January 29, 2003 | Center for Data Analysis Report on Poverty and Inequality
In discussions about poverty in America, concern is frequently expressed regarding working poor families with children. Many perceive the working poor as families that work full-time throughout the year yet still have incomes below the official federal poverty levels.
While some poor families fit this profile, most do not. Among poor families, when work does occur, part-time or part-year work is the norm. Examination of data from the U.S. Census Bureau's Current Population Survey (CPS) for 1999 reveals the following:
These findings indicate that public promotion of higher levels of employment and work among poor parents will substantially reduce child poverty. By contrast, policies that reward idleness will increase poverty.
Recent experience indicates that welfare reform policies can be extremely effective in increasing employment. For example, in 1996, Congress reformed the traditional Aid to Families with Dependent Children (AFDC) program, replacing it with a new program called Temporary Assistance to Needy Families (TANF). The TANF program required many adult recipients to engage in constructive activities directed toward self-sufficiency (for example, supervised job search, training, or community service work) as a condition for receiving aid. As a result of these requirements, welfare rolls shrank and employment among single mothers soared. Employment of never-married mothers, for example, increased by 50 percent. As employment among single mothers grew, poverty within that group fell by a third.
Work and Poverty among Families with Children
Each year, the U.S. Bureau of the Census calculates the number of families and persons living in poverty. A person is deemed "poor" if he lives in a family with an income below the official poverty income thresholds. For example, in 1999, the official poverty threshold was $13,423 for a family of three and $16,895 for a family of four.
Obviously, the count of poor persons will vary depending on what economic resources are included as part of the family's income. In this section, we will examine work and poverty using two different measures of income. The first is "money income." This is the most common measure of income employed by the Census Bureau; it includes most cash income received by the family but excludes a wide range of welfare aid such as food stamps, the Earned Income Tax Credit, and public housing.
We also analyze poverty using an "expanded measure of income" that includes food stamps, the Earned Income Tax Credit, and school lunch subsidies. This measure also deducts FICA, or Social Security taxes, from income.
Chart 1 and Table 1 both show the level of paid adult employment among poor families with children in 1999 using the "money income" standard. As the chart shows, roughly one-quarter of poor families with children had no adult employment during the year. Approximately one-quarter had adult employment between one and 999 hours during the year; and another quarter had between 1,000 and 1,999 hours. Slightly more than one-quarter had at least full-time/full-year employment with 2,000 or more hours of paid work. (The figure of 2,000 hours represents an average of 40 hours of work per week over 50 weeks.)
While the level of work among these poor families is greater than generally perceived, nearly three in four working-poor families had less than full-time/full-year employment. The median number of hours of work among all the families was 1,040 per year. The lack of full-time employment was a major factor contributing to poverty.
Table 2 presents the same data using the expanded definition of income. With the inclusion of the EITC, food stamps, and school lunches in calculating income, the number of poor families with children falls significantly-from 5.4 million in Table 1 to 4.4 million in Table 2. Since many of the added welfare benefits supplement the wages of low-income parents, the share of poor families with over 2,000 hours of employment falls from 27.8 percent in Table 1 to 23.5 percent in Table 2. The share of poor families with no employment increases to 32.5 percent in Table 2. The median hours of work among all poor families falls to 660 hours per year.
The differences between Table 1 and Table 2 are evidence of the effectiveness of some welfare programs (especially the EITC) in raising the incomes of working families-particularly those with full-time workers-above the poverty thresholds.
As in Table 1, married-couple families are far less likely to have no employment than single-parent families (15.9 percent compared to 39.7 percent). Nearly half of poor married-couple families had more than 2,000 hours of work in the year, compared to only 12.7 percent of single-parent families.
Over-Reporting Hours Worked
The number of families that work a substantial number of hours during the year yet remain poor appears surprising. Indeed, examination of the data strongly suggests that work levels shown in Tables 1 and 2 are significantly over-reported.
Throughout the CPS data on poor families, there is a significant discrepancy between reported hours of work and reported earnings. As Table 3 shows, when total adult annual earnings in a poor family are divided by the reported hours of adult work during the year, 42 percent of poor working families appear to have adult wage rates below the federal minimum wage of $5.15 per hour. This phenomenon is especially pronounced among poor families that report over 2,000 hours of adult employment during the year. Within that group, 52 percent reported effective wage rates that were below the minimum wage.
Table 4 shows the same data in a different form. The left half of the table shows the effective wage rates (annual earnings divided by annual number of hours worked) of adults in poor families. Among families with any reported adult employment, over one-fourth have effective wage rates of less than $4.00 per hour, and 42 percent have effective wage rates of less than $5.00 per hour.
The right half of Table 4 shows effective wage rates among poor families that report over 2,000 annual hours of adult employment. Of these, some 31 percent report wages below $4.00 per hour, and 51 percent report wages below $5.00 per hour.
One possible explanation for these low wage rates would be self-employment. Individuals who run their own small business may well have effective earnings below the minimum wage, especially during start-up years. However, the CPS data show that only 7.5 percent of working adults in poor families are self-employed. Among poor families reporting over 2,000 hours of employment, the level of self-employment is higher but still not great: 16.5 percent.
The most plausible explanation of the low effective wage rates among the working poor is that, among that group in the CPS, employment has been slightly over-reported and earnings have been somewhat under-reported. Overall, the number of families who work full-time/full-year and remain poor is almost certainly significantly lower than the figures shown in Tables 1 and 2.
Reducing Child Poverty by Increasing Work
Since nearly three-quarters of poor families with children have less than full-time/full-year employment, it follows that child poverty can readily be cut if the work levels in these families are increased. This section seeks to answer the following question: How much would child poverty be reduced if all poor families with children had at least one adult who worked full-time throughout the year?
To answer this question, we simulated an increase in hours worked using the CPS data file. In this simulation, for each poor family with less than 2,000 annual hours of adult employment, adult employment was increased to the 2,000-hour level and adult earnings were increased in a corresponding ratio. Thus, for example, if a family reported 1,000 hours of work yielding $6,000 in earnings, the simulation would increase the hours of work to 2,000 and the earnings would be increased proportionally to $12,000.
The simulation assumed that work is increased to 2,000 hours per year per family, not 2,000 hours per worker. Thus, for families with more than one adult, the simulation did not assume that each adult would work 2,000 hours; rather, the total employment level for all adults in the house combined was raised to 2,000 hours. For example, one parent might work 1,500 hours while the other worked 500 hours during the year.
If the effective adult hourly wage rate in the family was reported to be greater than the minimum wage in the CPS data, the reported hourly wage was retained in the simulation. (As above, the effective hourly wage rate equals total annual adult earnings divided by the total annual adult hours of work.) If the effective hourly wage rate shown in the CPS was less than the federal minimum wage of $5.15 per hour, hourly earnings were increased to equal the minimum wage. Thus, every family in the simulation had imputed earnings equal to at least $10,300 (2,000 hours of work times $5.15 per hour).
To measure the effects on poverty, the analysis used the expanded definition of income that adds the value of the EITC, food stamps, and school lunch subsidies to conventional "money income." Welfare benefits from the Temporary Assistance to Needy Families and Supplemental Security Income programs were assumed to be eliminated by the earnings increase. Any current payments from unemployment insurance, worker's compensation, and disability or retirement income were also eliminated from family income.
The values of the EITC and FICA taxes were adjusted to correspond to the earnings increase within the family. Food stamp benefits were also adjusted to match the increase in earnings within the family; however, not all families who were eligible for food stamps were deemed to receive them under the simulation. At present in the United States, only around 70 percent of the poor families with children who are potentially eligible to receive food stamps actually get them. The simulation assumed that this under-utilization of food stamp benefits would continue. Therefore, it was assumed that, after the simulated earnings increase, approximately 30 percent of families who were still eligible for food stamps would not receive them.
The Impact of
Increased Parental Work
Charts 2, 3, and 4 show the dramatic effects of the simulated increase in work and earnings on poverty.
More complete data showing the effects of increasing parental work to full-time/full-year are presented in Tables 5 and 6.
The analysis presented in this paper is relatively straightforward. Low work levels by parents are a major cause of child poverty. While it is true that most poor families with children are "working families," on average, the level of employment in poor families is quite low. Roughly three-quarters of all poor families with children have total parental work levels of less than 2,000 hours per year (the equivalent of one adult working 40 hours per week for 50 weeks). Nearly half of all poor families with children have less than 1,000 hours of paid employment throughout the year.
The government currently makes extensive efforts to supplement the incomes of low-wage working parents through programs such as the Earned Income Tax Credit, food stamps, and the school lunch program. These programs significantly raise the effective earning power of low-skill parents. For example, the typical single mother working full-time throughout the year at the minimum wage level of $5.15 per hour actually has an effective wage closer to $8.00 per hour once the value of these government income supports is counted as part of her income.
Despite the availability of this government support, nearly 4.4 million families with children remain in poverty. Much of the public discussion on reducing child poverty focuses on raising the hourly wage rates of parents. However, the analysis presented in this paper indicates that child poverty is more often the result of low levels of parental work than of parents' low hourly wage rates. Even at current wage rates, child poverty could be dramatically reduced simply by increasing the number of hours that parents work during the course of a year. This point is particularly important for policymakers, given the fact that government training programs have been relatively ineffective in increasing hourly pay rates.
Overall, if the number of hours of work among poor families with children was increased to 2,000 hours per year for each family, the percentage of families with children who were poor would be cut by 72 percent. While it is true that some families may not be able to maintain 2,000 hours of employment per year, the data still show that any policy that significantly increased the current low levels of parental work would have an enormous positive effect in reducing child poverty. Conversely, policies that ignore the current low levels of parental work are not likely to be successful in reducing poverty.
Robert E. Rector is a Senior Research Fellow, and Rea S. Hederman, Jr.,is Manager of Operations in the Center for Data Analysis, at The Heritage Foundation.
This study was conducted using the U.S. Bureau of the Census Current Population Survey (CPS) conducted in March 2000. This survey covers incomes for calendar year 1999.
All families included in this study had at least one related family member under the age of 18. Families without children and unrelated individuals were excluded. To be consistent with the Census Bureau's current practice in assessing poverty, related subfamilies are treated as part of the primary family. Unrelated subfamilies with children were treated as separate primary families.
Families were deemed to be in poverty if the family income was below the poverty income thresholds as specified the Census Bureau. These poverty income thresholds vary with family size, the number of children, and age of the householder.
Two definitions of income were used in the paper. The first is "money income." This corresponds to the Census Bureau's "definition one" of income, also called the "official definition of income." The money income concept was used for the calculations presented in Tables 1, 3, and 4.
A second definition of income was also used, which we have called "expanded income." This includes money income plus the value of the Earned Income Tax Credit, food stamps, school lunch subsidies, and capital gains, less capital losses and FICA taxes. The expanded income definition was used in Tables 2, 5, and 6. Earnings and hours of work of individuals under age 18 were excluded from all tables and figures in the report regardless of the definition of income used.
Annual earnings for each adult were taken from the person-level records of the Current Population Survey. If an individual held more than one job, the earnings from different jobs were added together to produce an aggregate earnings figure.
The average number of hours worked per week and the number of weeks worked during the prior year for each adult were also taken from the personal record in the CPS; the number of weeks of employment during the year and the average number of hours worked each week were multiplied together to produce a total annual employment figure for each adult. Effective hourly wage rates were calculated by dividing the annual earnings by the annual hours of employment. Again, as in all other figures in the paper, the earnings and hours worked by children were excluded; figures for earnings and hours worked pertain to adults only.
If there was more than one working adult in the family, all the earnings and hours of work of all the adult workers were counted. In Tables 3 and 4, earnings of all adults in the family and total hours of work for all adults were pooled to produce an aggregate effective wage rate for the whole family rather than a wage rate for each adult worker.
Increased Earnings in
In Tables 5 and 6, we have calculated the effect of raising the aggregate annual hours of adult employment to 2,000 hours in each poor family, using the following procedures. If there is only one employed adult in the family, the effective hourly wage for that adult was calculated, as well as the annual hours of work performed. A new annual earnings figure was then computed for the family by the formula A times B = C, where:
A = the greater of reported hourly wage of the earner or $5.15 per hour;
B = the greater of reported annual hours of work or 2,000 hours; and
C = new total annual earnings for the family.
According to this procedure, if the number of reported hours of work is over 2,000, there is no assumed increase in work. If the reported effective wage is above $5.15 per hour, there is no assumed increase in hourly wage.
If all adults in the family were over age 65, there was no assumed increase in work. Hourly wage rates were capped at $20.00 per hour.
Increased Earnings in
If there was more than one working adult in the family, the effective hourly wage rate for each working adult was calculated. If the effective hourly wage rate for an individual was below $5.15 per hour, it was raised to $5.15 and a new adjusted annual earnings figure was estimated for that individual.
After the annual earnings for each individual had been adjusted (where necessary), the earnings of all adults in the family were added together to produce an "earnings total" for the family. If the aggregate hours of work for all adults in the family was less than 2,000 hours per year, the aggregate hours of work was adjusted upward to reach 2,000 and the "earnings total" for the whole family was increased pro-rata. The share of total family earnings contributed by each earner would be the same before and after the adjustment.
It is important to note that in multi-earner families, the hours of annual work will be increased to 2,000 hours for all adult workers in the family taken together, not for each individual earner. For example, under the simulation, a family with two adults may have one adult who works 500 hours while the other adult works 1,500 for a joint total of 2,000 hours of work in the family. The simulation does not assume that each adult will work 2,000 hours individually.
According to this procedure, if the number of reported hours of work for the whole family is over 2,000, there is no assumed increase in work. If the effective hourly wage of each worker is above $5.15 per hour, there is no assumed increase in hourly wage. The adjustments described above produce a minimum annual earnings level in each family of $10,300.
If all adults in the family were over age 65, there was no assumed increase in work. Hourly wage rates were capped at $20.00 per hour.
Benefits and Taxes
After the aggregate family earnings were adjusted, the value of the Earned Income Tax Credit and FICA tax were recalculated based on the new earnings level. School lunch subsidies were imputed into family income according to federal eligibility criteria; school-age children in families with incomes below 130 percent of the poverty level were assumed to receive free school lunches throughout the school year.
Public assistance, including Temporary Assistance to Needy Families and Supplemental Security Income, was eliminated in the calculations for all families, since these benefits would generally not be available in families where the parents were fully employed. Unemployment income, disability income, and retirement income were also eliminated.
Receipt of Food Stamps
In United States, at present, not all households that are potentially eligible to receive food stamps actually receive them. In 1999, according to the U.S. Department of Agriculture, some 92 percent of poor single-parent households with children that were eligible to receive food stamps received them. Among poor married-couple families with children, some 57 percent of those eligible to receive food stamps did receive them in 1999. For purposes of the simulation, it was assumed that these patterns of receipt and non-receipt would continue after the hypothetical increase in work and earnings.
To determine which families would receive food stamps, a four-step process was used.
First, the total earned income of each family was increased according to the procedures described above.
Second, we determined whether a family would be eligible for food stamps, based on the family's increased earnings and other adjustments to income.
Third, among those families remaining eligible for food stamps, a random selection was made to determine which families would receive stamps. To perform the random selection, each eligible household was randomly assigned a number; numbers within certain ranges were deemed not to receive food stamps. Overall, as a result of this selection process, 92 percent of eligible single-parent families and 57 percent of eligible married-couple families were deemed as receiving food stamps.
Finally, among those eligible families that were selected to receive food stamps, the annual value of food stamp benefits was calculated based on the adjusted earnings and other income in the family.
New Net Family Income
After the above adjustments were completed, the family's new "total earnings," other cash income (if any), and applicable EITC, food stamp, and school lunch benefits were added together. FICA taxes were then subtracted from this sum to yield a new net family income variable. The new net family income variable was then compared to the official poverty income threshold for a family of the same size to determine whether the family was still poor.
The mean income of the 4.37 million poor families with children in 1999 was $10,204. After the simulated increase in hours worked, the mean family income would rise to $18,402.
One factor that contributes modestly to poverty among working families is larger family size. Since the official poverty income threshold increases with family size, families with more children need to earn more to keep the family's income above poverty. Poor families with children, on average, have 2.2 children per family. Poor married couples tend to have more children than poor single mothers (2.6 compared to 2.1). Poor families that report no adult work have fewer children (2.0), while poor families that report over 2,000 hours of work have, on average, more children (2.5).
To calculate the number of hours of work an individual performs during a year from Census data, it is necessary to multiply the self-reported number of weeks of work during the year by the self-reported average hours of work per week. Since the figures provided are imprecise, an over-reporting of total work can readily occur.
The value of a family's benefits from the food stamp program and from the Earned Income Tax Credit is a function of the family's earnings. We have recalculated the expected value of benefits from these two programs based on the adjusted value of earnings in the family. As a result of the earnings adjustment, each single parent with two children would typically have a minimum annual income of $16,123. This represents $10,300 in earnings, $3,816 from EITC, $2,030 from food stamps, and $765 from school lunches, less $788 in FICA taxes. The poverty income threshold for this family was $13,423. Thus, a family of three working full-time at the minimum wage would typically have a total income 20 percent above the poverty level. However, as noted in the text, not all families who are eligible to receive food stamps actually get them. Under the simulation, the minimum annual income for a family of three who did not get food stamps would be $14,093. This represents $10,300 in earnings, $3,816 from EITC, and $765 from school lunches, less $788 in FICA taxes. This sum would be 5 percent above the poverty income level.
Randy Rosso, Trends in Food Stamp Program Participation: 1994 to 1999, U.S. Department of Agriculture, Food and Nutrition Service, Office of Analysis, Nutrition and Evaluation, October 2001, pp. 143-136.
The procedures for allocating receipt and non-receipt of food stamps are described more fully in the Appendix.
All the figures in Charts 2, 3, and 4 and all the figures in the text that refer to reductions in poverty and increases in income utilize the expanded definition of income.
See note 1, supra.
For example, a large-scale U.S. Department of Labor-funded evaluation of the Job Training Partnership Act (JTPA) found that training programs increased the hourly wage rates of female trainees by 3.4 percent and the hourly wage rates of male trainees by 0.0 percent. See Howard S. Bloom et al., "The National JTPA Study, Overview: Title II-A Impacts on Earnings and Employment at 18 Months," Abt Associates, prepared for the U.S. Department of Labor, January 1993, p. 6.
U.S. Bureau of the Census, Current Population Survey, March 2000.
The Census Bureau considers an "unrelated subfamily" to be two or more people related to each other and living in the same household, but not related to the householder.
The poverty thresholds for 1999 were taken from U.S. Bureau of the Census, Current Population Survey, Poverty and Health Statistics Branch/HHES Division.
See U.S. Bureau of the Census, Money Income in the United States, 1999, Appendix A.