured children was small. Movements of children out of and into Medicaid or uninsurance include both “churning”--that is, exits entries by the same children over a relatively brief interval--and “turnover,” or exits and entries by different children.
Whether we classify a particular transition into Medicaid or uninsurance as an example of churning or not may depend on how we define the reference period for determining prior Medicaid enrollment or uninsurance. There’s no question that a child’s becoming uninsured after being covered for only six months should be included within the concept of churning.(11) But what about a child who becomes uninsured after 13 months of coverage--or 24 months? Whatever we take churning to mean conceptually, moreover, measuring churning with available data may be quite another matter.
With data from the SIPP one could evaluate each transition into Medicaid or uninsurance within a period and determine whether and how long a child may have last been enrolled in Medicaid or uninsured. For transitions occurring in calendar year 1993, for example, one could look back as far as 12 months for children making transitions into Medicaid or uninsurance in January 1993 and as far back as 23 months for children making transitions in December 1993. For transitions occurring in calendar year 1994, one could look back 24 months for children making transitions in January 1994 and up to 35 months for children making transitions later in the year. Maintaining a consistent operational definition of churning would require that we look back no farther for children making transitions late in the period than we are able to do for children making transitions early in the period. This would imply that for transitions occurring in 1993 we look back at most 12 months and for transitions occurring in 1994 we look back at most 24 months.
Here we take a somewhat different approach to measuring the relative amount of churning and turnover in the transitions that we observe into Medicaid enrollment and uninsurance. We do so in order to pull together the findings that we presented earlier and in Technical Appendix A. From the data presented in Tables 9 and 11, we can count the number of transitions of children into Medicaid and uninsurance over a 23 month period, from November 1992 through September 1994. We also know the number of children enrolled in Medicaid or uninsured at the beginning of the period, October 1992 (or any month along the way). Finally, from data presented elsewhere, we know how many children under 19 were ever enrolled in Medicaid and how many were every uninsured in each of FY93 and FY94 and over the full two-year period. Subtracting enrollment or uninsurance at the beginning of a period from the number of children who were ever enrolled or ever uninsured during the period gives us a measure of aggregate turnover. Specifically, it tells us how many additional children entered Medicaid or uninsurance over the period, which we can express as a percentage of the initial enrollment. Subtracting the aggregate turnover from the total number of transitions into Medicaid or uninsurance over the period gives us a measure of excess transitions, which represent churning. In effect, then, we are counting each transition as an instance of either turnover or churning on the basis of whether the individual making the transition was in that state at the beginning of the period. This is not a time-based definition as we discussed above, but it provides a measure that is not only very intuitive but indicative of how much change occurred in the Medicaid or uninsured population over the period.
To obtain the population uninsured at the beginning of a period requires a small calculation.(12) Children uninsured in the first month of a period include children who were uninsured before the period began and children for whom this is their first full month of uninsurance.(13) Children ever uninsured in the period include those with one or more months of uninsurance. To count how many children became uninsured during the year we must compare the number ever uninsured during the year to those uninsured at the beginning of the year. If we equate all children who were uninsured in the first month--that is, the point-in-time estimate for that month--with the number uninsured at the beginning of the year, then our estimate of children who became uninsured during the year does not include those who became uninsured in the first month. To count children who became uninsured in the first month as well as the remaining 12, we need to redefine the population at the beginning of the year as children uninsured in the first month minus those who began spells of uninsurance in that month.(14)
Our estimates of the components of change in the population of uninsured children are presented in Table 16. For FY93 we start with about 9.5 million uninsured children in October 1992. From this we subtract the estimated 551,000 children who became uninsured in that month to obtain the number of children who were uninsured at the beginning of the year: 8.9 million.(15) Altogether 16.1 million children were ever uninsured during FY93. Subtracting the initial population from this number yields our estimate of the number of children who became uninsured in FY93: approximately 7.2 million. This estimate of the turnover in the uninsured population represents 80
TABLE 16 (PDF document.)
percent of the number who were uninsured at the start of the period and 44 percent of the number who were ever uninsured during the period. We obtain somewhat higher estimates of turnover in FY94, with the children who were added to the uninsured population in that year representing 90 percent of the number uninsured at the beginning of the year and 47 percent of those who were ever uninsured during the year. For the two-year period FY93 and FY94 we find that 12.1 million children became uninsured over this time or 136 percent of the number who were uninsured at the beginning of FY93 and 58 percent of the number who were ever uninsured during the two years. That the 12.1 million new uninsured is not 15 million (the sum of the FY93 and FY94 additions) implies that about 3 million of the 7.6 million children who were added to the uninsured after the beginning of FY94 were uninsured at some point during FY93. These children will figure into our estimate of churning over the two-year period.
Nearly 9.3 million new spells of uninsurance were started in FY93 and nearly 9.5 million in FY94, for a two-year total of 18.7 million. In FY93 the number of new spells exceed by 2.1 million the number of children who were added to the uninsured in that year. These excess spells were started by children with one or more other spells during the year--that is, spells that were already in progress at the beginning of the year or spells that began during the year. These excess spells represent 23 percent of all the new spells that started during the year. Similarly, for FY94 we estimate that 1.9 million of the 9.5 million spells of uninsurance that started during the year were associated with children who had one or more other spells of uninsurance during the year. For the two-year period, the number of excess spells is more than just the sum of the numbers of such spells in FY93 and FY94. The total of 6.6 million includes spells started by children with no other spells in FY94 but with spells in FY93 that ended before the start of FY94. Measured in this way, churning accounts for 35 percent of all the new spells that were started in FY93 and FY94. By implication, if we were to define churning in FY94 in terms of spells started by children who had other spells of uninsurance in either FY94 or FY93, rather than just FY94, the number of such spells would equal the difference between the 6.6 million two-year total and the 2.1 million spells started in FY93 by children who had one or more other spells of uninsurance during FY93. This difference of about 4.5 million represents 47 percent of the new spells started in FY94. In other words, 47 percent of the new spells of uninsurance started in FY94 represented second, third, or even higher order spells by children who had been uninsured once previously in either FY93 or FY94. If we were to look for earlier spells as far back as two years for all children who started new spells in FY94, rather than just those who started new spells at the end of FY94, we would obtain an even higher estimate of churning. On the other hand, if we were to limit our definition of churning to include only those new spells started within a year of the last spell, then the estimate of churning in FY94 would lie somewhere between 20 percent and 47 percent.(16)
Table 17 presents analogous estimates of the components of change in the population of children enrolled in Medicaid in FY93 and FY94. While the number of children enrolled in Medicaid at a point in time is about one-third higher than the number of children who are uninsured, both the number of new spells started during a year and the number of new enrollees that these spells represent are smaller than the corresponding numbers for the uninsured. In FY93, new enrollees represented 48 percent of the number enrolled at the beginning of the year and 32 percent of the number of children ever enrolled during the year. For FY94 these figures were 40 percent and 29 percent, respectively, while the new enrollees over the two-year period were 75 percent of the initial
TABLE 17 (PDF document.)
enrollees and 43 percent of the children ever enrolled during the period. Clearly, then, Medicaid enrollees are a more stable population than uninsured children, with fewer new children joining and those who do enroll staying longer. Except for FY93, however, the rate of churning appears comparable for Medicaid and uninsurance. While the new spells started by children with earlier spells of Medicaid enrollment during the year were only 14 percent of all new spells in FY93, they were 23 percent of all new spells in FY94--higher than the corresponding figure for the uninsured. Likewise, new spells started over the two-year period FY93 and FY94 by children with earlier spells of enrollment during that same period represented 33 percent of all new spells compared to 35 percent for uninsurance. As with uninsurance, if we were to define churning in FY94 in terms of spells started by children who had other spells of Medicaid enrollment in either FY93 or FY94, the number of such spells would equal the difference between the 4.4 million two-year total and the 980,000 spells started in FY93 by children who had one or more other spells of Medicaid enrollment during FY93. This difference of about 3.4 million represents 51 percent of the new spells started in FY94. That is, 51 percent of the new spells of Medicaid enrollment started in FY94 represented second, third, or even higher order spells by children who had been enrolled in Medicaid once previously in either FY93 or FY94.
Our principal findings with respect to the dynamics of uninsurance, Medicaid eligibility, and Medicaid participation among children may be summarized as follows:
Overall, these findings detail the high rate of movement into and out of uninsurance each month and the high level of interaction between Medicaid and uninsurance. Medicaid plays a vital role in providing health insurance coverage to children who would otherwise be uninsured. Children who leave the status of uninsured are more likely to enroll in Medicaid than the relative frequencies of Medicaid and other sources of coverage would suggest. At the same time, loss of Medicaid coverage accounts for a disproportionate share of the movement of children from insured to uninsured.
As a focus for future research on the dynamics of health insurance coverage it would be useful to know how dynamic patterns vary among different segments of the income and poverty distribution. Investigation of such variation is complicated by the fact that income is itself a dynamic characteristic. Classifying children by their income at a single point in time is a simple approach to employ in analysis and to explain to readers, but it constitutes an oversimplification that is likely to misrepresent the more complex reality that children experience. Research that can address this complex reality would greatly advance our understanding of movement among health insurance statuses.
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., July 1998.
(1) For a spell starting in the sixth month of FY93, the longest duration that we can observe in our data is 19 months.
(2) An obvious strategy would be to screen out spells reported to have started in the first month or ended in the last month of a SIPP reference period and calculate the distribution of spell length among the remainder. With so many spells being reported at four months, however, any assumption that the true distribution of these spells resembles that of the much smaller number that appear to be free of a seam effect would have to be considered tenuous. Nevertheless, the exercise would be informative about the distribution of duration among at least the small subset of spells that is free of this particular source of error.
(3) We define the ending month of a spell of uninsurance as the last month in which a child was observed to be uninsured.
(4) In the next section we examine the health insurance coverage that precedes and follows spells of Medicaid-eligible uninsurance as well as uninsurance generally.
(5) We could correct for this by giving the longer spells a relative weight of 2 in the pooled tabulation. But the longer spells are not a random sample of all such spells that were initiated during the year, so we were reluctant to combine the longer and shorter spells.
(6) In these and most of the other tables in this report, we follow spells past a child’s 19th birthday. That is, a spell is not terminated by a child leaving the population of children under 19. Exits from the SIPP population do terminate spells, however. Children may exit the population by dying, moving out of the country, becoming institutionalized, or enlisting in the military and moving into military barracks. This last reason would appear to account for most of the exits from the survey universe among the entire population of children, but this may not apply equally to uninsured children.
(7) In Tables 8 through 11, unlike the earlier tables, children who turn 19 are considered to have left the population. We treated children turning 19 in this way so that we could reproduce the monthly estimates of uninsured children under 19. For example, the number of children without insurance in October 1992, minus the number who became insured or left the universe between October 1992 and November 1992, plus the number who became uninsured in November 1992, equals the total number of children under 19 who were uninsured in November 1992.
(8) This distribution of the source of coverage among children becoming insured in each month is very similar to what we reported in Table 5 as the distribution of insurance coverage after completion of spells of uninsurance that started in FY93. The spells that contribute to these estimates are somewhat different in the two tables, however. Table 8 includes all spells ending in a given month without regard to their duration whereas the figures in Table 5 are limited to spells that concluded in 1 to 12 months or, in the bottom panel, 13 to 18 months. In terms of the implications for the distribution of spell durations, however, selecting all spells that end in a given month or period is similar to selecting all spells that start in a given month or period. In either case the representation of spells is not affected by their duration. While Table 8 includes some spells that ended after more than 18 months, these are a relatively small proportion of the total--probably no more than 16 percent, if, as we expect, the distribution resembles that reported in Table 1.
(9) If 9 percent of the children uninsured in October 1992 continued to leave uninsured status each month, only 39 percent would remain uninsured after 11 months, compared to the 62 percent that we observe in Table 7. Similarly, if 4 percent of the children enrolled in Medicaid in October 1992 continued to leave Medicaid each month, 66 percent would remain enrolled in Medicaid after 11 months compared to the 80 percent that we observe in Table 7.
(10) In these tables, unlike Tables 8 through 11 but like the earlier tables, children who turn 19 are retained in the estimates. Children described as not in the population are limited to those who left the survey universe and whose insurance coverage after that point is unknown.
(11) Some have suggested, however, that very brief spells between periods of Medicaid enrollment should be counted as continuous enrollment rather than churning.
(12) These observations apply to Medicaid enrollment as well.
(13) SIPP data do not tell us when in a month a child became uninsured or insured. A child whose coverage ended part way through a month is not identified as uninsured until the next month. A child whose coverage started during a month is considered covered in that month, however.
(14) We cannot use the number of children uninsured in the prior month as the initial population because this number includes children who are not counted as ever uninsured during the year.
(15) Table 9 does not report the number of children who became uninsured in October 1992. The estimate of 550,000 new spells started in that month was obtained from the tabulations that underlie Table 1.
(16) The 35 percent estimate for FY93 and FY94 reflects new spells started by children with earlier spells ending two to 23 months earlier or about one year on average. This may provide some indication of the magnitude of churning that we would find with a uniform 12 month reference period.