Child Care Subsidy Duration and Caseload Dynamics: A Multi-State Examination

By Kendall Swenson

Introduction

This report provides an examination of the length of time that low-income families receive government-funded child care subsidies that pay for part or all of the cost of their care arrangements.¹ Statistics of subsidy duration provide a description of the interval of time that families utilize subsidies and document the calendar months when they are more or less likely to enter and exit the programs. These statistics are useful to researchers and policymakers because the patterns may be related to adult employment and child care stability outcomes, and they provide valuable information to program administrators who want to better understand the caseload dynamics of the subsidy programs.

The existing literature has examined the child care subsidy duration patterns of a few states, but these patterns have not been documented for the entire United States. This report partially fills in this knowledge gap by providing statistics of child care subsidy duration for 35 U.S. states using administrative data submitted by state subsidy programs to the U.S. Department of Health and Human Services (HHS). In general, the data presented in this report show that families utilize child care subsidy programs for relatively short time periods in most states, usually less than a year, but frequently return to the subsidy programs after they exit. They are more likely to enter, leave, or return to the subsidy programs during particular times of the year and these usage patterns often coincide with the school year calendar. Since the administrative data used for the analysis does not have information on the employment situation of the families before and after they receive subsidies, and since it lacks information about the non-subsidized care arrangements of the participating families, it is challenging to determine why these patterns occur and to identify whether the breaks in subsidy usage coincide with a discontinuity of the child care arrangements. Despite these limitations, documenting child care subsidy patterns is an important part of understanding how the subsidy programs are currently being utilized and to what extent caseload dynamics differ across various states with dissimilar administrative policies and procedures.

¹ The author would like to thank Karen Aschaffenburg, Kimberly Burgess, Ajay Chaudry, Nina Chien, Elizabeth Davis, Joseph Gagnier, Minh Le, Susan Hauan, Taryn Morrissey, Andrew Williams, and Sharon Wolf for helpful comments on earlier drafts of this report.

Literature Review

Previous research on subsidy durations reveals that low-income families generally receive child care subsidies for relatively short periods of time, although reentry to the programs after an exit is common in many states (Meyers, Peck et al. 2002, Witte and Queralt 2005, Ha and Meyer 2010, Forry, Welti et al. 2012, Ros, Claessens et al. 2012). In their study of five states Meyers, Peck et al. (2002) reported median spell durations of participating children between three and seven consecutive months. They also found that children that reentered the subsidy programs after an exit had subsequent spell durations that were about the same lengths as their previous episodes of participation. These results were somewhat similar to those found in other studies, although comparing duration lengths across the studies is challenging because many of them utilized different methodologies and analyzed data from different time periods or populations (Davis, Grobe et al. 2012). For example, some of the studies measured subsidy duration by following adult heads of households, while others followed the children receiving the care. Some studies defined a break in subsidy participation with one month of non-use, while other studies defined a break in subsidy duration with two consecutive months of non-use. They also used different methodologies to adjust the data when records were unavailable before and after the time-periods examined, and they examined data from dissimilar types of subsidy cases.

The short spell durations and the on-and-off patterns observed in the literature could be influenced by a variety of factors. One possibility is that the usage patterns reflect unstable employment patterns among the children’s parents, who are disproportionately low-income single mothers. Several studies of single mothers have documented employment patterns, including short durations and job cycling, that appear to be similar to the duration patterns observed in the child care subsidy literature (Acs 2001, Andersson, Holzer et al. 2005, Andersson, Freedman et al. 2012). Other studies have examined this issue more directly by matching the employment records of child care subsidy recipients to administrative data from the child care subsidy programs. These studies found that in some cases the timing of child care subsidy exits appears to coincide with employment changes or exits. However, these studies also show that many families appear to remain eligible for the subsidies after they leave the program, suggesting that child care subsidy durations also are likely influenced by factors unrelated to employment (Grobe, Weber et al. 2006, Ha and Meyer 2010). In addition, since maternal employment is often contingent on securing child care subsidies it is possible that some of the employment exits are influenced by problems related to child care arrangements or the inability or unwillingness of the recipients to recertify their eligibility for the subsidy programs. In other words, the primary reasons for subsidy exits are not always known even when accurate employment and subsidy records are concurrently observed by researchers.

The results of several qualitative studies suggest that low-income families have difficulty securing and retaining child care arrangements and they often find that following subsidy administrative policies is burdensome and challenging. Adams, Snyder et al. (2002) conducted a series of interviews with state and local child care administrators and identified a number of administrative practices that appear to reduce the duration of child care subsidy usage. The study found that families often faced considerable administrative burden when trying to apply for or recertify their eligibility status. For example, families sometimes had to interact with more than one agency during the application process, had to make more than one trip to an administrative office, and sometimes had to wait for weeks or months to get an appointment with a social worker. In addition, families receiving Temporary Assistance for Needy Families (TANF) sometimes had additional difficulties with redetermination because of the temporary nature of their employment or training activities. The study also found that agencies had different policies regarding the ways in which families could recertify their eligibility status including mail, phone, or fax.

Other studies emphasize the complexity of understanding child care subsidy duration patterns. For example, a related study by Chaudry (2004) followed 42 low-income families in New York City over a three year period and described the challenges they encountered securing the care needs of their children. The stories retold by Chaudry are relevant to the subsidy duration literature because they show that families encounter challenges in securing child care for a variety of reasons. While many of the families experienced complications associated with dealing with local social service organizations to obtain subsidies, their child care arrangements also were sometimes interrupted by problems associated with their ever-changing work schedules and with the personal relationships they had with their providers.

In summary, the literature on child care subsidy duration suggests that subsidies are often used for relatively short periods of time and that reentry to the programs after an exit is common. While some of the exits are likely related to the volatile employment patterns of the parents that participate in the programs, many families that leave the program appear to be still employed and income-eligible for the subsidies.

Data

The data analyzed for this report are from the ACF-801 child care subsidy administrative records.² The ACF-801 data consist of monthly records submitted by state child care programs to HHS, and this analysis uses data from federal Fiscal Years 2004 through 2010 that are linked longitudinally by matching the Social Security Numbers (SSNs) of the family heads. The ACF-801 data include all families that received subsidies from the Child Care and Development Fund (CCDF) including those funded through the Child Care Development Block Grant (CCDBG), those funded with transfers from the Temporary Assistance for Needy Families (TANF) program, and those funded with state matching and maintenance of effort (MOE) funds related to the CCDBG. States also have the option of including families receiving subsidies from other funding sources such as the Social Services Block Grant (SSBG), direct TANF³ funds, or state-funded sources, but not all states include these records in the ACF-801 data. Some states pool several child care subsidy funding sources and operate a single program, while other states operate separate child care subsidy programs. For example, some states administer separate subsidy programs for families receiving assistance or job training from the TANF program and states may or may not submit these records along with their ACF-801 data submissions.⁴

Some states report state-created unique case identifiers of the family heads instead of Social Security Numbers (SSN). A decision was made for this analysis to exclude the records that do not have SSNs for the heads of households since preliminary analysis of the data revealed that many of the non-SSN unique numbers appeared to be reported inconsistently across the months (or fiscal years) and their inclusion could bias the analysis. All SSNs were scrambled to protect the identity of the recipients.⁵

The 16 states that were excluded from the analysis are displayed in Table 1 by reason of omission. States that submitted samples instead of their full subsidy caseloads to HHS for some or all of the years examined were left  out because longitudinal analysis is not possible with incomplete data.⁶ The states of Connecticut and Florida submitted their full populations to HHS, but they were excluded because they did not submit the SSNs for the heads of household for some or all of the months in the period of time analyzed for this report.⁷ One complex issue encountered during the analysis was the presence of multiple records in the data with the same SSNs during the same months. Some of these cases appear to reflect transitions of families from one office or administrative process to another. For example, in some cases the multiple records appear to reflect redeterminations or transitions in and out of the Temporary Assistance for Needy Families (TANF) program or changes in geographic locations. However, it appears that many of the multiple records are a result of misreporting and a decision was made to delete all families containing a duplicate SSN for any of the months in the analysis. The data submitted by Oregon, Mississippi, and Arkansas consisted of large numbers of duplicate SSNs and these states were omitted. After these exclusions, 35 states remained for analysis. Another challenge with examining data from states with different types of caseloads is that the states varied in the percentage of their caseloads that received care because they were in protective services. Since these children are likely to have different characteristics than other children they were excluded from the analysis.

Table 1. States Excluded from the Analysis by Reason

² A sampled version of the ACF-801 data is available to the public on the Research Connections Web site http://www.researchconnections.org. For preservation of confidentiality, the public-use version does not include the unique identifiers needed to construct measures of child care subsidy duration.

³ States have the option of transferring TANF funds to the Child Care Development Fund (CCDF) or spending the funds directly on child care subsidies. Funds spent directly on child care subsidies are sometimes called TANF-Direct funds and states have the option of including these families in their ACF-801 data submissions, but not all states do.

⁴ The majority of statistics released by the U.S. Department of Health and Human Services adjust the caseload counts using a “pooling factor” to estimate the number of families funded by CCDF. These adjusted numbers are derived by multiplying the unadjusted caseloads reported on the ACF-801 data by the pooling factor. Since the data are not able to identify which families were funded by CCDF and which were funded by other funding sources, the data in this report are not adjusted with a pooling factor or re-weighted to align with any published caseload targets. Excluding the pooling factor does not alter the caseload characteristics or spell duration patterns at the state level, but does increase the size of the caseload reported.

⁵ One limitation of analyzing pseudo-SSNs is that the analysis was unable to exclude records that had invalid SSNs. Previous analysis has found that a small percentage of SSNs on the ACF-801 files were not valid, but the frequency of invalid SSNs was small and their inclusion in this analysis was not expected to significantly bias the results presented.

⁶ The states of Alaska, Minnesota, and Pennsylvania currently submit full populations to HHS but were excluded because they did not submit full samples for all of the years included for this analysis.

⁷ States are not required to collect Social Security Numbers and Lead Agencies need to make it clear to applicants that providing SSNs is optional. Despite these options, most states receive SSNs for the majority of their participants.

Limitations of the Data

Producing a series of subsidy duration statistics with the ACF-801 data to be used for cross-state comparisons is desirable, though challenging for a variety of reasons. One limitation of the ACF-801 data, as previously discussed, is that some states include families receiving subsidies funded from SSBG, TANF-Direct, or state-funded programs in their caseload submissions, while other states do not. This issue is particularly salient because families receiving TANF have shorter spell lengths, on average, than other families and states that exclude some or all of their TANF families from their CCDF caseloads may artificially appear to have longer spell durations than states that include their TANF records in the ACF-801 data. These changes are especially challenging when states alter over time the populations they include in their ACF-801 data submissions. Thus, care should be exercised when making conclusions with the results. States also modify their policies and budgets across time, which may produce findings that are difficult to interpret without having intimate knowledge of the individual state programs. To mitigate these challenges this report primarily describes trends and patterns that appear consistent across many states, and less attention is placed on states with outlying characteristics. Hopefully, future researchers can build on the findings in this report with more in-depth examinations of the individual states with outlying characteristics because these states may have implemented innovative administrative practices that offer informative lessons for policymakers in other states.

A second limitation of the ACF-801 data is that it does not collect information about the circumstances of the families when they begin or exit the subsidy programs. One reason for examining child care subsidy duration patterns is that they are likely related to the usage patterns of child care arrangements. Many low-income families are unable to afford the full costs of care by themselves and the discontinuity of subsidies often corresponds with changes in child care arrangements. However, changes in care arrangements do not always occur when subsidies are initiated or discontinued⁸ and the ACF-801 data provide no information about why these transitions transpired. Therefore, the link between changes in subsidy receipt and continuity of care arrangements cannot be directly made with the data.

⁸ For example, in the absence of subsidies some families may be able to pay the full cost of care to the providers themselves, or they may be able to convince their providers to lower their payment obligations.

Definition of Child Care Subsidy Duration

This report defines duration periods, also called spells, as the number of continuous months that families⁹ receive child care subsidies, preceded and followed by a month of non-receipt.¹⁰ All months of subsidy receipt are included in the spells regardless of the number of hours of participation. Spells began the months when families started using subsidies for either the first time or after at least a one-month break of non-use. Spells are considered completed when there is at least one month that the families do not receive subsidies. In some places in this report the definition is altered slightly to require that new spells are preceded and followed by at least two months of non-receipt. The durations presented in this paper are statistical spells, which may differ from the methodologies that individual state administrators use to define program entries, exits, and lengths of participation. In some cases participants return to the subsidy system after they exit. In order to contrast families that have and have not experienced recent spells of participation the report restricts the analysis in some places to families that are either new to the subsidy system or are returning after a 36-month absence from the programs. In order to present statistics that measure subsidy duration across several individual spells of participation, the report presents in some places the total number of months that families received child care subsidies over a three-year period, regardless of whether the months of receipt were continuous.

⁹ This study used families as units of analysis, which differs from some other studies that used children as units of analysis.

¹⁰ Using months as units of measurement exaggerates somewhat the amount of time that some families receive child care subsidies if they begin subsidy participation during the middle or end of a month, or if they end subsidy use at the beginning or middle of a month. For example, if a family began receiving subsidies on August 15^th and ended subsidy use on November 14^th, the duration would be measured as four months even though the duration was 13 weeks, which is closer to three months after dividing the 13 weeks by 4.3 weeks/month.

Spell Beginnings

Figure 1 displays the number of families from all 35 states that began subsidy use either for their first time, or after at least a one-month break in usage. Some families are counted multiple times if they had a break in subsidy use and returned. As shown in the figure, the data reveal that families are more likely to begin new episodes of child care subsidy use during particular months of the year than in other months and these patterns appear to be related to the school year calendar. For example, new spell beginnings are at their highest frequency during the months of August, September, and October. December appears to be a month when spell beginnings are less frequent. These patterns vary somewhat across states, although the majority of states display at least some seasonal patterns, often showing peak months in either August or September. For example, Figure 2 presents the spell beginnings of Wisconsin and Kansas, which have higher than normal entries during the late summer months.¹¹

Families frequently return to the subsidy programs after periods of non-use. This issue is explored in Table 2. Column two of the table presents the percentage of families that were either entering the subsidy programs for the first time, or were reentering after at least 36 months of non-receipt. As shown in the table, a large percentage of the entering families in all 35 states had previously received subsidies. These percentages ranged from a low of 41 percent in the District of Columbia to a high of 66 percent in New Jersey.¹² When the families from the 35 states were analyzed together, about 43 percent of them entered the programs without any recent subsidy receipt and about 57 percent of them were returning within 36 months of a previous period of participation.

A second pattern that is apparent in Table 2 is the large percentage of families that began new subsidy spells after only a single month of non-receipt following a previous period of participation. For example, of the families that began new spells in the 35 states (see column 3), about 17 percent of them were returning to the programs after one month of non-receipt. In some cases, these spells may be administrative reporting errors that reflect occasional omitting or misreporting of family records. In other cases, however, the patterns likely reflect families that did not receive subsidies for a variety of reasons.

Figure 1. Number of Families in the 35 States Beginning New Spells of Subsidy Receipt by Month of Entry

Figure 2. Number of Families in Wisconsin and Kansas Beginning New Spells of Subsidy Receipt by Month of Entry

Table 2. Percentage of Families Beginning New Spells of Child Care Subsidy Use that Had Previously Received Subsidies by State: FY 2007 to FY 2010

¹¹ The data for the remaining 33 states can be found in Appendix Table 1.

¹² The percentages in this sentence were calculated by subtracting the values in column two from 100 percent.

Spell Endings

Figure 3 displays the months that families exited the subsidy programs for Federal Fiscal Years 2007 to 2010.¹³ Like spell beginnings, the timing of spell endings is more prevalent during certain months of the calendar year than in others. The caseloads in the table represent the number of families that received child care subsidies before experiencing at least a one-month break in subsidy participation. As shown in the figure, subsidy exits are more frequent in May and August than in other months.

Figure 3. Number of Families in the 35 States that Ended Subsidy Spells by Month of Exit (FY 2007 to FY 2010)

¹³  This figure presents data from a somewhat different time period from that shown in Figure 1. It excludes a very small number of families that began subsidies before FY 2004. Caseloads for the month of September 2010 are not included.

Length of Spell Duration

The amount of time that families receive child care subsidies can be presented in different ways. One possibility is with a Kaplan-Meier survival curve, like that shown in Figure 4. The x-axis in the figure represents the number of consecutive months that families received child care subsidies and the y-axis signifies the percentage of the caseload that “survived” to the corresponding month in the x-axis.¹⁴ For example, slightly less than half (47 percent) of families received subsidies for six or more months, and about 25 percent participated for at least 12 months before experiencing at least one month without receipt.¹⁵ Uninterrupted subsidy participation for at least two years was somewhat uncommon; these spells represented about 11 percent of the families that began subsidy use in FY 2007 (i.e., between October 2006 and September 2007). About six percent of families received child care subsidies for at least 36 consecutive months.

Figure 4. Kaplan-Meier Survival Curve: The Percentage of Families in the 35 States that Received Child Care Subsidies to Particular Lengths of Time without Experiencing at Least One-Month of Non-Receipt (FY 2007)

Another way to analyze length of subsidy receipt is to examine the percentage of families that experienced spell durations of various lengths. This is shown in Figure 5 for families that began spells of subsidy use in Fiscal Year 2007 for either their first time or after at least 36 months of non-receipt. When the families from the 35 states are analyzed together, about 21 percent of them completed spells in one or two months, and another 25 percent completed spells in three to five months. About 29 percent of families had spell lengths of 12 months or greater.

One of the most common methods of analyzing subsidy durations is with medians. Figure 6 presents the median spell lengths of families in the 35 states that began new spells of participation for either their first time or after at least one month on non-receipt. The time period is restricted to families beginning new spells between January 2004 and August 2009. The data from the 35 states presented in the figure reveal that median spell lengths have seasonal patterns that generally fluctuate between four and seven months.¹⁶ Families that began subsidy usage in the months of August to January had higher medians spell lengths than families beginning program use during other months of the year. In general, median spell durations were particularly low when they began in June compared to the spell durations that began in the late summer and early fall months.

Figure 5. Number of Continuous Months that Families Received Child Care Subsidies before Leaving the Subsidy Program or Experiencing at Least One Month of Non-Receipt (Percentages of All Families Entering the Programs in FY 2007)