Disparities in U.S. Parental Employment Insecurity and Child Well-Being Across Income Groups: Before, During, and After the Great Recession

Abstract

The Great Recession was the worst economic downturn in the U.S. since the Great Depression of the 1930s. This research focusses on children in five income-quintile groups before, during, and following the Great Recession by analyzing changing disparities across these groups using a new Index of Parental Employment Insecurity, nine additional well-being indicators, and the overall Child Well-Being Index. The new Index of Parental Employment Insecurity captures employment insecurity more fully than the official U.S. monthly unemployment rate by using a comprehensive approach spanning 15-month periods (combining January-to-December measures for a calendar year with measures for March of the following year), based not only on two traditional unemployment indicators, but also two indicators of “hidden unemployment” and two indicators of “underemployment”. The other nine well-being indicators, also with children or youth as the unit of analysis, are median family income, health insurance coverage, Prekindergarten enrollment, three health indicators – very good or excellent health, obesity, and activity limitations – two indicators reflecting family experience – one-parent families and residential mobility – and idle or disconnected youth. For each of the 10 indicators in 1994, children in the lowest-income, lower-middle-income, and middle-income families had lower levels of well-being, often much lower, than children in the highest-income families. These had gaps narrowed, as of 2014, for all three of the lowest-income groups by about two-thirds for health insurance coverage and by about two-fifths for Prekindergarten enrollment. Compared to the highest-income group, however, the gaps had widened for the middle-income group for most indicators, the direction of changes in gaps was mixed for the lower-middle-income group, and the gaps had narrowed, usually by about one-fourth, for the lowest-income group, leaving gaps for the lowest-income group for seven of the ten indicators in 2014 that were at least three-fourths as large as they had been in 1994. In addition, median family income for the highest-income group grew by $32,500 reaching $173,600 in 2014, and the corresponding disparities expanded for the middle-income, lower-middle-income, and lowest-income groups, respectively, to $113,300, $139,400, and $160,800.

Appendices

Appendixes

1.1 Appendix A: Comparing Components of Index of Parental Employment Insecurity

Results for specific components of the Index of Parental Employment Insecurity are presented in Appendix Table 2 and Fig. 13. Appendix Table 2 shows that children were more likely to have a parent who was unemployed during 2014 than to have a parent who was unemployed during March 2015, with differences for various income-quintile groups of 3–5% points. Similarly, children were more likely to have a parent who was underemployed during 2014 than to have a parent who was unemployed during March 2015, with differences for various income-quintile groups of 3–8% points. Thus, children in specific income-quintile groups were nearly or more than twice as likely to have unemployed parent or an underemployed parent in 2014 than to have a parent who was unemployed during March 2015. In addition, children in each income-quintile group were nearly as likely to have a hidden unemployed parent and nearly as likely to have an underemployed parent in March 2015 as to have an officially unemployed parent in March 2015.

Table 2 Percent Children with Parental Official or Annual Unemployment, Hidden Unemployment, or Underemployment, by Children’s Family-Income Quintile: 2014 and March 2015

Fig. 13

Percent of Children with Parental Employment Insecurity for Discreet Categories of Official and Annual Unemployment, Hidden Unemployment, and Underemployment: by Children’s Family Income Quintile: 2014 and March 2015

Table 1 Income Ranges in 2014 for Five Children’s Family Income Groups

Of course, some children are included in more than one of these categories either because one parent was included in the same or different categories in both 2014 and March 2015, or because a second parent in the home was included in the same or different categories during a period when the first parent was not included as employment insecure. Fig. 13 presents results which disentangle these categories. The results indicate for various income-quintile groups that the annual parental unemployment rate in 2014 (excluding children with a parent who was unemployed in March 2015) was nearly as larger than the rate of official parental unemployment in March 2015. Similarly, the rate of parental underemployment for 2014 and/or March 2015 (excluding children with an unemployed or hidden unemployed parent in 2014 and/or March 2015) was as larger or larger than the rate of official parental unemployment in March 2015. Finally, the rate of parental hidden unemployment for 2014 and/or March 2015 (excluding children with an unemployed parent in 2014 or March 2015) was nearly as large as the rate of official parental unemployment in March 2015. Thus, both this annual parental unemployment measure and this hidden parental underemployment measure contribute as much or more to the overall magnitude of the Index of Parental Employment Insecurity than does the official parental unemployment measure.

Appendix B: Calculating the Child Well-Being Index

The overall measure of child-well being presented here extends the approach of Land and colleagues to provide the basis for assessing both the magnitude and trends in disparities across children in the five income groups. Based on data for 28 key national indicators of child well-being that measure seven quality of life domains, the numerical value of the CWI for a given year is calculated in the four following steps (Land et al. 2001; Land 2005a, b).

First, assign a value of 100 to each indicator for the baseline year. Second, for each subsequent year calculate the percentage change from the baseline year for that indicator. Third, for a specific year equally weight the values for the indicators in a domain (calculate the arithmetic mean) to obtain the percentage average change for the domain as a whole. Fourth, for a specific year equally weight the values of each domain (calculated as the arithmetic mean of the values for the seven domains) to obtain the overall CWI value for the year. Thus, the CWI is an evidence-based measure indicating the average amount of change that was experienced by children across the seven domains between a baseline year and a subsequent year.

A corresponding set of CWIs can be calculated for each of the five child income-quintile groups by applying the same procedures to each group separately (Land et al. 2001; Land 2005a, b). Using this method, the results measure the average percentage change experienced by each group, using its own starting point as the baseline. However, because the baseline for each group is set to a value of 100, disparities cannot be assessed at the baseline year or subsequent years.

To directly assess the magnitude of disparities, and their trends through time, this research uses an alternative approach for specific income-quintile groups. A value of 100 is assigned to each indicator for the baseline year for the total population. Then for each specific group, the percent difference between that group and the total population is calculated for the baseline year. Thus, differences in values across income-quintile groups in the baseline year reflect their deviation from the mean, which is used to compare groups and measure disparities.

Using these results as the starting point, the subsequent steps in the original procedure are followed to calculate a modified CWI for each specific group. Thus, the value for a group-specific CWI for any subsequent year using this method reflects the extent to which that group differs on average in the quality of life from the total population in the baseline year, that is, the total population in the baseline year serves as a benchmark for measuring disparities. Furthermore, differences across groups in a specific year reflect the extent to which these groups differ from each other in that year, using the indicator values for the total population in the baseline year as the starting point. Two groups will, therefore, have CWI values that are equal in a given year, only if they are experiencing the same average quality of life. However, their CWI values will differ as long as there is a disparity between the groups in their average quality of life.

Results from this new approach contrast with the approach developed by Land et al. (2001), which leads to identical CWI values for two groups in a given year if these groups have experienced similar percent changes since the baseline year, even if disparities across groups remain large. But the new approach leads to identical CWI values for two groups in a given year only if the groups have identical levels of well-being.

For the present analysis, 10 indicators can be included in the CWI, because data are available only for these indicators to calculate rates for specific income groups. Three are indicators of family economic resources (median income, parental employment insecurity, and health insurance coverage), three are indicators of health (obesity, activity limitations, and very good or excellent health), two are indicators of social relations (one parent family and moved during the past year), and two are indicators of community engagement (Prekindergarten enrollment and idle or disconnected youth). These ten indicators and four domains of well-being are used to calculate values of the CWI for the present research.

Appendix C: Measuring Income Quintiles, Index of Parental Employment Insecurity, and Housing Costs

1.1 Income Quintiles

The March Current Population Survey (CPS) and its Annual Social and Economic Supplement (ACES) collect very detailed income data that provide the basis for precisely calculating family-income quintile values for children which are used in this report to assign each child to a specific family-income quintile group. The authors calculated these results for income quintiles from March CPS data provided by Flood et al. (2017). The most recent data analyzed for this report were collected in March 2015 for annual income in calendar 2014. March CPS data provide the basis for the analyses of income variables.

1.2 Index of Parental Employment Insecurity: Six Component Measures

The new Index of Parental Employment Insecurity is based on six specific employment concepts using data collected in the Annual Social and Economic Supplement of the March Current Population Survey. The most recent data analyzed for this report were collected in March 2015. The index is calculated as the percent of children with at least one parent who, as indicated below, was classified as unemployed, as hidden unemployed, or as underemployed. March CPS data used by the authors to calculate these results were obtained from Flood, et al. (2015).

The first concept is official unemployment as of March, which classifies persons as unemployed, if they were not employed at the time of the March interview, and they had been looking for work during the past 4 weeks. The authors used the EMPSTAT variable to identify parents as officially unemployed in March.

The second concept is hidden unemployment as of March, which classifies persons as hidden unemployed, if they were not employed at the time of the March interview and had not looked for work during the past 4 weeks, but did want a job. The authors used the WANTJOB variable to identify parents as hidden unemployed in March.

The third concept is underemployment as of March, which classifies persons as underemployed, if they are working part-time but want a full-time job. The authors used the WHYPTLW variable to identify parents as underemployed if they reported working part time because they could only find part-time work or they worked part time because of slack work.

The fourth concept is unemployment during a calendar year, which classifies persons as unemployed, if they report looking for work or being on layoff for at least 1 week during the preceding calendar year. The authors used the WKSUNEM1 variable to identify parents as unemployed during a calendar year.

The fifth concept is hidden unemployment during a calendar year, which classifies persons as hidden unemployed, if they looked for work (and were not employed or on layoff) for at least 1 week during the preceding calendar year. The authors used the NWLOOKWK variable to identify parents as hidden unemployed during a calendar year.

The sixth concept is underemployment during a calendar year, which classifies persons as underemployed, if they are working part-time but want a full-time job. The authors used the variable WHYPTLY to identify parents as underemployed if they reported working part time because they could not find a full-time job or because of slack work during a calendar year.

1.3 Housing Costs

Housing costs were calculated with data from the 2013 American Housing Survey conducted by the U.S. Census Bureau (2015), retrieved with assistance from Jean Mullin. Housing costs include mortgage or rent and utilities. Housing costs were reported in broad categories. The housing cost for the lowest-income group was calculated as the mean housing cost for households with incomes of Less Than $10,000 and with incomes of $10,000 to $19,000. The housing cost for the lower-middle-income group was calculated as the mean housing cost for households with incomes of $20,000 to $29,999, with incomes of $30,000 to $39,999, and with incomes of $40,000 to $49,999. The housing cost for the middle-income group was calculated as the mean housing cost for households with incomes of $50,000 to $59,000 and with incomes of $60,000 to $79,000. The most recent available data from the American Housing Survey were collected for 2013.

To measure income values across time that reflect real purchasing power, and to present housing costs consistent with income data, all results in this report are adjusted for inflation using the Consumer Price Index (CPI) and presented as 2014 dollar values (U.S. Bureau of Labor Statistics 2015).

It should be noted that monthly housing costs may be more volatile than annual housing costs because, for example, of differences in the cost of utilities during various months of the year, which introduces some noise into the annual estimates of housing costs calculated by multiplying monthly housing costs by 12. Such statistical noise is dampened, however, by the use of broad income categories in the American Housing Survey. It should also be noted that housing costs may change differentially depending on labor market conditions and other factors. The aim here is not, however, to assess the nature and magnitude of changes in housing costs, but instead to provide a contextual measure to shed light of the relative importance that changes in median family income had for various income-quintile groups, focusing on one of the most important expenditure categories of families.