Administrative Burdens and Economic Insecurity Among Black, Latino, and White Families

BACKGROUND

A robust body of research shows that income support programs have beneficial short- and long-term effects on low-income families, especially those with children. This study focuses on benefits receipt among households encountering financial difficulties, with or without minors. However, households that include children are more likely to be low income than those without (Shrider et al. 2021) and to be eligible beneficiaries of income transfer programs. Indeed, to qualify for TANF, households must have at least one child under the age of eighteen present (U.S. Department of Health and Human Services 2021), and two-thirds of SNAP benefits go to families with children (Center on Budget and Policy Priorities 2017). Recipients of UI, in contrast, are somewhat less likely to live in a household with a child: 44 percent of UI recipients coreside with an individual under the age of eighteen (Carey et al. 2021). Because most families receiving benefits include children, much of the research examining the effects of these programs focus on such households.

Income support programs are influential in reducing poverty in the short run and often result in improvements in both health and economic productivity in the long run (Hoynes and Schanzenbach 2018). In 2015, SNAP benefits alone lifted nearly four million children out of poverty (Hoynes and Schanzenbach 2018; Wheaton and Tran 2018), and the collective value of income transfers and tax credits removed 7.4 million in 2017 (Hoynes and Schanzenbach 2018; Shapiro and Trisi 2017). Additionally, SNAP access during childhood is related to better health outcomes during adulthood (Hoynes, Schanzenbach, and Almond 2016).

The impact of TANF on family well-being is relatively modest, in part because federal funding has been held at its original 1996 level (Floyd, Pavetti, and Schott 2017). However, research suggests that sizable increases in total income from higher TANF ($1,000) payments are associated with significant increases in student achievement (Duncan, Morris, and Rodrigues 2011; Hoynes and Schanzenbach 2018) and TANF coverage may improve a family’s daily routine and decrease the likelihood that a child repeats a grade (Wang 2015). During the Great Recession, about 40 percent of households that included a person receiving UI benefits would have been considered poor before accounting for this income transfer (Gabe and Whittaker 2012). During the COVID-19 pandemic, UI benefits played a pivotal role in reducing hardship among the unemployed and stimulating economic activity (Falk et al. 2021).

Given the importance of income support programs for family and child well-being, racial and ethnic differences in benefit receipt may perpetuate inequalities in economic opportunity. Studies show that unequal access to resources, particularly among families with children, contributes to disparities in health, education, material hardship, and subjective well-being (Duncan, Morris, and Rodrigues 2011; National Academies of Sciences, Engineering, and Medicine 2019; O’Brien et al. 2020; Cross 2020). Black and Latino families in the United States have long faced unequal access to resources relative to White families. Although labor-market inequalities influence many of these disparities in income, the decentralization of the American welfare state also tends to perpetuate racial-ethnic inequalities (Michener 2018; Soss et al. 2008; Hardy and Logan 2020; Herd and Moynihan 2018). Black families are more likely to live in states that have not expanded Medicaid, have lower minimum wages, do not offer supplements to the federal EITC, spend less on TANF cash assistance, and offer less-generous UI benefits (Michener 2018).

ADMINISTRATIVE BURDENS IN THE AMERICAN WELFARE STATE

Administrative burdens are commonly classified into three broad categories: learning costs, psychological costs, and compliance costs (Moynihan, Herd, and Harvey 2015). A learning cost refers to the responsibility imposed on citizens to learn about a program, ascertain whether they are eligible, and understand the nature of benefits and how to access services. Psychological costs are incurred when citizens have to deal with the stigma associated with participating in an unpopular program; they may also involve a loss of autonomy and increased stress as a result of navigating program procedures. Compliance costs include the completion of applications for enrollment or reenrollment and responding to or avoiding discretionary bureaucratic requests.

Typically, researchers focus on specific instances of administrative burdens: paperwork, extra hurdles for applying to benefits, excessive bureaucratic discretion, and so on. In this study, we adopt a broader conceptualization of administrative burdens: all factors that contribute to the lack of benefit receipt among individuals who are eligible for the given benefit. We are, in effect, capturing the aggregate consequences of a series of first-order burdens that affect program participation among the eligible. This broader conceptualization deviates from typical administrative burden analyses that focus on a single burden or a single program; instead, it adopts as a counterfactual that all families eligible to receive a given benefit should receive the benefit. This counterfactual is not without precedent. Data from European countries find near-universal coverage of unemployment benefits in Germany, Finland, Austria, and Belgium (Thévenot, Maestri, and Maquet 2016). The UK Universal Credit automatically updates recipients’ benefit values when their earnings change (Coady et al. 2021); in Estonia, thirteen national data registries connect to efficiently identify eligible recipients for unemployment benefits and to distribute benefits (e-Estonia 2018; Helmes 2020). We acknowledge that many factors drive the lack of take-up of benefits—from stigma to low information about a given benefit to burdensome application processes. In viewing the counterfactual as a near-automatic distribution of benefits, however, we can conceptualize each of these broadly as administrative burdens and can investigate how such burdens shape access to resources throughout the past thirty years.²

Notably, our full-coverage counterfactual does not attribute residual lack of coverage—after accounting for specific administrative burdens—to behavioral differences overall or between subgroups. Historically, the United States has prevented some subgroups from gaining equal access to income support programs through a series of policies and practices that give rise to racially discriminatory outcomes. For example, TANF’s predecessor, Aid to Families with Dependent Children (AFDC), restricted benefit receipt for African Americans to certain times of the year to guarantee the supply of Black labor for agriculture and domestic work (Floyd et al. 2021). Thus potential behavioral differences, such as whether to apply for benefits, can be conditioned by program and policy design. Adopting a broader conceptualization of administrative burdens addresses this potential endogeneity issue and accounts for racist policies and practices within the welfare state that may obstruct program participation. It also allows us to benchmark trends across time and place and offers a tool that can be carried forward in future research to continue monitoring progress using publicly available data.

WHY FOCUS ON JOB LOSS?

A primary goal of income support programs is to protect families from negative economic shocks (Figinski 2017). In this analysis, we examine receipt of income support from SNAP, TANF, or UI after a person in a low-income family experiences job loss.³ Job loss is relatively common: in a typical (that is, nonrecession) year, between 1 and 2 percent of workers are laid off, low-income and less-educated workers at even higher rates.⁴ Past analyses reveal how job loss drives an immediate decline in individual earnings and overall household income (Couch and Placzek 2010; East and Simon 2020). For example, Chloe East and David Simon (2020) find that no-fault job loss reduces individual earnings by between 49 and 66 percent over the first six months; after one year, monthly earnings are still 38 percent lower than the month before the layoff event, and negative effects on earnings are traceable more than five years beyond the event (see Couch, Jolly, and Placzek 2011).

Focusing on job loss in this analysis is a particularly useful scope condition for understanding administrative burdens: as an income shock, it marks a critical moment when low-income families will generally turn to income transfer programs for support. Relative to other events that may trigger eligibility for income support programs—namely, family transitions such as childbirth, divorce, family member death—job loss is more common and arguably more straightforward to measure in national surveys.

Research has used job loss as a trigger event to examine the dynamics of the U.S. social safety net. For example, East and Simon (2020) use data from the Survey of Income and Program Participation (SIPP) to examine receipt of major public assistance programs following an involuntary layoff. Consistent with the literature (Rothstein and Valletta 2017), they find that UI is the most substantial source of income support for individuals who lose their jobs, followed by SNAP benefits; they find other income support programs, notably TANF, provide less help (see also Bitler, Hoynes, and Kuka 2017). Whereas East and Simon (2020) primarily focus on earnings replacement after job loss across the income distribution, this study focuses on the consequences of uneven access to income transfers among families that are likely eligible to receive the benefits, focusing on racial-ethnic and over-time differences.⁵

DATA AND METHODS

We use the panel component of the CPS ASEC from 1990 to 2019.⁶ Households in the CPS ASEC sampling frame are sampled for four consecutive months, then are not interviewed for the next eight months, and then are then included in the CPS again for the next four months. As a result, roughly half the respondents in one year’s CPS ASEC are also featured in the subsequent year’s CPS ASEC. The panel component of the ASEC is not often used (for recent exceptions, see Hardy, Smeeding, and Ziliak 2018; Lundberg 2021) but provides a powerful source of data to measure year-on-year transitions in employment, poverty, income, benefit receipt, and more. Moreover, it features larger sample sizes and more comprehensive income data relative to other sources of the panel data, such as the SIPP or PSID.

We first limit our sample to respondents who are observed in consecutive years within the ASEC. Although we use data from both years in which each respondent is observed, we evaluate all outcomes in the second of the two years. This limits our sample to some forty thousand individuals per year who are observed in the ASEC for their second year. Following common practice in the inequality and poverty literature, we focus on outcomes at the family unit level (more specifically, the Supplemental Poverty Measure unit, which is equivalent to the household in more than 95 percent of cases). We refer to the family unit as family regardless of whether it includes children.

Our analytical focus is on families with a working-age adult that transition from employment in year one to nonemployment in year two. Our definition of nonemployment includes working-age adults who are jobless and actively looking for work (unemployed) but also those who are jobless and do not report to be looking for work (nonemployed).⁷ We do not include as nonemployed any working-adult adults who are retired or in education. Because we are interested in lower-income families (those more likely to seek government assistance in the event of job loss), we limit the sample to families with a pretax, pretransfer income that is below half the national median in the second year in which they are observed. The rationale is that wealthier families experiencing job loss are less likely to face economic insecurity relative to lower-income families that experience job loss. In figures A.2 through A.5, we also present results for the full sample without cutting based on pretax-transfer incomes. We focus on pretax-transfer incomes in the second year (after job loss), rather than the first year, in our primary analysis given that the pretax-transfer incomes after job loss are more indicative of the family’s potential need for income transfers. Focusing on pre-job-loss income may capture some families who are better off in the year after job loss if another family member increases work intensity or receives higher wages, and also leads to the possibility of missing families who experience the largest losses in income and might be eligible for higher levels of benefit receipt. We present results that cut the sample based on incomes in the year before job loss in figure A.4.

As to outcomes of interest, among the low-income families with a working-age adult who experience job loss, we are primarily interested in two sets of outcomes: the lack of receipt of transfer benefits among those who appear to be eligible to receive the transfers, and the replacement rate of transfers. We discuss these in turn.

First, we measure our primary proxy of administrative burdens: nonreceipt of SNAP, TANF, and UI transfers among families eligible to receive the benefits in the year in which a family member becomes jobless. These are binary indicators of whether at least one member of the family receives a positive value of the given benefit in the second year (regardless of whether the family also received the benefit in the year before job loss). As we document in figure A.1, most families who receive SNAP, TANF, or UI benefits in the year of job loss did not receive the given benefit in the year before job loss.⁸

We measure eligibility for SNAP and TANF benefits using indicators from Urban Institute’s TRIM3 simulations, which takes into account immigration status, family structure and size, state policy rules, and more when identifying likely eligibility.⁹ TRIM3 does not provide an estimate of eligibility for UI benefits. Thus, we create a proxy of UI eligibility manually in the ASEC based on whether the newly jobless individual worked at least twenty-six weeks in the prior year; worked at least ten hours per week, on average, in the prior year; and reports being unemployed (jobless and actively searching for work).

Our second indicator captures the consequences of administrative burdens. We measure the replacement rate of income transfers as the share of lost market earnings that are compensated by increases in transfer benefits for a family experiencing job loss. For example, if a family loses $20,000 in annual earnings due to lost wages after job loss, but subsequently receives $5,000 in additional social assistance (beyond what the family received in the year prior to job loss), the replacement rate is 25 percent. We bottom-code replacement rates at zero (no negative values) and top-code at 100 percent (perfect replacement). Of course, the U.S. legal maximum replacement rate is typically below perfect replacement. For example, state laws usually cap the UI maximum benefit amount between 70 and 80 percent of prior income, just as SNAP recipients are expected to spend 30 percent of their resources on food, limiting the benefit amount to no more than 70 percent of their income (D.C. Department of Human Services 2022; Lacoma 2022). Under these restrictions, replacement rates between 70 and 80 percent likely represent a best-case scenario.

To assess the broader contribution of administrative burdens to economic insecurity, we also use the indicators of program eligibility and benefit receipt to produce a counterfactual replacement rate if all eligible individuals received the benefits they are eligible to receive. To calculate this, we first compute the total sum of transfers that a family unit would receive if it indeed received all SNAP, UI, and TANF benefits for which it is eligible. We then substitute this value for the observed value of benefit receipt. We then construct the counterfactual replacement rate using these substituted benefit values.¹⁰

We supplement our primary results with three sensitivity tests. First, we address the benefit underreporting of TANF and SNAP in the CPS ASEC. Benefit underreporting will likely lead us to underestimate replacement rates. If benefit underreporting has intensified from the 1990s onward, or varies meaningfully by race-ethnicity, it may alter our conclusions about trends and disparities in benefit access after job loss. To address this issue, we present a set of results in which we adjust for underreporting in TANF and SNAP using Urban Institute’s TRIM3 simulations.¹¹ In short, TRIM3 uses information about each individual and household in the CPS ASEC to predict the value of benefits that a recipient is likely eligible to receive. The simulations use individual or household data on race, ethnicity, immigration status, marital status, household structure, state of residence, income, state-level policy rules, and more to estimate program eligibility. One limitation of TRIM3 is that the data are currently available only from 1993 through 2017; thus we limit many of our analyses to these years. A second limitation is that TRIM3’s benefit simulations may allocate SNAP and TANF benefits too strongly (relative to administrative records) toward households at the bottom of the income distribution (Stevens, Fox, and Heggeness 2018). As a result, the use of TRIM3 could provide an overoptimistic scenario of TANF and SNAP benefit receipt among our focal populations. We therefore present results both with and without TRIM3 benefit adjustments and evaluate our conclusions in either scenario.

Second, it could be that different population characteristics across states, rather than administrative burdens, drive incomplete access to benefits. To test this possibility, we also estimate our counterfactual replacement rates after reweighting each state-year’s population to have the same population characteristics as the United States as a whole. Specifically, we equalize the mean levels of high school graduation, college graduation, family structure, employment rates, gender, household size, and number of children of each state to match the mean values among all states (see figures A.2 through A.5).

Third, it could be that variation in states’ benefit levels, rather than differential access, could drive any differences we observe in counterfactual replacement rates. Given that SNAP benefits do not vary across states (other than Alaska and Hawaii), and that UI benefits largely vary depending on an individual’s prior earnings, it is primarily policy-driven variation in TANF benefit levels that threaten our analysis. Thus, we also produce estimates of counterfactual replacement rates after equalizing all states’ TANF benefit levels (see figures A.2 through A.5).

FINDINGS

Figure 1 presents the share of all family units in the United States with at least one member experiencing a transition from employment to nonemployment by race-ethnicity and year. In 1990, just under 10 percent of all families had one member transition from being employed in 1989 to not employed in 1990. For Whites, the rate was slightly lower than the national mean (around 8 percent); for Blacks and Latinos, it was higher (around 12 percent and 14.5 percent, respectively).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Family Units with at Least One Employment to Nonemployment Transition from Previous Year

Source: Authors’ calculations from the U.S. Current Population Survey (Flood et al. 2022).

Note: Error bars represent 95 percent confidence intervals.

In the following two decades, the share of families experiencing job loss steadily declined until the onset of the Great Recession, then spiked around 2008 and 2009 before steadily declining again. In 2010, the share of families experiencing job loss from the prior year was just above 5 percent. Across all years, however, racial-ethnic differences persist. Blacks and Latinos consistently face notably higher rates of family job loss relative to Whites. By 2019, in fact, the share of Blacks or Latinos experiencing family job loss was around double the rate of that of Whites, around 10 percent to 5 percent.

ADMINISTRATIVE BURDEN EFFECT

Figure 2 provides descriptive trends in the nonparticipation of SNAP, UI, and AFDC-TANF benefit receipt among low-income families who experience joblessness and are likely eligible to receive the given benefit. The upper left panel shows that SNAP benefits, the only of these transfers primarily administered at the federal level, have low and declining rates of nonparticipation among likely eligible families experiencing job loss. In 1990, an estimated 45 percent of likely eligible SNAP families experiencing job loss did not receive the benefit, though this figure fell to around 20 percent in 2019. In many years, Blacks see lower than average nonparticipation rates of SNAP benefits among the eligible, whereas Latinos often see higher than average. This may reflect the “chilling effects” of immigration enforcement that disproportionately affect Latino families (Friedman and Venkataramani 2021) That is, many Latino immigrants who are eligible to receive benefits may forgo them to avoid real or perceived negative immigration consequences for themselves or those close to them who may be undocumented (Friedman and Venkataramani 2021; Watson 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Administrative Burden Effect

Source: Authors’ calculations from the US. Current Population Survey (Flood et al. 2022).

Note: Rates of benefit nonparticipation among low-income families who experience joblessness and are likely eligible to receive the given benefit.

The upper right panel of figure 2, in contrast, suggests higher nonparticipation rates among the eligible for UI benefits. In 1990, UI missed an estimated 40 percent of likely eligible families experiencing job loss. By 2019, this had increased to about 60 percent. Across most years, Black and Latino families experiencing job loss are less likely to receive UI benefits despite likely being eligible (see also Kuka and Stuart 2021).

The lower left panel shows steep declines in access for AFDC-TANF benefit receipt. In 1990, cash assistance from AFDC missed an estimated 20 percent of likely eligible families experiencing job loss, outperforming SNAP and UI. In 2019, however, nonparticipation climbed to 80 percent of likely eligible families experiencing job loss. Differences across race-ethnicity are not generally statistically significant in most years, though White families have lower rates of take-up in more recent years.

Put simply, administrative barriers and other factors that limit access to benefit receipt among the eligible are leaving considerable income support undelivered for families experiencing job loss. This is likely to affect the economic insecurity of these families. To contextualize the consequences, we visualize counterfactual income replacement rates if all families eligible for the benefits actually receive the benefits.

The thick solid line in each panel of figure 3 shows the estimated increase in replacement rates if families were to access the benefits they are eligible to receive. The dashed line presents the same, but after adding in TRIM3 benefit adjustments. The upper left panel, for example, documents that the average low-income family experiencing job loss could have had a 13 percentage point higher replacement rate—or 11 percentage points with TRIM3 adjustments—in 1993 if they received all the transfers for which they were eligible. In 2019, however, the rate declined to 10 percentage points, or 6 percentage points with TRIM3. Thus, families are receiving a larger share of benefits for which they are eligible in more recent years compared to the early 1990s. Subsequent investigation confirms that this trend is driven primarily by rising SNAP participation among eligible families (see figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Percentage Point Increase in Replacement Rates

Source: Authors’ calculations from the US. Current Population Survey (Flood et al. 2022).

Note: No administrative burden effect. Sample limited to households in pre-tax/transfer SPM poverty after experiencing job loss. TRIM3 adjustments applied for SNAP and TANF benefits.

Black families (upper right) and Latino families (lower left) experienced larger potential increases in replacement rates than White families if they had received all benefits after job loss, although racial differences mostly even out after TRIM3 adjustments. Recall that TRIM3 over-allocates some income transfers toward the bottom of the income distribution, so the true replacement rate may be somewhere between the two lines in figure 3. Figures A.2 through A.5 present results when not cutting the samples based on income, or when cutting the sample based on income in the year before job loss. These alternative results offer similar conclusions: the potential increase in replacement rates if families facing job loss received all SNAP, TANF, and UI benefits for which they are eligible has declined over time, and is generally larger for Black and Hispanic families relative to White families, though TRIM3 adjustments continue to narrow or eliminate those racial differences.

In figures A.2 through A.5, we provide evidence that variation across time and race-ethnicity in potential increase in replacement rates is primarily driven by benefit access, rather than differences in TANF benefit levels across states or differences in compositional features of states’ populations.

CHARACTERISTICS OF STATES WITH LARGE ADMINISTRATIVE BURDEN EFFECTS

Our prior results identify administrative burden effects across time by the race-ethnicity of individuals. As a final step, we investigate how the racial-ethnic composition of place is associated with administrative burden effects. Specifically, we identify demographic characteristics of states with smaller and larger administrative burden effects, measured here as the mean percentage increase in the replacement rate in a state in the absence of the administrative burden effect (similar to figure 3) averaged over 1993 through 2017. Our assessments of the state-level correlates of administrative burden effects are purely descriptive associations; the methods we use are not designed to infer causality.

Figure 4 plots the mean administrative burden effect by state, depending on the modeling decisions made: our baseline estimates with no adjustments, with TRIM3 benefit adjustments, with homogenous state-level demographic characteristics, and with equal state TANF benefit levels. Higher values (darker colors) represent stronger administrative burden effects or, more precisely, higher percentage point increases in the replacement rate of families experiencing job loss if they were able to access all benefits for which they are likely eligible.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Mean Change in Replacement Rate If Families Received All Income Transfers They Are Eligible to Receive, 1993–2017

Source: Authors’ calculations from the US. Current Population Survey (Flood et al. 2022). Allow space for key in pagination

Note: Darker colors indicate a stronger administrative burden effect (defined as percent replacement rate if families received all transfers for which they were eligible minus the observed replacement rate). Homogenous demographics refers to results after reweighting state populations to match the national means of adults with only a high school degree, adults with a college degree or more, household structure (single parent with children, single adult without children, two-parents with children, multiple adults without children), number of children in the home, number of adults in the home, female, and employment rates. Equal Benefit Levels refers to an adjustment that alters AFDC-TANF benefit levels in each state-year to match the national median TANF benefit levels among recipients in the given year, effectively eliminating state-variation in TANF generosity (and thus limiting variation primarily to differential access).

In percentage point terms, the largest administrative burden effects are concentrated in the Southwest, the smallest in the upper Midwest. Arizona, for example, consistently performs the worst of all states: in our base estimates, its average increase in replacement rates would be 19 percentage points if its families received all income transfers they are eligible for after job loss. Arizona was also the first state to reduce its lifetime limit of cash assistance from TANF to twelve months, down from the national maximum of sixty (Parolin 2021). On the opposite end, Utah is the best performing state: the average increase in replacement rates is 8 percentage points.

These geographic differences are likely to overlap with important demographic differences in the administrative burden effect. Consider, for example, that poor-performing states such as Arizona and Texas tend to have larger shares of Latino residents than other states. Even if differences in administrative burden effects do not vary widely across individuals of different races and ethnicities (see figure 3), we may see more notable disparities across places with higher proportions of Black or Latino residents.

Figure 5 investigates this possibility. Specifically, it displays the bivariate association of a state’s demographic characteristics—race-ethnicity and share of children in single-parent homes—with the mean administrative burden effect—the percentage point gain in replacement rates with full transfer receipt—in the state from 1993 through 2017. We again test four variations of the estimates, aligning with the four panels of figure 4. These bivariate associations provide us a descriptive view of the characteristics of states with higher or lower percentage point increases in the counterfactual replacement rate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Bivariate Association of Demographic Characteristic, 1993–2017

Source: Authors’ tabulation (Flood et al. 2022).

Note: Mean administrative burden effect, counterfactual increases in replacement rates. The figure plots point estimates and 95 percent confidence intervals from separate bivariate estimates that regress the mean percentage point increase in a state’s replacement rate should family units receive the transfers for which they are eligible between 1993 to 2017 (administrative burden effect) on the labeled state-level covariate. Homogenous demographics refers to results after reweighting state populations to match the national means of adults with only a high school degree, adults with a college degree or more, household structure (single parent with children, single adult without children, two parents with children, multiple adults without children), number of children in the home, number of adults in the home, female, and employment rates. Equal TANF Benefit Levels refers to an adjustment that alters AFDC-TANF benefit levels in each state-year to match the national median TANF benefit levels among recipients in the given year, effectively eliminating state-variation in TANF generosity (and thus limiting variation primarily to differential access).

With respect to race-ethnicity, the findings in figure 5 show that the larger the share of White state residents, the lower the administrative burden effect. In contrast, states with a larger share of Latino residents, in particular, tend to face larger administrative burden effects. States with larger shares of Asian residents also experience larger administrative burden effects, or higher potential increases in replacement rates if families received all income transfers after job loss. The coefficients for the share of Black parents and single parents are generally insignificant.

The findings are consistent whether we apply TRIM3 adjustments for benefit underreporting, adjust to match the demographic characteristics (excluding race-ethnicity) of states, and equalize TANF benefit levels. In short, families experiencing job loss in states with higher shares of Latino residents in particular have relatively more to gain if barriers to accessing benefits for which the families are likely eligible to receive were fewer.

DISCUSSION AND CONCLUSION

Government income support programs play a critical role in reducing hardship and economic insecurity in the event of job loss. However, that income support is not provided automatically. Potential claimants must navigate a myriad of bureaucratic processes to claim the benefits, and many potential recipients either do not initiate or complete the process. These administrative burdens lead to large sums of unclaimed income transfers each year. In this article, we examine variation in the receipt of income transfers after a family member experienced job loss. We underscore three key findings.

First, individuals from low-income families experiencing job loss and who are eligible to receive income transfers are less likely to receive UI or TANF benefits in the late 2010s than in the early 1990s, but more likely to receive SNAP benefits. Thus administrative burdens, broadly defined, have worsened for UI and TANF but not for SNAP.

Second, and relatedly, individuals currently receive more of the benefits available to them than they used to. In other words, the potential increase in replacement rates for receiving all available TANF, SNAP, and UI transfers is smaller in the late 2010s than in the early 1990s. From this perspective, the contribution of administrative burdens to economic insecurity has decreased, largely due to rising SNAP participation among the eligible. However, some of the trend is also due to declining TANF benefit levels for the average state. When the value of potential income transfers declines, removing all administrative burden effects has less consequence because the income transfers do less, by definition, to improve replacement rates.

Third, our evidence suggests that these trends have not occurred evenly across the population. Black and Latino families experience job loss at a greater rate than White families, and also generally lower income replacement rates as a result of administrative burden effects. However, across all racial-ethnic groups, administrative burden effects have declined slightly over time, likely due in part to rising accessibility of SNAP benefits. Moreover, the racial-ethnic differences narrow markedly when adjusting for benefit underreporting with TRIM3. Shifting focus from the race-ethnicity of individuals to the racial-ethnic composition of place, we find that states with higher proportions of White residents tend to have smaller administrative burden effects.

Our operationalization of administrative burdens shifts focus from specific bureaucratic steps that reduce benefit take-up to coverage rates among likely eligible individuals more broadly. This approach has several advantages. First, measuring population coverage is relatively simple and can be readily used to track variation in the intensity of administrative burdens across programs, places, time, and population subgroups. Cross-sectional snapshots of coverage rates not only reveal programs or places that should be targeted for policy intervention, but can also be used to identify aspects of program design or implementation that are effective in boosting enrollment and may be adopted by other programs or jurisdictions. Second, panel data at the household level, as this study uses, can be used to measure coverage rates after a semi-exogenous eligibility (or benefit level) trigger, such as job loss. This type of data can be useful for tracking trends in program access over time among a set of families experiencing relatively similar income shocks.

Beyond these practical and empirical advantages, our approach also shifts the locus of action away from the household and toward the state. Identifying the learning costs, psychological costs, and compliance costs that households face is essential to understanding what does and does not work in program design. Yet these behavioral- or knowledge-based explanations for why households fail to enroll in social programs assume a level of agency with state interactions that may undermine efforts to expand coverage. If an eligible household remains uncovered even after addressing the myriad costs to enrollment, it can be attributed to an affirmative choice to decline assistance. If we instead take universal coverage as the working counterfactual, as this study does, any eligible households not receiving coverage continues to be viewed as a failure of the state that motivates additional interventions. This is important for two reasons. First, for many families, and particularly Black and Latino families, past discriminatory and unfair interactions with the welfare state may lessen their willingness to actively pursue the benefits they are entitled to receive. Reducing enrollment costs is not enough to surmount the legacy of exclusion that colors families’ perceptions of government programs. Second, heterogeneity in the capacity of households to navigate interactions with the state is extreme; an enrollment process deemed low cost or accessible to the median eligible family may still be too difficult for many. Taking full coverage as the goal makes it the explicit responsibility of the state to ensure that all families receive the support they need and deserve, regardless of their cognitive abilities, language facility, or physical capacities.

Moving forward, improving data collection is key to measuring administrative burden intensity and evaluating policy interventions. Data reporting should be standardized and streamlined so that policymakers and the public can quickly evaluate coverage across programs and jurisdictions. A dashboard that includes basic information, such as coverage rates among eligible individuals, time from application to benefit receipt, success rates from initial application, and so on, would be useful for setting benchmarks to evaluate the performance of state and federal governments. This administrative data can and should be paired with improved qualitative and quantitative data collection at the household level. Household panel surveys, for example, could benefit from incorporating a battery of questions regarding experiences navigating program enrollment: whether the household considered applying for a program, number of attempts, outcome, and so on. This information could be used to identify which stages of the initial enrollment or recertification process should be targeted for intervention.

Administrative burdens contribute to lower income replacement rates for families experiencing job loss, our study suggests, particularly for Black and Latino families and in states with large shares of non-White residents. Improving the evaluation of administrative burden effects, and identifying solutions toward reducing them, may be important for promoting greater economic security and reducing racial-ethnic inequalities in economic well-being.