Survey of Income and Program Participation Small Grants
Abstracts of Funded Projects
Multiple-Father Fertility: Prevalence and Connection to the Criminal Justice System
Eirik Evenhouse, Mills College and Siobhan Reilly, Mills College
Multi-partnered fertility is a fast-growing, but poorly understood, phenomenon that is highly prevalent in some vulnerable subgroups. It presents knotty problems in policy design for poverty reduction, child-support, and marriage. There is emerging evidence that it poses a threat to family well-being. For example, it appears to discourage father-child contact and to reduce fathers’ payment of child support. The existence of children from prior relationships reduces the probability of marriage by both men and women, and often makes it economically disadvantageous for low-income couples to marry. Multi-partnered fertility may also hold advantages (for example, by expanding or diversifying sources of household income), but these have had scant formal exploration.
Even less is known about the causes of multi-partnered fertility than about its consequences. Given its rapid growth, and its challenges for family well-being and for policy design, the causes and correlates of multi-partnered fertility deserve careful study.
The SIPP offers researchers the possibility of measuring multiple-father fertility in a large survey because, from its start in 1985, its detailed household relationship data have enabled the analyst to differentiate full siblings from half siblings. From each household’s relationship matrix, one can determine how many fathers are represented among a mother’s children. Pooling the 11 SIPP surveys that are currently available yields observations on approximately 60,000 mothers over a two-decade period (1985-2004). The 2008 SIPP will expand that sample.
This study has four objectives. One is to provide the first nationally representative time-series estimates of the prevalence of multiple-father fertility from 1985-2008 (in the population as a whole and in key subgroups), based on SIPP’s household relationship matrices. Another is to gauge the time-consistency of SIPP’s relationship and its agreement with other family structure indicators in SIPP. A third goal is to examine the relationship between the propensity of mothers to have children by more than one man and local rates of male involvement with the criminal justice system (the arrest and incarceration data are from the FBI’s Uniform Crime Reports and the Bureau of Justice Statistics, respectively). Finally, the study will lay the groundwork to use SIPP data to investigate the impact of other policy variables (such as child support enforcement or TANF/AFDC) on multi-partnered fertility.
________________________________________________________________________________
Gaps in Health Coverage among Mexican Immigrant Children: How Much Does Legal Status Matter?
Deborah Roempke Graefe, Pennsylvania State University
The aim of this project is to document the patterns and trajectories of health insurance coverage among Mexican-origin children of immigrants and to compare these patterns and trajectories both with those of U.S. native racial/ethnic groups, including non-Hispanic white, African American, and Hispanic children, and within-group by immigrant documentation status. Little is known about the effects of immigrant documentation status on the health of Mexican parents and their children, although the Mexican-origin population is among the most likely to report having neither health insurance coverage nor a regular source of health care. And while more than 5 million U.S. children have Mexican immigrant parents, and this group is growing, health insurance coverage for undocumented immigrants remains a contentious topic of current health reform policy debates. This project uses the 1996, 2001, 2004, and 2008 SIPP panels, which permits tests of differences between these groups during periods of distinct national economic growth and recession, including the current period during which lack of health insurance has grown. The longitudinal nature of SIPP permits documentation and evaluation of health insurance coverage on a monthly basis, and the use of Gold Standard Restricted-Use Data makes possible the development of an algorithm for estimating parent and child immigrant legal status. This algorithm will be applied to public use files for comparing coverage patterns and trajectories by immigrant documentation status. Descriptive results, including cumulative proportions experiencing transitions to and from coverage, based upon life table analyses, will demonstrate the patterns and trajectories for each comparison group. This study provides a foundation for needed future research to evaluate the role of documentation status in observed patterns and trajectories of health insurance coverage against alternative family, socioeconomic, demographic, immigrant assimilation, and context-of-residence public health policy explanations.
________________________________________________________________________________
Earning Profiles and Employment Trajectories as Outcomes and Precursors: Validating Analysis Based on SIPP Synthesized Beta Data
Lingxin Hao, Johns Hopkins University
Wage and income disparities by skills, race, and other social groupings have been widely studied and the findings are consistent to support the conclusion of rising economic inequality since 1970s. However, the disparities in the shapes of individual workers' entire earnings profiles and employment trajectories have yet to be established and the social consequences, such as levels of family well-being, of the individuals' previous labor market mini-histories are less known. A reason for these limitations is a lack of long panel data. The matched survey data and administrative data in the Survey of Income and Program Participation Synthesized Beta Data (SSB) provide a rare, valuable opportunity to address these challenges. The available long and uninterrupted individual histories of earnings and employment for many birth cohorts now make possible both cohort and age comparisons of the shapes of earnings and employment distributions.
The project has three primary objectives. First, it analyzes individual earnings profiles and employment trajectories as a function of age controlling for social stratification factors, as well as the differential effects of this age function by levels of skills, race/ethnicity, gender, and nativity. Second, it examines the effects of the characteristics of earnings profiles and employment trajectories of husband versus wife as precursors for current family well-being measures. Third, it develops direct and indirect validation methods to assess the degree of validity of estimates based on the SSB data against the estimates based on the confidential Gold Standard Restricted-Use Data.
Findings from this project will identify which social groups and which birth cohorts have the most disadvantaged earnings profiles and employment trajectories as well as at which life stages the disadvantage deepens. The types of earnings profiles and employment trajectories that predict low levels of family well-being will be pinpointed. Such information will provide policy implications for stopping and reversing the rising inequality trend. In addition, the validation of SSB results will inform the academic community and the Census Bureau about the usability of SSB while considering intersecting trajectories along the life course of individuals.
________________________________________________________________________________
Veteran Status, Disability, Poverty and Material Hardship
Colleen Heflin, University of Missouri, Andrew S. London, Syracuse University, and Janet M. Wilmoth, Syracuse University
Veterans are a sizeable and policy-relevant demographic group in the United States (U.S Census Bureau 2003). This research has three objectives: (1) to estimate poverty and the inability to meet the basic needs (i.e., food, housing, and bill-paying hardship) of veterans and nonveterans with and without disabilities over a fifteen-year period; (2) to examine the extent to which veteran status is associated with the risk of poverty and material hardships, taking into account disability status and various socio-demographic characteristics; and (3) to determine the extent to which veteran status moderates the association between disability status and poverty/material hardships.
This research uses the 1992 to 2004 panels of the Survey of Income and Program Participation (SIPP). Food hardship, bill-paying hardship and housing hardship will be examined using information from the Adult Well-Being Topic Modules in addition to the federal poverty measure. The SIPP also contains information on military service, income from veteran-related programs, and a series of disability measures.
The military as an institution provides extensive benefits and services to individuals who are deemed deserving of such support because they have served or sacrificed for the state. This social contract provides an array of services and benefits that have the potential to reduce poverty and material hardship among recipients. Yet little, if any, research has examined that possibility. Evidence of such poverty- and hardship-reducing effects among veterans can inform policy, benefit and service development for other needy or vulnerable populations, and stimulate additional research that could do the same.
______________________________________________________________________________
Partially Indentifying the Impact of Food Stamps on Food Insecurity among Children: Addressing Endogeneity and Misreporting Using the SIPP
Brent Kreider, Iowa State University, Craig Gundersen, University of Illinois, and John Pepper, University of Virginia
Using data from the 2004 panel of the Survey of Income and Program Participation (SIPP), this project looks at the impact of the Supplemental Nutrition Assistance Program (SNAP) – formerly known as the Food Stamp Program – on child food insecurity. By far the largest food assistance program in the United States, SNAP’s central goal is to alleviate food insecurity. In this light, policymakers have been perplexed to find positive associations between food insecurity and the receipt of food stamps. In this project, confronts the two main issues confounding identification of the causal impacts of SNAP on food security: (1) Endogenous selection into the program; and (2) extensive systematic misreporting of participation status. To do so, partial identification bounding methods to account for these two identification problems in a single unifying framework are applied and extended. Imposing relatively weak nonparametric assumptions on the selection and reporting error processes, tight bounds on the impact of SNAP on child food insecurity are provided. The SIPP is especially well suited for this project because, along with containing information on food insecurity, it includes all of the information needed to establish eligibility into the program (other datasets are less comprehensive), and SNAP participation is observed at repeated intervals over time.
________________________________________________________________________________
Divorce and Women's Risk of Health Insurance Loss
Bridget Lavelle, University of Michigan
The full scope of the economic consequences of marital disruption may be broader than previous studies suggest, extending beyond monetary resources to health insurance coverage. Little is known about how marital transitions affect health insurance coverage. This study focuses on the linkages between divorce and women’s risk of health insurance loss, because women are nearly twice as likely as men to be insured as dependents, often on a spouse’s insurance policy. To enhance our understanding of health insurance dynamics and the economic consequences of divorce in the U.S., this project analyzes monthly marital histories and health insurance experiences of nonelderly women in the large nationally representative 2001 and 2004 Panels of the Survey of Income and Program Participation. Three key research questions are addressed: (1) Does divorce increase women’s risk of health insurance loss? (2) What is the temporal pattern of insurance loss and recovery relative to the time of divorce? (3) How does the risk of insurance loss after divorce vary by source of health insurance and employment status prior to divorce? Women at the greatest risk for insurance loss may be those who are insured as spousal dependents prior to divorce, and those who do not have access to coverage through their own employers. Fixed-effects models are estimated to examine to what extent and for what length of time divorce jeopardizes women’s health insurance coverage, and what factors moderate this risk.
________________________________________________________________________________
The Effects of Parental Income on Children's Well-Being and Future Success: An Analysis of the SIPP Matched to SSA Earnings Data
Bhashkar Mazumder, Federal Reserve Bank of Chicago
A vast literature in the social sciences has analyzed the importance of parental economic resources on children’s well-being. However, few studies have been able to access parental earnings histories over long periods of time for a very large sample of families in the US. Similarly, few studies have been able to distinguish the relative importance of parental income in specific periods of the life course of the child. A growing literature has shown that there are critical periods in childhood development where material resources may be especially valuable (e.g. Cunha and Heckman 2007). Finally, studies have not yet been able to convincingly demonstrate that the statistical associations between parental income and children’s outcomes truly reflect causal processes and are therefore amenable to policy interventions.
This project addresses these issues by assembling a rich intergenerational dataset that combines various SIPP panels with SSA administrative earnings records. Measures of parental earnings will be taken over many years and at various points of the child’s life course. Further, the project will use two plausibly exogenous sources of income variation: parental job loss due to work place closings and changes in the Earned Income Tax Credit that increased family resources. The project will estimate the effects of parental income on measures of well-being during childhood as well as children’s future income as adults. In addition to providing broad descriptive measures, there will be a particular focus on estimating causal effects and on identifying whether the timing of parental income over the life course matters. A critical part of the analysis is demonstrating the value of the SSA administrative earnings data (Summary Earnings Records (SER) and Detailed Earnings Records (DER)) for research purposes. The project will inform Census about the quality of income data in the SIPP and in the SSB.
________________________________________________________________________________
Food Stamp and TANF Program Participation with Matched Administrative and SIPP Survey Data
Bruce D. Meyer, University of Chicago and NBER
Benefit receipt in major household surveys is often underreported. This mis-reporting has important implications for our understanding of the economic circumstances of disadvantaged populations, program takeup, the distributional effects of government programs, and studies of other program effects. We propose to examine administrative data on the Food Stamp Program (FSP) and Temporary Assistance for Needy Families (TANF) participation matched to Survey of Income and Program Participation (SIPP) household survey data. First, is an assessment of the level of agreement between the administrative and SIPP survey data. Then, how mis-reporting, both false negatives and false positives, vary with individual characteristics will be assessed. From these results likely biases can be inferred in several types of FSP and TANF analyses. The determinants of program receipt will then be directly examined using combined administrative and survey data. The combined data will allows for the examination of accurate participation using individual characteristics missing in administrative data. Differences between our conclusions and those obtained used survey data alone will be noted.
This project also examines the accuracy of SIPP imputations of FSP and TANF receipt using the matched data. In 2004, more than ten percent of months of receipt of both benefits from the FSP and TANF were imputed. Often researchers drop imputed observations before conducting their analyses. Since this project has the rare situation of observing a close approximation to true receipt, we can test whether including or dropping the imputed observations brings us closer to the estimates obtained with accurate data. Specifically, the determinants of program takeup will be examined to determine whether they are more accurately predicted when including imputed observations or when excluding them.
________________________________________________________________________________
Non-Enrollment by Children Eligible for Public Health Insurance
Karoline Mortensen, University of Maryland and Hanns Kuttner, Hudson Institute
The goal of this project is to provide a more detailed understanding about uninsured children who are eligible for public programs but not enrolled. The project will use the 2004 SIPP public use files. Our analysis incorporates the longitudinal nature of the SIPP data and the rich topical modules to address three related issues regarding enrollment in public insurance. First demonstrated is the role imputation plays in the beginning and ending of spells without health insurance observed in SIPP. The share of beginnings and endings of spells involve imputation are analyzed to determine whether some spells are an artifact of imputation rather than spells of uninsurance. Next, is a description of the relationship between not having health insurance while eligible for public coverage and other dimensions of children’s well-being, including their family structure, health issues, and their experience of non-health privation. The hypothesis is that some share of spells can be attributed to frictions in enrollment in Medicaid and low demand for health insurance, as well as being in stressed families and being disconnected from public programs. Finally, there is an estimation of the effect of all of these characteristics on the relative odds of ending a spell without health insurance coverage among eligible children.
________________________________________________________________________________
The Effect of Long-Run Earnings Volatility on Static and Dynamic Health Insurance Coverage
Matthew S. Rutledge, University of Michigan
This project proposes to use the Survey of Income and Program Participation (SIPP) Gold Standard Restricted-Use Data, which links health insurance and demographic data from the SIPP to earnings histories from Social Security records, to examine the relationship between a worker’s long-run earnings volatility and insurance status. It is hypothesized that a worker with a volatile earnings history is more likely to be uninsured than a worker with a stable earnings profile, controlling for the level of earnings and years of experience. Further, an insured worker with a volatile earnings history is more likely to lose insurance, and an uninsured worker is less likely to gain coverage, the more volatile her earnings history. No previous project has examined the effect of long-run earnings volatility on either static or dynamic health insurance coverage. In addition, whether the results are robust to the SIPP Synthetic Beta dataset and the Panel Study of Income Dynamics are tested.
________________________________________________________________________________
The Effects of Recessions on the Labor Market Earnings of Vulnerable Workers: A Longitudinal Analysis Using the SIPP Gold Standard Restricted-Use Data
H. Luke Shaefer, University of Michigan
This project would use the Census Bureau’s SIPP Gold Standard Restricted-Use Data to examine the effects of recessions on the labor market earnings of specific groups of vulnerable workers, focusing on the low-skilled and minorities. By examining the 1990-91 and 2001 recessions, the goal is to identify the extent of earnings losses, and the extent and speed at which the earnings of vulnerable workers recovered following recessions, both in absolute terms and relative to more advantaged workers. Little consensus exists regarding these issues, making this study important. This study is further relevant in the context of the current recession, making it timely for informing policy. This project would use the UM-RDC to access the Gold Standard Restricted-Use Data, which links observations from the Survey of Income and Program Participation (SIPP) with the Social Security Administration’s Detailed Earnings Records (DER). This offers a nationally representative sample including annual earnings data for adult respondents from 1978-2003. Analyses will simultaneously be conducted on the SIPP Synthetic Beta.
The sample will be limited to adults ages 25-55 with reasonable labor force attachment. Vulernable workers will be identified using skill level, race and ethnicity. The hypothesis is that, both in absolute and relative terms, vulnerable workers will be hit hardest by recessions. However, it is also hypothesized that vulnerable workers adversely affected by recessions through job displacement will regain pre-recession earnings levels faster during recoveries, when compared to more-advantaged displaced workers.
This project begins descriptively, looking at the longitudinal earnings patterns of workers over the recessions. Next, it examines the absolute association of the two recession periods on vulnerable workers’ annual earnings, controlling for numerous factors. Finally, a difference-in-differences approach will be used to compare differential annual growth rates of earnings prior to, during, and following the exogenous shock of recessions associated with vulnerable workers, relative to less-vulnerable groups. This goes beyond extant knowledge in assessing the differential impact of recessions on the labor market earnings of vulnerable workers, as compared to less vulnerable workers.
_______________________________________________________________________________
WIC Participation by Children who Participate as Infants: The Roles of (Re-)Certification and Changes in Family Composition
Christopher A. Swann, University of North Carolina--Greensboro
The WIC program provides health screenings, nutrition education, and vouchers for nutritious foods to infants, children, and pregnant and postpartum women. Although children make up the majority of WIC recipients, estimates show that participation among eligible children is significantly lower than infants. To better understand this drop in participation, this project studies exits from WIC, and any subsequent re-entry, by children who participate in WIC as infants. The aims are a better understanding of (1) the importance of (re-)certification, changes in the number of WIC-eligible family members, and other child, family, and state characteristics on the probability of exit from the WIC program; (2) whether these factors play different roles for children who exit due to ineligibility compared to those who exit while eligible; and (3) the factors that are associated with re-entry.
Discrete-time logistic hazard rate models are used to study the time in months until exit by children who participate as infants. The models are estimated using data from the 2004 and 2001 panels of the Survey of Income and Program Participation (SIPP). SIPP includes longitudinal information on WIC participation and data required to determine eligibility so that eligibility and participation spells may be constructed. Birth months are used to identify recertification months. Finally, SIPP contains a host of additional control variables.
One of the issues that arises when analyzing duration data with SIPP is seam bias – the bunching of transitions at interview months. SIPP interview procedures were modified for the 2004 panel in order to reduce seam bias. In addition to the study of WIC exits, the project compares data on WIC spells and estimates of hazard rate models across the 2001 and 2004 panels to explore whether the changes to the interview procedures appear to translate into changes in empirical estimates.
________________________________________________________________________________
The Effect of Unemployment on Family Composition, Doubling Up, and Well-Being
Emily Wiemers, University of Michigan
Moving in with family and friends, or doubling up, is one way individuals and families cope with job loss. High vacancy rates in rental units combined with high foreclosure rates provide suggestive evidence that doubling up is increasingly common during the current economic downturn, but there is little work on how prevalent this form of resource sharing is and to what extent it helps families smooth consumption and maintain well-being through difficult times. This project uses data from the Survey of Income and Program Participation to explore the relationship between household composition and unemployment. Three related questions are examined: (1) do families double up during spells of unemployment; (2) how does doubling up affect measures of well-being such as income, food insecurity, health, and overall consumption; and (3) since benefits from some public transfer programs depend on household composition, how does doubling up interact with public transfer programs designed to help households smooth consumption. This project contributes to understanding how unemployment, household composition, and well-being are related. Given the way family composition interacts with benefit receipt, understanding these relationships has important implications for the effectiveness of programs designed to alleviate poverty, and is especially relevant as unemployment rolls continue to expand.
________________________________________________________________________________
Trends in Economic Instability Across Household Surveys
Scott Winship, The Pew Charitable Trusts
Understanding trends in economic instability is vital if economic and social policies aimed at mitigating economic risk are to be effective. Despite the popular perception that economic instability has been rising—and a research literature that often supports this perception—recent studies have cast doubt on the conventional wisdom. At the same time, previous research that has used disparate measures, datasets, and methodological choices provides little guidance to account for different findings across studies. The diversity of findings leaves unanswered the question of whether different datasets yield similar conclusions when analyzed the same way. This study seeks to remedy these problems by estimating comparable trends in economic instability, using multiple measures, across three of the most-used household surveys—the SIPP, CPS, and PSID. The result will be a fuller understanding of trends in economic risk and of the utility of household surveys in measuring volatility.
________________________________________________________________________________
