Families: update



Download 0.69 Mb.
Date21.07.2021
Size0.69 Mb.
#85299





Children in working poor families:

update and extensions

submitted to:
The Foundation for Child Development
by:
Richard Wertheimer Melissa Long

Justin Jager


Contact: Richard Wertheimer, Ph.D.

Child Trends

4301 Connecticut Avenue, NW Washington, DC 20008 (202) 362-5580 dwertheimer@childtrends.org

October 2002


Table of Contents
ACKNOWLEDGMENTS.......................................................................................................................... 1
EXECUTIVE SUMMARY ........................................................................................................................ 2
CHAPTER 1. INTRODUCTION ............................................................................................................ 7
WELFARE REFORM AND WORK ................................................................................................................. 7

OVERVIEW OF THE STUDY ........................................................................................................................ 8

Parental work effort ............................................................................................................................ 8

Earnings.............................................................................................................................................. 8

Defining poverty ................................................................................................................................. 8

External circumstances ....................................................................................................................... 9

KEY QUESTIONS ADDRESSED IN THIS REPORT ......................................................................................... 10

1. To what extent was being in a working family associated with a reduced likelihood of poverty for a child during the late 1990s into 2000? .......................................................................................... 10

2. How common was it for children in poor families to have working parents during the 1995-2000 period? .............................................................................................................................................. 10

3. How were working poor families different from poor families not meeting the working standard during the years 1995-2000? ............................................................................................................ 10

4. How were working poor families different from working, near-poor and more affluent families in the years 1995-2000? ....................................................................................................................... 11

5. How common were movements into and out of working poverty during the 1990s? ................... 11

7. How sensitive are our findings to the definition of poverty? ........................................................ 11

8. Compared with children in other families, how were children in working poor families developing shortly after welfare reform was implemented? Similarly, how are children in families leaving welfare developing compared with families staying on welfare? ..................................................... 12

PREVIOUS STUDIES ................................................................................................................................. 12
CHAPTER 2. DEFINING WORKING POOR FAMILIES................................................................. 15
MEASURING POVERTY ............................................................................................................................ 15

MEASURING WORK ................................................................................................................................. 16
CHAPTER 3. DATA AND METHODS ................................................................................................ 17
DATA: SURVEY OF INCOME AND PROGRAM PARTICIPATION AND THE CURRENT POPULATION SURVEY 17

METHODS ............................................................................................................................................... 18
CHAPTER 4. RESULTS FOR 1995-2000 USING THE CURRENT POPULATION SURVEY (CPS)

.................................................................................................................................................................... 20


HOW MANY CHILDREN LIVE IN WORKING POOR FAMILIES? ..................................................................... 20

TO WHAT EXTENT DOES BEING IN A WORKING FAMILY REDUCE THE LIKELIHOOD OF POVERTY FOR A CHILD?.................................................................................................................................................... 21

Differences by family structure ......................................................................................................... 22

Differences by race/ethnicity ............................................................................................................ 23

Differences by parental education .................................................................................................... 23

HOW COMMON IS IT FOR CHILDREN IN POOR FAMILIES TO HAVE WORKING PARENTS? ............................ 25

Differences by race/ethnicity ............................................................................................................ 26

Differences by parental education .................................................................................................... 26

HOW ARE THE CIRCUMSTANCES OF CHILDREN IN WORKING POOR FAMILIES DIFFERENT FROM THOSE IN FAMILIES NOT MEETING THE WORKING STANDARD? ............................................................................... 27

Patterns by race/ethnicity ................................................................................................................. 27

Patterns by family structure and parental education........................................................................ 29

Patterns by parental wage rate......................................................................................................... 30

Patterns of transfer receipt, health insurance coverage, home ownership, and paid child care utilization .......................................................................................................................................... 30

HOW ARE CHILDREN IN WORKING POOR FAMILIES DIFFERENT FROM THOSE IN OTHER MORE PROSPEROUS WORKING FAMILIES?............................................................................................................................... 33

Patterns by family structure.............................................................................................................. 33

Patterns by race/ethnicity ................................................................................................................. 33

Patterns by family structure and parental education........................................................................ 34

Patterns by parental wage rate......................................................................................................... 35

Patterns by transfer receipt, health insurance coverage, home ownership, and paid child care utilization .......................................................................................................................................... 35

SUMMARY .............................................................................................................................................. 36


CHAPTER 5. THE DYNAMICS OF POVERTY AND WORKING, 1987-1997.............................. 38
PATTERNS OF MOVEMENT INTO AND OUT OF POVERTY ........................................................................... 38

Patterns by family structure.............................................................................................................. 41

Patterns by educational attainment of better educated parent ......................................................... 43

Summary ........................................................................................................................................... 46

CHANGES IN WORK BEHAVIOR AND FAMILY RESOURCES, 1996-1997 ..................................................... 47

Changes in work behavior ................................................................................................................ 48

Changes in family resources ............................................................................................................. 52

Summary ........................................................................................................................................... 54


CHAPTER 6. ANALYSIS OF THE SENSITIVITY OF RESULTS TO THE DEFINITION OF POVERTY ................................................................................................................................................ 56
THE ALTERNATIVE POVERTY DEFINITION................................................................................................ 56

IMPUTING CHILD CARE EXPENSES ........................................................................................................... 57

RESULTS ................................................................................................................................................. 57

SUMMARY .............................................................................................................................................. 62


CHAPTER 7. THE WELL BEING OF CHILDREN IN WORKING POOR FAMILIES AND OTHER FAMILIES ................................................................................................................................. 63
VARIABLES MEASURING HOW WELL THE CHILD IS DEVELOPING ............................................................. 64

VARIABLES ASSOCIATED WITH HOW WELL A CHILD WILL DEVELOP ........................................................ 65

Variables measuring the home environment..................................................................................... 65

Variables measuring interaction with the community ...................................................................... 66

RESULTS ................................................................................................................................................. 67

Results by work and poverty status ................................................................................................... 67

Results by TANF leaver/stayer status ............................................................................................... 73

SUMMARY .............................................................................................................................................. 77

Children in working poor families versus other children ................................................................. 77

Children in TANF leaver families versus children in stayer families ............................................... 77


CHAPTER 8. DISCUSSION .................................................................................................................. 79
SUMMARY .............................................................................................................................................. 79

DISCUSSION ............................................................................................................................................ 79

Negotiating the transition from welfare to work............................................................................... 79

Negotiating the transition out of poverty .......................................................................................... 83


APPENDIX A. IMPUTING CHILD CARE COSTS ............................................................................ 86
ADJUSTMENT OF IMPUTATION EQUATIONS TO REPLICATE SIPP FINDINGS .............................................. 88

IMPUTATIONS IN YEARS OTHER THAN 1997 ............................................................................................ 88


REFERENCES ......................................................................................................................................... 89

Acknowledgments
Financial support for this project was provided by the Foundation for Child

Development.


The authors acknowledge the helpful comments provided by Kristin Moore, Ph.D. on an earlier draft of this report.
Susan Eshleman provided outstanding programming support on the analysis of the 1996

Survey of Income and Program Participation.



Executive Summary
In this study, we define working poor families as families in which annual income is below the official poverty threshold and the family meets a work standard that is dependent on family structure. The work required of single-parent families with children under age 6 is 20 hours per week, while the work required of two-parent families is 35 hours per week. We have translated this into an annual work requirement of 1,820 hours per year for two-parent families and 1,040 hours per year for single-parent families.

Thus, any family with at least this many hours of work during a calendar year is defined as working.


Prior to federal welfare reform, in 1995 and 1996 about 4.5 million children lived in working poor families. During the first two years of welfare reform, this number increased—to 4.7 million in 1997 and 5.0 million in 1998. However, the number of children in working poor families decreased to 4.7 million in both 1999 and 2000. The number of children in poor families not meeting the work standard fell steadily over the entire period, from 9.5 million in 1995 to 6.4 million in 2000. This pattern presumably reflects the combined effects of a strong economy and federal welfare reform—both of which encouraged and facilitated greater work effort by persons with relatively low earning potential.
In Chapter 1 we list seven questions that would serve as the framework for this report. We begin our discussion with a summary of our answers to those questions.
1. To what extent is being in a working family associated with a reduced likelihood of poverty for a child?
Throughout the 1995-2000 period, being in a working family dramatically reduced the likelihood of poverty for a child. Among children in working families only about 8-9 percent were poor. In contrast, among children in families not meeting the work standard

63 percent were poor in 1995-1997, with a substantial decline to 54 percent in 2000. Nonetheless, even in 2000, children in families not meeting the work standard were nearly 7 times as likely as children in working families to be poor.


Poverty rates for children varied substantially by family structure, race/ethnicity, and parental education. Nonetheless, meeting the work standard was associated with a substantial reduction in the risk of child poverty across all demographic and educational groups.
2. How common is it for children in poor families to have working parents?
Even though living in a working family substantially reduces the likelihood of a child being poor, many children live in poor families that meet the working standard. The enactment of federal welfare reform appears to have substantially increased the percentage of poor children who live in working families. In 2000, about 43 percent of all poor children lived in families that met the working standard—a sharp increase from

1995 when 32 percent met the work standard, and a pattern which generally holds across racial/ethnic groups.

3. How are working poor families different from poor families not meeting the working standard?
Compared with children in poor families not meeting the work standard in 2000, children in working poor families are:

• more likely to live with two parents,



about as likely have a parent who has completed 12 years of education,1

less likely to be receiving AFDC/TANF or Food Stamps,

about as likely to be covered by health insurance (an important change since 1995

when the working poor were less likely to be covered),

• more likely to live in a family that owns a home (married-couple families only),2 and

more likely to be in paid child care.

Among children in both working poor families and poor families not meeting the work standard and whose parents paid for child care, child care expenses consumed a much large share of family income than among children in more affluent families.


4. How are working poor families different from working, near-poor and more affluent families?
Compared with children in two categories of more prosperous working families in 2000, children in working poor families are:

• less likely to live with two parents,



less likely to have a parent who has completed 12 years of education,

less likely to have a parent who earns more than $10.00 per hour,

more likely to be receiving AFDC/TANF or Food Stamps,

less likely to be covered by health insurance (married-couple families only),

less likely to live in a family that owns a house, and

less likely to be in paid child care.

Moreover, among children in working poor families with paid child care, child care expenses consumed a much large share of family income than among children in more affluent working families. For example, for single-parent, working poor families with paid child care, nearly two-thirds paid 40 percent or more of their income for child care, and 41 percent paid at least half.



1 This is a change from 1995 when single mothers in working poor families were more likely to have completed 12 years of education.

2 This pattern is the opposite of 1995 when single-mother working poor families were more likely to own their own homes but married-couple working poor families were not.

5. How common are movements into and out of working poverty?


Over time, according to analysis of the Survey of Income and Program Participation (SIPP) there is a rough balance between the number of children entering and leaving poverty each year. (For example, between 1996 and 1997 2.6 million children left poverty and 2.3 million children entered it. This is roughly 20 percent of the total children in poverty at this time.) However, during the 1990-1991 recession, the number of children entering poverty was significantly higher than the number of children leaving poverty; in contrast, during the 1993-94 and 1996-1997 periods of substantial economic growth, the number of children leaving poverty was significantly higher than the number entering poverty.
Consistently over the entire period, increasing parental work effort to meet or exceed the working standard is a much better strategy for moving children out of poverty than not meeting the working standard in either year. However, in every year tabulated, this strategy was successful at removing the child from poverty only about half the time. For the other half, increasing work effort led the family into the ranks of the working poor.
Conversely, meeting the working standard for two consecutive years is a good strategy for avoiding moving into poverty. For nonpoor children whose parents met the working standard in both 1996 and 1997, the likelihood of moving into poverty was only 3 percent. This is significantly lower than the likelihood for (a) those nonpoor children whose parents met the working standard in the second year only (10 percent), and (b) those nonpoor children whose parents failed to meet the working standard in either year (21 percent).
These two findings generally hold regardless of whether the child is in a married-couple family or a single-mother family. In fact, there was no significant difference by family structure in the likelihood of poor children leaving poverty as a consequence of their parent(s) meeting the working standard in 1997, given that they did not meet the working standard in 1996. Moreover, with only a few exceptions, nonpoor children in single- mother families were not significantly more likely than nonpoor children in married- couple families to enter poverty in the following years, regardless of the work behavior

of their parent(s).


Similarly, the pattern of association between parental work behavior and the likelihood of a child exiting or entering poverty held regardless of the educational attainment of the better educated parent. However, there were indications that parental education may play a key role in the likelihood of a child exiting or entering poverty, after controlling for the dynamics of parental work behavior. Higher levels of education appear to be associated with a better chance of escaping poverty and a better chance of avoiding poverty. However, these relationships were not consistently statistically significant.

6. What were the changes in work behavior and economic resources of children in TANF families, children in poor families not meeting the work standard, children in working poor families, children in working, near-poor families, and more affluent working families?


For families that received AFDC/TANF in 1996 and the overlapping group of poor families not meeting the work standard, parents generally increased their work effort between 1996 and 1997—typically both hours worked per week and weeks worked per year. To a somewhat lesser extent, this was also true of parents in working poor families. There was little substantial change in work for working families above the poverty line.
Family income rose significantly for children in families receiving TANF in 1996, families that were poor and did not meet the work standard in 1996, families that were poor and did meet the work standard in 1996, and working families with incomes between 100 percent and 200 percent of the poverty threshold in 1996. They did not change significantly for more affluent working families.
Food Stamp receipt and health insurance coverage fell between 1996 and 1997 for families that received AFDC/TANF in 1996 and also both categories of families that were poor in 1996. Food stamp but not health insurance coverage fell for working families with incomes between 100 percent and 200 percent of the poverty threshold. Health insurance coverage for the more affluent working families was nearly universal and did not change year-to-year, while Food Stamp receipt was virtually non-existent among these families in both years.
7. How sensitive are our findings to the definition of poverty?
We have defined poverty differently and tested the sensitivity of our analysis of children in working poor families to using an alternative definition instead of the official poverty definition. More specifically, we have defined economic resources to include not just

pre-tax money income but also the value of Food Stamps and other non-cash benefits and any refundable earned income tax credit received by the family. We have also excluded from economic resources any positive federal payroll or income tax liability and child care expenses. However, we have continued to use the poverty thresholds of the official poverty definition.


Using this definition results in 18 percent of children being classified as poor in 1996 and

14 percent in 2000. These percentages are slightly lower than the percentage classified as poor using the official definition (20 percent in 1995 and 15 percent in 2000). Moreover, the 1993-1994 decline in the alternative poverty rate is statistically significant.


However, use of the alternative poverty definition has little effect on the share of poor children living in working families. Moreover, findings comparing working poor families with other poor families and other working families (e.g., differences in parental education, receipt of public assistance, health insurance coverage, etc.) are essentially unchanged.

8. Compared with children in other families, how are children in working poor families developing? Similarly, how are children in families leaving welfare developing compared with families staying on welfare?


According to the limited number of variables directly measuring how well a child is developing, children in working poor families and children in poor families that do not meet the work standard appear to fare worse than children in more affluent working families across all three domains—health, cognitive development, and behavior. They also are less likely than children in poor families to be identified as gifted and are more likely to be overweight.
In the area of home environment, children in working poor families also seem to be at a disadvantage compared to at least one other group on the following measures: meals with father, parental involvement, parental aggravation, and father’s educational expectations for his child.
Finally, in the area of interaction with the community, children in working poor families were less active in extracurricular activities than more affluent children and less likely to attend kindergarten, private school, or religious school, while their parents were less likely to hold positive views of their community and more likely to hold negative views.
We found no difference between TANF leaver and stayer families in variables measuring how well a child is developing.
However, there is a mixed picture in the area of home environment. For meals with mother, father involvement, and father’s educational expectations for his child, children in leaver families appear to be at an advantage compared with children in stayer families. On the other hand, appear to be at a slight disadvantage according to the parental aggravation index.
Finally, there are only a few differences between TANF leaver and stayer families for variable measuring interaction with the community. Parents in leaver families were less likely than parents in stayer families to hold negative attitudes towards the community and were less likely to attend religious schools. They were also more likely to receive child care services during the first three months of life.

Chapter 1. Introduction
In 2000, 11.6 million of all children (about 16 percent) in the U.S. lived in families below the official poverty threshold, which was around $17,500 for a family of four. For black children the poverty rate was 31 percent, for Hispanic children the rate was 28 percent, while, for white non-Hispanic children, the rate was 9 percent (Dalaker, 2001). Child poverty rates also varied substantially by family structure. In 2000, only 8 percent of children in married-couple families were poor, while 40 percent of all children in single- mother families were poor (U.S. Census Bureau, 2001a).
All of these percentages have declined sharply between 1996 and 2000. For example, in

1996 (prior to federal welfare reform), over 20 percent of all children lived in poor families and 40 percent of all black and Hispanic children lived below the poverty threshold (Dalaker, 2001).


Although 11.1 million children lived in families3 below the poverty line in 2000, only 6.0 million (about 54 percent) received some form of means-tested cash transfer payment such as Temporary Assistance to Needy Families (TANF) (U.S. Census Bureau, 2001b). This implies that 5.1 million children were living in poor families which were not receiving payments from means-tested cash transfer programs.
Welfare reform and work
Welfare reform placed a much heavier emphasis on work. Under the terms of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) (Greenberg & Savner, 1996), states must:

• Require a parent receiving assistance under the Temporary Assistance to Needy Families (TANF) program to work once the State determines the parent or caretaker is ready to engage in work, or once the parent has received assistance under the program for 24 months, whichever is earlier;

• Require a parent receiving assistance and not exempt from work requirements to participate in community service if not employed; and

• Meet an all-families and a two-parent-family work participation rate requirement, there being a fiscal penalty for failing to meet the requirement. The work required of single-parent families with children under age 6 is 20 hours per week, while the work required of two-parent families is 35 hours per week4.


“Increasing employment and earnings of needy families . . . [and] decreasing . . . child poverty” are two explicit objectives of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA).5 However, imposing a stringent work

3 In order to maximize the coverage of children considered in this report, statistics on children in families calculated by the authors include children in unrelated subfamilies, and children who are primary and secondary individuals. This is broader than the Census Bureau’s definition of “related children” which excludes these children.

4 Unless the two-parent family receives federally funded child care, in which case the work required is 55 hours per week.

5 U.S. Congress, Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Sec. 411.

requirement does not guarantee that a family will escape poverty. In 2000, nearly 1 million householders living in families with related children had earnings below the official poverty threshold for a family of four, even though the householder worked full- time, full-year (U.S. Census Bureau, 2001c).


Overview of the study
As we found in our earlier report, a substantial number of children continued to remain poor in spite of considerable work effort by their parents during this new welfare reform era (Wertheimer, 1999). In this study, we continue our exploration of the relationship between the poverty rates of children and the work behavior of their parent(s), thus, updating and extending the research we originally undertook as welfare reform was getting underway. By doing so, we can continue to learn about both the prospects and limitations for reducing poverty among children by increasing the work effort of parents.
Parental work effort
One way of exploring this relationship is to compare the poverty rates of children whose parent(s) make a substantial work effort with those whose parent do not make a substantial work effort. For children living in married-couple families, we previously found a very strong relationship between parental work and poverty because even low- wage parents can escape poverty if they work enough hours.6 In contrast, for children living in single-parent families, even a full-time, full-year work effort may not be enough to allow children to escape poverty if the parent is able to secure only low-wage employment. To explore the magnitude of this problem, we calculate the percentage of poor children living in families where the parent(s) makes a substantial work effort over the period 1995-2000.
Earnings
Because increased work effort alone may not be enough to allow children to escape poverty (especially in single-parent families), we, as before, also explore the relationship between the poverty rate of children and the earnings capacity of the higher-earning parent in their family. Clearly, increased work effort will be far more effective at reducing child poverty if the parent’s wage rate exceeds the legal minimum. Since the parental wage rate is likely to be related to that parent’s educational attainment, we also explore the relationship between the child poverty rate and parental education for those children whose parent(s) already make a substantial work effort.
Defining poverty
The discussion so far has implicitly assumed that the official poverty threshold was an appropriate measure of a minimally acceptable level of well-being for children. However, the official poverty threshold fails to take into account several key factors


6 Since September 1, 1997, full-time, full-year earnings (3,640 hours per year) for two people earning the minimum wage ($5.15 per hour) has been $18,746. The poverty line for a family with two adults and two children was $17,463 in 2000.

influencing the well-being of children. Two of the most important of these factors are work-related expenses (especially child-care expenses) and health insurance coverage.


Child care expenses. If single parents of young children living in poor families are not working at all or working relatively little, the availability and cost of adequate child care may not be an important issue. Similarly if a married-couple family with children includes only one worker, child-care expenses may not be an important issue. However, if these parents are required to increase their work effort substantially, they are likely to have to make some kind of arrangement for the care of their children. If relatives are available and willing to care for the children, this may not impose a financial burden on the family. However, if paid child care is required, the family faces an extra, mandatory expense that would not have been necessary if their work effort had not increased.7 To explore this issue, we calculate the percentage of children in poor families whose working parents pay for child care and the ratio of those expenses to family income.
Health insurance. Historically, children whose families received public assistance were also eligible for health insurance from the Medicaid program. However, children in poor families whose parent rely primarily on earnings rather than assistance were not necessarily covered by either Medicaid or by health insurance provided by an employer. Recently, though, eligibility for Medicaid has been broadened, and states are beginning

to offer their own health insurance programs with federal assistance from the State Child Health Insurance Program (SCHIP). To explore the consequences of these changes, we contrast the percentage of children covered by health insurance in working poor families with the coverage rate of children in poor families not making a substantial work effort and other more prosperous working families over time.


External circumstances
Another assumption implicit in the discussion above is that the level of work effort of the parents is under their own control. In a rapidly growing economy with a low rate of unemployment, it is relatively easy for a person with little education and few skills to obtain employment—at least at a low wage job. However, in a slowly growing economy or in an economy in recession, it may be much more difficult to find and keep a job and

to work as many hours as the employee wishes. In our previous study we examined the relationship between work and poverty over the period 1987 through 1994, which includes a mild recession. While an extension of this study to incorporate the recent recession would be a useful addition to our state of knowledge, the current study is focused on a period (1995-2000) in which the economy was constantly expanding, and, thus, is not helpful in exploring this particular issue.




7 Other work-related expenses include payroll and income taxes on earnings and commuting expenses.

Key questions addressed in this report


1. To what extent was being in a working family associated with a reduced likelihood of poverty for a child during the late 1990s into 2000?
Welfare benefits have been set in all states at levels below the poverty threshold. In

2000, the annual maximum TANF benefit for a one-parent family of three persons in the median state was $5,052 per year, which was only 36 percent of the poverty line. If Food Stamp benefits are added, the combined benefits in the median state were $9,072 per

year, which was 65 percent of the poverty threshold (U.S. House of Representatives,

2000). Even in the most generous state (Alaska), combined TANF and Food Stamp benefits were only 116 percent of the official poverty threshold (Food Stamps aren’t included as apart of income when determining if a family is poor).8 Consequently, nearly all families who rely solely on TANF and Food Stamps alone are poor—even if Food Stamp benefits are counted as part of income.


Working can dramatically increase the likelihood that a family escapes poverty. For example, a single parent with three children who was paid $8.50 per hour in 2000 for a

40 hour week, 52 weeks per year, would have earned $17,680 per year, which exceeded the 2000 poverty threshold for a family of four, which was $17,524.


Our previous study established a strong connection between increased work, even at low wages, and a reduced risk of poverty. To update this analysis, we estimate the difference in the poverty rate for children in working families versus children in non-working families and how that difference varies depending upon family structure, parental education, and other key variables over the period 1995-2000. This update is important because the post-welfare reform influx of persons previously on welfare into the labor force could have changed the composition of working poor families.
2. How common was it for children in poor families to have working parents during the 1995-2000 period?
Although working may reduce the likelihood of poverty, as we have seen in our previous study, working is no guarantee of escaping poverty. We will estimate over the 1995-

2000 period the percentage of children in poor families where the parents make a substantial work effort and how that percentage varies depending on family structure, parental education, and other key variables.


3. How were working poor families different from poor families not meeting the working standard during the years 1995-2000?
Our previous study found that working poor families enjoyed some advantages (e.g., parental education) but also some disadvantages (e.g., health insurance coverage) when compared with poor families not meeting the work standard. We update our estimates of differences between working poor families and poor families not meeting the working


8 However, the official poverty threshold does not take into account differences in the cost of living. The relatively high TANF and Food Stamp benefits in Alaska reflect the judgment of the state and federal governments, respectively, that the cost of living in Alaska is exceptionally high compared with other states.

standard across a variety of dimensions including family structure, parental education, health insurance coverage and home and car ownership.


4. How were working poor families different from working, near-poor and more affluent families in the years 1995-2000?
Since the ultimate goal of welfare reform is to increase the likelihood of families escaping poverty through work, we also look at the differences between working poor families and working families that are near-poor (with incomes between 100 percent and

200 percent of the poverty threshold) and more affluent families (with incomes at least



200 percent of the poverty threshold). We estimate differences between working poor families and these better-off working families across a variety of dimensions including family structure, parental education, and average hourly wage.
Questions 1 through 4 are addressed in Chapter 4 of the report.
5. How common were movements into and out of working poverty during the 1990s?
Examining the movements of children into and out of poverty is important for at least two reasons. First, long-term poverty has been shown to have more negative effects on child development than a brief exposure to poverty (Duncan & Brooks-Gunn, 1997). Second, looking at movements into and out of working poverty focuses attention on the fact that moving into the labor force or moving out of poverty is not a one-way street.
6. What were the changes in work behavior and economic resources of children in TANF families, children in poor families not meeting the work standard, children in working poor families, children in working, near-poor families, and more affluent working families between 1996 and 1997?
The answer to this question will indicate whether changes in work behavior of parents and economic resources accompanied federal welfare reform.
Questions 5 and 6 are addressed in Chapter 5.
7. How sensitive are our findings to the definition of poverty?
Examining the sensitivity of our results to the definition of poverty is important because the official definition of poverty has some serious limitations for a study focused on the working poor. First, the official definition of poverty includes in a family’s economic resources only money income before taxes. Thus, the official definition excludes non- cash transfers such as Food Stamps and the refundable portion of the Earned Income Tax Credit (EITC). Second, the official definition of poverty does not deduct from a family’s economic resources work-related expenses such as payroll and income taxes and child- care expenses.
Low-income families relying primarily on public assistance are more likely than other families with children to be receiving non-cash transfers. They are also less likely to have significant work-related expenses since their earnings and hours of work are either limited or zero. In contrast, low-income families relying primarily on earnings are less likely to be receiving non-cash transfers and more likely to be paying payroll taxes and other work-related expenses such as child care and commuting costs. On the other hand,

low-income families relying primarily on earnings are more likely to qualify for the EITC, which has been made increasingly generous over the past ten years. Thus, they may receive a refundable tax credit from the federal government which, in recent years, may be a significant source of income not counted by the official definition of poverty.


In short, there are federal tax and transfer programs that are a significant economic resource to low-income families and that affect families relying primarily on public assistance quite differently than they affect families relying primarily on earnings. Yet, the official definition of poverty fails to take these programs into account.
The analysis in Chapter 6 explores how the findings of the study are affected by taking some of the most important of these programs into account.
8. Compared with children in other families, how were children in working poor families developing shortly after welfare reform was implemented? Similarly, how are children in families leaving welfare developing compared with families staying on welfare?
The 1996 Survey of Income and Program Participation (SIPP) contains (1) variables that measure how well a child is developing and (2) variables that have been shown to be associated with how well a child is likely to develop. The first section of this chapter uses these measures to contrast how well children in working poor families are developing compared with (1) children in poor families not meeting the work standard, (2) children in working families with incomes between 100 percent and 200 percent of the poverty line, and (3) children in more affluent working families (over 200 percent of the poverty line). This analysis, included in Chapter 7, permits us to assess whether children in working poor families are likely to develop differently from other children.
The second section of Chapter 7 performs an analogous set of comparisons for children in families who left welfare between 1996 and 1997 compared with children in families who stayed on welfare both years.
Previous Studies
We provide a thorough review of the literature in a companion paper (Croan, Hatcher, Long, & Wertheimer, 2002). Here we provide the conclusion of that review.
While the research literature on the working poor has substantially advanced in many areas over the past few years, serious definitional problems remain surrounding both poverty and work. While the official poverty threshold and its associated measurements of resources have been justifiably challenged, the federal government has not changed its definition of poverty, although it has produced statistics using many alternative poverty definitions. This profusion of measures continues to hinder both comparisons across studies and the development of indicators measures consistently over time. Definitions of work range just as widely as definitions of poverty. This matters because some definitions have set the work standard so low as to include nearly everyone who works at all, while some definitions exclude people (especially single mothers) who are making a substantial work effort even though they are not working full-time, full-year. The Child Trends definition picks a middle ground that is tied to the expectations of policy-makers at the time that federal

welfare reform was originally enacted and, thus, provides a standard with policy significance.


On the other hand, there has been substantial progress in identifying the reasons why the working poor remain poor. Adults in working poor families generally have low wage rates and often do not work full-time and year-round, although they often would prefer to do so. The working poor are also less well-educated than more prosperous adults and tend to work in lower-skill occupations. Finally, the working poor are less likely to be married and, thus, less able to pool two earnings streams.
There is substantial movement into and out of poverty. These movements are associated not just with changes in eligibility for government programs but also with changes in personal circumstances, such as divorce or a job change.
Both poverty and work behavior are strongly influenced by the strengths of the economy, although low-wage workers seemed to benefit less from the economic boom of the 1990s than higher-wage workers.
The earned income tax credit has become the program most successful at moving working families out of poverty. The effect of welfare reform is much less clear. It has been noted that the apparent success of welfare reform may be reliant on the expanding economy which accompanied it and might not be sustained—or even reversed—in a less favorable economic climate. Food stamps have the potential to help working poor families, but welfare reform appears to have led to a decline in already low participation rates.
The effect of all of the changes accompanying welfare reform on families and children is not clear. While working poor families and their children generally are somewhat better off than poor families not making a substantial work effort (e.g., somewhat better educated, more likely to have two potential earners), they are also less likely to be covered by health insurance or to be benefiting from food stamps. They also are at a substantial disadvantage compared with more prosperous working families. In addition to being less educated, they are less likely to have their children in formal child care arrangements and pay a larger share of their incomes for child care services.
There is relatively little information on how welfare reform may have affected children in working poor families. Welfare-to-work experiments find mixed effects on children with some apparent benefits to younger children and some adverse effects on teenagers.9 More recent data find no substantial difference in child well-being measures between welfare leavers (many of whom are still poor) and children in families currently on welfare. However, these data include neither the many working poor families who have not left welfare nor the working poor families who have

never been on welfare.



9 These findings are discussed in more detail in Chapter 8.

Finally, there is a need for studies focused on the effects of the recent economic slowdown. It is still not clear that the reformed welfare system as it currently is structured can handle the strains that would be caused by a serious recession.


In short, the biggest remaining gaps in the literature are a lack of measures defined consistently and available over time, the relatively little research focused on the well- being of all working poor children, and the lack of studies focused on welfare reform in a period of economic slowdown.

Chapter 2. Defining working poor families
In defining the working poor there are two key issues:

• How to measure poverty; and



How to measure work.

In a recent review of the literature (Croan et al., 2002), we explored these issues in depth, and, therefore, we will only briefly summarize our findings here.


Measuring poverty
As described in detail in our review, most of the studies focused on the measurement of poverty have taken as a point of departure the landmark study carried out by the National Research Council (Citro & Michael, 1995). This study explored many weaknesses of the official definition of poverty and pointed out that the current definition does not reflect:

• increases in the standard of living since the measure was first developed in 1963,



government policies that alter an individuals’ or families’ pretax money income (e.g.,

payroll taxes, in-kind benefit programs, and refundable tax credits),

• increases in medical care costs and benefits,

geographic variations in price,

changes in the composition of families and households (e.g., larger expenditures on

child support, increasing number of households cohabiting or headed by someone 65

or over), and

• changes in work patterns (e.g., child care expenses, transportation expenses).



To address these weaknesses, the NRC Panel made a number of recommendations for revision and proposed a new definition of poverty that would identify the current poor population more accurately. These changes focused on three main areas: the poverty thresholds, the measurement of resources, and data sources.
Since the publication of the NRC report, additional research has been undertaken to operationalize the Panel’s recommendations and propose conceptual or methodological amendments. These studies generally find that although certain manipulations to the poverty measure have the effect of either increasing or decreasing the poverty rate, the trends in poverty rates based on the NRC measures, are similar to the trends produced using the official poverty measure (Croan et al., 2002). The official definition of poverty has been used in Chapters 1-5 and Chapter 7.
To explore the possibility that our findings may be sensitive to the definition of poverty, we have devoted Chapter 6 to an alternative definition of poverty that takes into account the fact that families with working parents are subject to payroll taxes and possibly

federal income taxes that parents not meeting the work standard are more likely to escape and are more likely than parents not meeting the work standard to have child-care expenses. This alternative measure also takes into account the dollar value of Food

Stamp benefits (a non-cash benefit that is a very close substitute for cash), the Earned

Income Tax Credit (EITC), which in recent years has been shown to be an important tool for increasing the income of working families. Finally, we have included in our measure of economic resources the monetary value of several other non-cash transfer programs, including school lunches, low-income energy assistance, and housing assistance.


Because our study is focused on children, and nearly all children are raised in families, we measure poverty on a family basis—just as does the official poverty definition and most of the alternatives that have been proposed.
Measuring work
In addition to conceptual differences in measuring poverty, there have also been substantial differences in the thresholds used in measuring both work and poverty. The original Child Trends study (Wertheimer, 1999) employed a definition of work based on work requirements built into federal welfare reform legislation. This definition differed from most of those in other studies at the time by being based on the work effort of both parents in the family rather than on the work effort of a single person. The threshold was set at 1,820 hours per year for married-couple families and 1,040 hours per year for

single-parent families.


The definition of work in other studies (many based on the work effort of a single adult) has varied dramatically. On the liberal side, studies have identified a person as working if they worked at any time during the past year. On the more stringent side, studies have

counted individuals as working only if they worked full-time (at least 35 hours per week)



all year (at least 50 weeks per year). Other definitions of working have included devoting 27 weeks or more during the past year to working or looking for work, and working at least 27 weeks in the previous year, usually for at least 20 hours per week. Details and citations are provided in our literature review (Croan et al., 2002).
Since poverty is defined based on annual income, we are measuring work effort as total hours worked per year. Because our study is focused on children, it is desirable that we use a family-based definition of work effort. Consequently, for two-parent families, we add together the hours worked per year by both parents.
Setting the threshold for counting the family as a working family is a normative decision (which may account for the wide range of definitions in the literature). Since the working poor have become an object of increasing attention due primarily to welfare reform and its focus on encouraging increased work effort, we continue to base our definition on the work requirement imposed upon states by PRWORA in 1996. The work required of single-parent families with children under age 6 is 20 hours per week,

while the work required of two-parent families is 35 hours per week. We have translated this into an annual work requirement of 1,820 hours per year for two-parent families and

1,040 hours per year for single-parent families. Thus, any family with at least this many hours of work during a calendar year is defined as working.
This definition, thus, has the advantage of being based on the view of the Congress and President as to what constitutes a reasonable work effort and is also consistent with consensus view of the panel of experts convened by the Foundation for Child Development.

Chapter 3. Data and Methods10

Data: Survey of Income and Program Participation and the Current



Population Survey
We have chosen the 1996 panel of the Survey of Income and Program Participation (SIPP) as one primary data source for this study. SIPP has limited longitudinal data, which allows us to examine the dynamics of working poverty and also includes detailed information on child care and factors influencing the well-being of children. However, SIPP is several years out of date.
In an effort to provide more current information, we have also used the March Current Population Survey (CPS) for March 1996 through March 2001. This chapter describes some of the key features, the strengths, and the limitations of both SIPP and the CPS.
Both SIPP and the CPS are designed to produce estimates that are representative of the entire U.S. population. In an attempt to get more accurate estimates of both income and participation in government programs, members of each SIPP panel are interviewed once every four months for at least one year and, for most SIPP panels, for at least two calendar years. At each interview, highly detailed information about employment, income, and program participation is gathered for each of the four preceding months for each member of the household unit. These monthly longitudinal data are supplemented with data collected once or twice per panel on specific topics including child care, asset ownership, and factors measuring or relating to child well-being.
In contrast, the CPS, in its March supplement, collects data on income by source and annual weeks of employment for the preceding calendar year, and collects both demographic information and current employment information for the week of the survey. Hence, information on family composition and data on work and employment focus on different time periods. Moreover, the CPS requires respondents to recall information from a more distant past.
SIPP’s greatest strength is its intensive effort to obtain detailed income and employment data on a monthly basis from successive interviews four months apart. This should increase the likelihood of comprehensive and accurate recall of this critical information.
On the other hand, the sample size of SIPP is smaller than the sample of the CPS. This leads to larger standard errors than those obtained from the CPS. In addition, the longitudinal design of SIPP makes it vulnerable to attrition bias. Members of the sample are lost with each successive wave of interviews, and there is evidence that the sample members lost may differ in systematic ways from the sample members who are retained (Huggins & Winters, 1995). More specifically, the retained sample members have higher incomes in their first interview than the sample members who are not interviewed in subsequent waves.
Certain design features of SIPP make explicit measurement problems which are only implicit in an essentially cross-section survey such as the CPS. For example, because

10 This chapter is an updated version of a chapter originally appearing in our earlier study (Wertheimer, 1999)

members of each household are interviewed three times per year, it is possible to learn about changes in family composition that occur over the course of a year, which is impossible to detect in the CPS. For example, if a husband and wife divorce during the course of a year, one family can split into two families.
Because families and households are not necessarily stable over time, SIPP is organized around persons and, unlike the CPS, is not organized in a hierarchical structure with household, family, and person records. Connections with other members of the family and household are maintained by a set of cross-references that are free to change from one interview to the next. Information collected at the family level (e.g., family income) is, nevertheless, recorded as a person-level variable.
Like families, persons can also appear and disappear. A baby born in February 1996 is not present during the first set of interviews for the 1996 SIPP panel but is present subsequently. Persons who are present for the first interview can disappear at any time during the period covered by the SIPP interviews either because they have died, moved, or because interviewers have been unable to obtain information about them in subsequent interviews.
Finally, even though persons are interviewed every four months, for many questions, the interviewees are asked for monthly data. Thus, detailed income and employment data are recorded on a monthly basis for all 12 months in a calendar year.
We discuss next how we handled these complexities.
Methods
We have designed our analysis to address the issues raised in Chapter 1. Thus, the cross- tabulations of SIPP and the CPS presented in this report consist of the percentages of various segments of the population with specific combinations of characteristics, e.g., among children living in families below the poverty line, the percentage whose parent(s) meet the TANF work standard, by family structure.
Because the primary focus of this study is on children, the basic unit of analysis of the study is the child rather than the family. This approach is also consistent with SIPP’s organization around persons rather than families. Thus, rather than tabulate the number of working poor families, we tabulate the number of children living in working poor families.
For our SIPP analysis, the persons included in our sample universe are those who were living as of the first month of interviews in 1996 and for whom there is complete information for an entire 12-month period. We use a corresponding sample weight constructed by the Census Bureau for the purpose of making the sample representative of the U.S. population for that year. We classify a person as a child if they were less than

18 years old as of the first month of the 12-month period.


For each year’s CPS analysis, the children included in our sample universe are those who are at least one year old and less than 19 years old as of the survey date (March). This is a close approximation to those who were ages zero to 18 in the previous year—the year for which the income and employment data were collected.

As stated earlier, information on annual family income (which is required for determining if a family is above or below the poverty threshold) is reported directly in the CPS. Poverty status is calculated by comparing annual family income with the poverty threshold for the family as structured at the time of the March interview and taking into account family size.


With SIPP calculating family income and poverty status is more complex. To calculate family income for a child using SIPP, we add up the child’s family income for all 12 calendar months, regardless of changes in family structure. For example, if a child’s mother and father separated and formed separate households at the end of June and the child resided with the mother only thereafter, only the income of the mother and other members of that newly defined family would be included for July through December. However, income of both parents would be counted for January through June.
To calculate the poverty status of the child, we compare the annual family income with the average poverty threshold over the course the year. This calculation takes into account any changes in the threshold associated with changes in the size of the family over the course of the year.
Using a technique suggested by the Bureau of the Census, we have estimated standard errors for all tabulated statistics. The Census technique takes into account the fact that both SIPP and the CPS use a clustered sample design that produces less variation across observations than a purely random sample would have produced. Thus, the estimates of the standard errors produced by this technique are higher than would have been produced had we used an estimation procedure that assumed a purely random sample.
These standard errors can be used to test the statistical significance of differences in percentages between two groups. For example, we can determine if the difference between the poverty rate of children in single-mother and two-parent working families is statistically significant. Generally, whenever we draw attention to a difference in percentages in our discussion, the difference is statistically significant at the .05 level or better.

Chapter 4. Results for 1995-2000 using the Current Population



Survey (CPS)
The Current Population Survey (CPS) provides the most up-to-date data set for tracking children in working poor families, children in poor families not meeting the work standard, children in working families with incomes between 100 percent and 200 percent of the poverty line, and children in working families with incomes above 200 percent of the poverty line. With the CPS, we have created continuous time series for these groups for the period 1995-2000, which encompasses the period in which federal welfare reform was implemented.
However, the CPS has several drawbacks when compared with the Survey of Income and Program Participation (SIPP). Because of the greater attention paid to collecting income information in SIPP and the fact that annual income is measured month by month and collected every four months, income is better reported in SIPP than in the CPS. Consequently, the poverty rate as estimated by SIPP is lower than the poverty rate estimated by the CPS.
Similarly, information on employment is collected in far greater detail in SIPP than in the CPS, and data are collected on a monthly basis based on interviews conducted every four months. Consequently, the employment information in SIPP is probably more accurate than the employment data collected in the CPS.
Finally, the CPS collects no data on use of child care services and their cost—the most important component of work-related expenses. We have dealt with this problem by using statistical relationships estimated using data from SIPP to impute use of paid child care and child care costs for each CPS family. In addition, the Census Bureau has imputed family-level estimates of the value of housing subsidies, school lunch, energy assistance, federal income and payroll taxes, and receipt of Earned Income Tax Credit (EITC) refundable credits.
In short, each survey has different strengths, and results in this report based on the CPS are not entirely comparable with similar analyses carried out elsewhere in this report using data from SIPP.
How many children live in working poor families?
Prior to federal welfare reform, in 1995 and 1996 about 4.5 million children lived in working poor families. During the first two years of welfare reform, this number increased—to 4.7 million in 1997 and 5.0 million in 1998. However, the number of children in working poor families decreased to 4.7 million in both 1999 and 2000. The number of children in poor families not meeting the work standard fell steadily over the entire period from 9.5 million in 1995 to 6.4 million in 2000.
This pattern presumably reflects the combined effects of a strong economy and federal welfare reform—both of which encouraged and facilitated greater work effort by persons with relatively low earning potential. As some families escaped the ranks of the working poor through either increased work effort or higher wages, they were replaced by other families who increased their work effort enough to meet the work standard but did not

earn enough to escape poverty. The dynamics of this process in the critical 1996-1997 period are explored elsewhere in this report using data from the Survey of Income and Program Participation.


To what extent does being in a working family reduce the likelihood of poverty for a child?
As shown in Figure 4.1, according to data from successive Current Population Surveys, the percentage of children in families whose incomes were below the official poverty

Figure 4.1 Among all children, percentage living in poor families: all families, families meeting the work standard, and families not meeting the work standard, 1995-2000


70%

60%


50%

62% 62% 62%
58%

55%




54%



40%
Percent


30%



20%

20% 20% 19% 18%
16% 15%

10%
8% 8% 8% 9% 8% 8%


0%

1995 1996 1997 1998 1999 2000


All families Working Not meeting work std

threshold has declined from about 20 percent in 1995 and 1996 to 15 percent in 2000.11

During this same period, among children living in families meeting the working threshold, the percentage who were poor remained relatively constant in the range of 8-9 percent. In contrast, among children in families not meeting the working threshold, about

62 percent were poor in 1995-1997 with a substantial decline thereafter to 54 percent in



2000.
11 As noted earlier, unlike the Census Bureau, we include in this calculation children in subfamilies and primary and secondary individuals who are less than 18 years old.

Thus, between 1995 and 1997, children living in families not meeting the working threshold were 7.6 times as likely to be poor as children living in working families. However, by 2000, this ratio dropped to 6.9—still a very large difference.
Differences by family structure
Working also reduces the likelihood of poverty when children in married-couple families are considered separately from those in single-mother families. Between 1996 and 2000, among children living in married-couple families that met the work standard, only about

Table 4.1. Number of children (thousands) living in poor families and as a percentage of all children by family structure and whether hours worked by parents met the work standard, 1995-2000



1995 1996 1997 1998 1999 2000


All families

14,010

13,820

13,590

12,970

11,750

11,090

Working

4,462

4,474

4,687

5,029

4,696

4,722

Not meeting work std

9,545

9,350

8,903

7,941

7,054

6,372

Married-couple families

4,514

4,444

4,290

4,166

3,922

3,781

Working

2,352

2,344

2,398

2,573

2,311

2,404

Not meeting work std

2,162

2,100

1,892

1,593

1,611

1,377

Single-mother families 7,719 7,410 7,437 7,113 6,129 5,629

Working

1,867

1,833

2,012

2,212

2,170

2,116

Not meeting work std

5,852

5,577

5,425

4,901

3,959

3,513







1995

1996

1997

1998

1999

2000

All families

20%

20%

19%

18%

16%

15%

Working

8%

8%

8%

9%

8%

8%

Not meeting work std

62%

62%

62%

58%

55%

54%

Married-couple families

9%

9%

9%

9%

8%

8%

Working

5%

5%

5%

6%

5%

5%

Not meeting work std

52%

52%

53%

48%

51%

50%

Single-mother families 46% 45% 45% 42% 38% 36%

Working

21%

20%

22%

22%

21%

20%

Not meeting work std

74%

74%

74%

73%

68%

68%

5 percent were poor. In contrast, among children in married-couple families not meeting the work standard, the percentage who were poor ranged between 48 percent and 52 during the 1995-2000 period.


Among children in single-mother families meeting the working standard, between 20 and

22 percent were poor over this same period. In contrast, among children in single-mother families not meeting the working standard, the percentage who were poor was much higher—although it decreased from 74 percent in 1995 to 68 percent in 2000.

Differences by race/ethnicity
As shown in Table 4.2, black non-Hispanic and Hispanic children have the highest likelihood of poverty within each working status category. Nevertheless, working

Table 4.2. Percentage of children living in poor families, by race/ethnicity and whether hours worked by parent(s) met the work standard, 1995-1998






1995


1996


1997


1998


1999


2000


White, non-Hispanic



















Working

5%

4%

5%

5%

4%

4%

Not meeting work standard

50%

52%

53%

49%

47%

47%

Black, non-Hispanic



















Working

16%

16%

15%

18%

16%

15%

Not meeting work standard

73%

71%

67%

63%

66%

63%

Asian, non-Hispanic

Working

7%

5%

8%

7%

6%

7%

Not meeting work standard

56%

63%

63%

55%

41%

49%

Hispanic




Working

21%

22%

20%

20%

18%

17%




Not meeting work standard

70%

71%

71%

69%

59%

55%

Source: March Current Population Survey, 1996-2001.

reduces the likelihood of poverty regardless of children’s race or ethnicity. However, the reduction is smaller for black and Hispanic children.


For white, non-Hispanic children, the likelihood of poverty is about 10 times as high for families not meeting the working standard (about 47 percent in 2000) as for working families (about 4 percent in 2000). For Asian children the ratio is about 7 to 1 (49 percent and 7 percent, respectively). For non-Hispanic black children, the likelihood of poverty

is about four times as high for families not meeting the working standard (about 63 percent) as it is for working families (about 15 percent). For Hispanic children, the likelihood of poverty is about three times as high for families not meeting the working standard (about 55 percent) as it is for families not meeting the working standard (about

17 percent).
Differences by parental education
As shown in Table 4.3, higher levels of parental education are associated with lower rates of child poverty, within each family structure and working status category of the parents. In 2000, among children living in married-couple families meeting the working standard, the likelihood of being poor is 27 percent if the better educated parent has completed fewer than 12 years of education but only 2 percent if the better educated parent has completed at least 16 years of education. Among children living in married-couple families not meeting the working standard, the likelihood of being poor is higher, but a higher level of education is still associated with a lower likelihood of being poor.

Table 4.3. Percentage of children living in poor families, by whether the hours worked by parent(s) met the work standard, family structure, and



the educational attainment of the better educated parent, 1995-2000


1995


1996


1997


1998


1999


2000


Working

Married-couple

Less than 12 years 29% 30% 27% 25% 27% 27%

12 years 7% 7% 8% 9% 8% 7%

13-15 years 4% 3% 4% 4% 3% 3% At least 16 years 1% 1% 1% 1% 1% 2% Single-mother

Less than 12 years 46% 45% 45% 45% 46% 48%

12 years 22% 24% 28% 27% 24% 24%

13-15 years 15% 13% 13% 14% 14% 11% At least 16 years 4% 5% 4% 3% 3% 2%



Not meeting work standard

Married-couple

Less than 12 years

68%

68%

72%

66%

66%

65%

12 years

51%

54%

55%

55%

56%

57%

13-15 years

45%

46%

44%

41%

48%

46%

At least 16 years

34%

33%

33%

19%

26%

27%

Single-mother

Less than 12 years

83%

83%

83%

83%

76%

74%

12 years

75%

73%

74%

74%

70%

70%

13-15 years

65%

65%

66%

61%

59%

59%

At least 16 years

45%

42%

38%

44%

39%

45%

Source: March Current Population Survey, 1996-2001.

Similarly, among children living in single-mother families meeting the working standard, the likelihood of being poor in 2000 is 48 percent if the mother has completed fewer than

12 years of education but only 2 percent if she has completed at least 16 years of education. Among children living in single-mother families not meeting the working standard, the likelihood of being poor is higher, but a higher level education is still associated with a lower likelihood of being poor.
Within education and family structure categories, working still reduces the likelihood of being poor. For example, for children in married-couple families in which the better educated parent had completed fewer than 12 years of education, the likelihood of poverty was 27 percent in 2000 for working families but 65 percent for families not meeting the working standard. For children in married-couple families in which the better educated parent had completed 16 years of education, the likelihood of poverty was 2 percent in 2000 for working families but 27 percent for families not meeting the

working standard. There are similar significant differences for children in single-parent families.


How common is it for children in poor families to have working parents?
Even though living in a working family substantially reduces the likelihood of a child being poor, many children live in poor families that meet the working standard. The enactment of federal welfare reform appears to have substantially increased the percentage of poor children who live in working families. As shown in Figure 4.2, in

2000, about 43 percent of all poor children lived in families that met the working

Figure 4.2. Among all children in poor families, percentage whose parent(s) met the work standard, by family structure

70%

60%

50%

52% 53%

56%
62%

59%

64%



40%


Percent

30%


32% 32%

24% 25%

34%


27%
39% 40%

35%
31%



43%
38%


20%


10%

0%

1995 1996 1997 1998 1999 2000


All types of families Married couple families Single mother families

standard—a sharp increase from 1995 in which 32 percent met the work standard. As shown in Table 4.4, among children in poor, married-couple families, nearly two-thirds (64 percent) lived in families that met the working standard—also sharply up from 1995 in which 52 percent met the work standard. Among children in poor, single-parent families, a significantly smaller percentage (38 percent) lived in working families. However, this too represents a substantial increase from 1995 in which only 24 percent lived in working families.



Differences by race/ethnicity
As shown in Table 4.4, between 1995 and 2000, among children in poor single-mother families the percentage who met the work standard increased sharply for all racial/ethnic groups. However, in 2000, children in poor, single-mother, black, non-Hispanic families were significantly more likely to live in working families (41 percent, respectively) than poor, single-mother Hispanic children (31 percent). The difference in the percentage meeting the work standard for white and Hispanic children in single-mother families was not statistically significant.

Table 4.4. Among children living in poor families, the percentage whose parent(s)



met the work standard, by race/ethnicity and family structure, 1995-2000


All races/ethnicities



1995



1996



1997



1998





1999



2000

All types of families

32%

32%

34%

39%

40%

43%

Married couple families

52%

53%

56%

62%

59%

64%

Single mother families

24%

25%

27%

31%

35%

38%

White non-Hispanic



















All types of families

36%

33%

36%

41%

41%

44%

Married couple families

51%

51%

54%

62%

58%

62%

Single mother families

27%

27%

28%

33%

35%

38%

Black non-Hispanic

All types of families

23%

26%

28%

32%

34%

36%

Married couple families

42%

51%

49%

63%

47%

54%

Single mother families

24%

25%

29%

32%

37%

41%

Asian non-Hispanic

All types of families

27%

21%

33%

33%

44%

44%

Hispanic




All types of families

37%

40%

41%

44%

46%

48%




Married couple families

60%

61%

62%

66%

65%

71%




Single mother families

20%

22%

25%

28%

33%

31%

Source: March Current Population Survey, 1996-2001.

However, 71 percent of children in poor, married-couple, Hispanic families had parents who met the working standard—significantly higher than for poor, white non-Hispanic children (62 percent) and poor, black non-Hispanic children (54 percent). As with single-parent families, there were sharp increases in the percentage of poor married- couple families who met the work standard, except for blacks where the apparent 1995-

2000 increase is not statistically significant.

Differences by parental education

As shown in Table 4.5, the generalization that children in poor, married-couple families are more likely to live in families meeting the work standard than children in poor, single-parent families continues to hold true within categories of educational attainment of the better educated parent. For example, in 2000, among children whose better- educated parent had fewer than 12 years of education, 70 percent of children in poor, married-couple families had parents meeting the working standard, compared with 36 percent of children in poor, single-mother families.
There is no consistent association between educational attainment and the likelihood that a poor parent meets the work standard.
How are the circumstances of children in working poor families different from those in families not meeting the working standard?
The two leftmost columns of Tables 4.6a and 4.6b address this question.
As shown in Table 4.6a, in 2000, children in working poor families are more than twice

as likely (51 percent) than children in poor families not meeting the working standard (22 percent) to live in a married-couple family. Conversely, children in poor families not

Table 4.5. Among children living in poor families, the percentage whose parent(s)

met the work standard, by family structure and parental education, 1995-2000

Single-mother

Source: March Current Population Survey, 1996-2001.




1995

1996

1997

1998

1999

2000

Married-couple

Less than 12 years



51%


57%


54%


58%


62%


70%


12 years

54%

51%

58%

63%

62%

61%

13-15 years

51%

48%

58%

61%

48%

56%

At least 16 years

49%

51%

53%

72%

57%

68%

Less than 12 years


19%

17%

19%

23%

29%

36%


12 years

27%

30%

35%

36%

39%

42%

13-15 years

30%

31%

27%

38%

41%

34%

At least 16 years

26%

31%

37%

22%

27%

21%




meeting the working standard are more likely to live in single-mother families (56 percent versus 45 percent). As shown in Table 4.6b, this pattern also held prior to welfare reform in 1995. However, children in single-mother families made up a

somewhat smaller percentage of poor families not meeting the work standard in 2000 (56 percent) than in 1995 (62 percent)—presumably reflecting the increasing work effort of single mothers over this period.


Patterns by race/ethnicity
In 2000, white non-Hispanic children, who accounted for 65 percent of all children in the U.S., made up 38 percent of children in working poor families and 36 percent of children in poor families not meeting the working standard. Thus, white non-Hispanic children

are underrepresented among both working poor families and poor families not meeting the working standard.


Black non-Hispanic children, who accounted for 16 percent of all children (excluding American Indians) in the U.S. in 2000, made up 24 percent of children in working poor families and 32 percent of children in poor families not meeting the working standard. Thus, black non-Hispanic children were overrepresented among working poor families and even more overrepresented among poor families not meeting the working standard.

Table 4.6a. Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by race/ethnicity, parental education, and average



hourly wage rate, 2000



Family structure

Poor Families Not Meeting the Work Standard


Working Poor

Families

Working

100-200% of poverty

Working

200% of poverty and above


Married-couple 22% 51% 65% 85% Single-mother 56% 45% 30% 11% Single-father 6% 4% 5% 4% Other 15% 0% 0% 0% Race/ethnicity

White non-Hispanic 36% 38% 50% 75%

Black non-Hispanic 32% 24% 19% 10% Asian non-Hispanic 4% 4% 4% 4% Hispanic 25% 32% 25% 10% Family structure and parental education

Married and education less than 12 years 28% 38% 18% 3% Single mother and education less than 12 years 38% 35% 14% 6% Family structure and average hourly wage rate

Married

Less than $5 per hour 16% 39% 10% 3%

$5 - $6.99 per hour 25% 24% 15% 3%

$7 - $9.99 per hour 29% 27% 27% 8% At least $10 per hour 30% 10% 48% 87% Single mother

Less than $5 per hour 28% 42% 6% 2%

$5 - $6.99 per hour 26% 32% 15% 5%

$7 - $9.99 per hour 21% 22% 43% 12% At least $10 per hour 26% 4% 36% 81%

Source: March 2001 Current Population Survey.

Asian, non-Hispanic children, who accounted for 4 percent of all children in the U.S. in

2000, made up 4 percent of children in working poor families and 4 percent of children in poor families not meeting the working standard. Thus, they were proportionally represented among both groups of poor families.
Finally, Hispanic children, who accounted for 16 percent of all children in the U.S. in

2000, made up 32 percent of children in working poor families and 25 percent of children in poor families not meeting the working standard. Thus, they were overrepresented



among poor families not meeting the work standard and even more overrepresented among working poor families.
As shown by comparing Table 4.6a with 4.6b, the racial/ethnic composition of both children in poor families not meeting the work standard and children in poor families meeting the work standard did not change appreciably between 1995 and 2000.
Patterns by family structure and parental education
As shown in Table 4.6b, in 1995, mothers of children in working poor single-mother families were significantly less likely to have completed fewer than 12 years of education (32 percent) than mothers in poor single-mother families not meeting the work standard

Table 4.6b. Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by race/ethnicity, parental education, and average



hourly wage rate, 1995

Family structure



Poor Families Not Meeting the Work Standard
Working Poor

Families

Working

100-200% of poverty

Working

200% of poverty and above


Married-couple 23% 53% 70% 86% Single-mother 62% 42% 25% 11% Single-father 3% 5% 5% 4% Other 12% 0% 0% 0% Race/ethnicity

White non-Hispanic 33% 40% 58% 80% Black non-Hispanic 35% 23% 17% 9% Asian non-Hispanic 4% 3% 3% 4% Hispanic 25% 32% 21% 7% Family structure and parental education

Married and education less than 12 years 39% 38% 16% 2% Single mother and education less than 12 years 44% 32% 17% 5% Family structure and average hourly wage rate

Married

Less than $5 per hour 27% 47% 12% 2%

$5 - $6.99 per hour 28% 33% 17% 4%

$7 - $9.99 per hour 17% 18% 35% 10% At least $10 per hour 27% 3% 36% 83% Single mother

Less than $5 per hour 46% 58% 13% 5%

$5 - $6.99 per hour 22% 30% 28% 7%

$7 - $9.99 per hour 19% 11% 42% 17% At least $10 per hour 12% 1% 18% 71%

Source: March 1996 Current Population Survey.

(44 percent). However, as shown in Table 4.6a, by 2000, while the percentage for working poor single-mothers had not changed significantly (35 percent), the percentage of mothers completing fewer than 12 years of education in poor single-mother families not meeting the work standard dropped to 38 percent. Thus, by 2000 single mothers in working poor families were no longer better educated than those in poor families not meeting the work standard.



Patterns by parental wage rate
Between 1995 and 2000, for both single-mother and married-couple families, the percentage of earners making less than $5.00 per hour12 (in current dollars) dropped substantially for both working poor families and poor families not meeting the work standard. The decreases are especially notable for poor single-mother working families (58 percent in 1995 but 42 percent in 2000) and poor single-mother families not meeting the work standard (46 percent in 1995 but 28 percent in 2000).
However, there is a consistent pattern that a higher percentage of working poor parents has hourly earnings below $5.00 per hours than for poor parents not meeting the work standard. For example, in 2000, the percentage of single mothers meeting the work standard but earning less than $5.00 per hours was 42 percent, compared with 28 percent for her counterpart not meeting the work standard.
Patterns of transfer receipt, health insurance coverage, home ownership, and paid child care utilization
As illustrated by comparing Tables 4.7a and 4.7b, the percentage of children in poor families not meeting the work standard who received AFDC/TANF or food stamps dropped substantially between 1995 and 2000. This is, of course consistent with the dramatic drops in both the TANF and Food Stamp caseloads associated with federal welfare reform. The 1995-2000 change in the percentage of children in working poor families receiving AFDC/TANF is not statistically significant, but the corresponding drop in the percentage receiving food stamps is significant.
In both years, as shown in the two leftmost columns of Tables 4.7a and 4.7b, children in working poor families were significantly less likely than children in poor families not meeting the working standard to receive TANF or Food Stamp benefits regardless of family structure. In 2000, among children in single-mother families, 13 percent of children in working poor families received AFDC at some time during the year, compared with 41 percent of children in poor families not meeting the work standard. For Food Stamps, the comparable recipiency rates are 46 percent and 62 percent, respectively.
Health insurance coverage for children in single-mother working poor families increased from 76 percent to 87 percent between 1995 and 2000. However, the change for married-couple, working poor families is not statistically significant. However, health insurance coverage for children in poor families not meeting the work standard did not increase substantially over the same period. For children in married-couple poor families not meeting the work standard, the percentage was essentially the same in both years (71 percent in 1995 and 73 percent in 2000). For children in single- mother poor families not meeting the work standard, the percentage actually decreased from 90 percent in 1995 to

85 percent in 2000. Thus, by 2000, health insurance rates were similar for children in working poor families and poor families not meeting the work standard.



12 For married-couple families, the higher-earning spouse’s wage rate is used for these calculations.

Table 4.7a. Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by receipt of AFDC and Food Stamps,

health insurance coverage, home ownership, 2000



Married-couple families

Poor Families Not Meeting the Work Standard
Working Poor

Families

Working

100-200% of poverty

Working


200% of poverty and above

Received AFDC 24% 5% 3% 0% Received Food Stamps 47% 25% 8% 0% Health insurance coverage 73% 70% 84% 95% Home ownership 30% 49% 61% 86% Single-mother families

Received AFDC

41%

13%

7%

2%

Received Food Stamps

62%

46%

15%

3%

Health insurance coverage

85%

87%

83%

89%

Home ownership

22%

24%

34%

61%


Table 4.7b. Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by receipt of AFDC and Food Stamps,

health insurance coverage, home ownership, 1995



Married-couple families

Poor Families Not Meeting the Work Standard
Working Poor

Families

Working

100-200% of poverty

Working

200% of poverty and above


Received AFDC 35% 7% 4% 1% Received Food Stamps 55% 34% 11% 1% Health insurance coverage 71% 66% 77% 95% Home ownership 35% 41% 60% 85%

Single-mother families

Received AFDC

69%

23%

10%

4%

Received Food Stamps

80%

51%

22%

6%

Health insurance coverage

90%

76%

79%

90%

Home ownership

15%

21%

34%

62%

Source: March 1996 Current Population Survey.

Note: Data for home ownership is as of 2001.



Home-ownership rates increased significantly between 1995 and 2000 for married- couple, working poor families from 41 percent to 49 percent but did not change significantly for married-couple families not meeting the work standard. Because of the increase for the working poor families, in 2000, children in married-couple working poor families were more likely than children in married-couple poor families not meeting the work standard to live in owner-occupied homes (49 percent versus 30 percent).
The pattern was opposite for single-mother families. Home-ownership increased from 15 percent to 22 percent for single-mother families not meeting the work standard but did

not change significantly for single-mother working poor families. (This could possibly reflect single-mother families moving in with parents who are homeowners.) Because of the increase in ownership for the former group, there was no significant difference in



rates of home ownership between the families of children in working poor, single-mother families (24 percent) and poor, single-mother families not meeting the working standard (22 percent) in 2000—even though there was a significant difference in 1995.
Child care. Unfortunately, the Current Population Survey (CPS) does not collect information on child care expenses. However, the 1996 Survey of Income and Program Participation (SIPP) does—along with many variables which it has in common with the CPS. Thus, in order to be able to estimate how much, if anything, families spent on child care, we imputed child care expenses onto each family’s record using statistical relationships estimated using SIPP. The model is based on a specification developed by the U.S. Census Bureau (Short, Garner, Johnson, & Doyle, 1999).
The first two models predict whether a family incurred child care expenses separately for single-parent and married-couple families. The second two models predict child care expenses for those families that incurred child care expenses—again for single-parent

and married-couple families. Explanatory variables include income, race/ethnicity, number of children by age, region, family income, and the percentage of family income earned by the mother. The specification of the models and the resulting equations used for imputation are described in detail in Appendix A. Because we were not able to estimate child care expenses directly from the CPS, results should be viewed with caution.


As shown in the two leftmost columns of Table 4.8a, only 7 percent of children in married-couple working poor families were in paid child care in 2000. Not surprisingly, this percentage was higher than for children in poor families not meeting the work standard (4 percent). Among children in single-mother working poor families, the percentage was substantially higher—22 percent. This percentage was also higher than the percentage receiving paid child care among children in poor single-mother families not meeting the work standard (7 percent).
Among children in working poor families and poor families not meeting the work standard whose parents paid for child care, child care expenses tended to consume a large percentage of family income. Among children in married-couple, working poor families that paid for child care, around two-thirds paid at least 30 percent of their total family income for child care expenses, and 20 percent reported paying over half of their income. For single-parent, working poor families, the financial burden was even greater. Nearly two-thirds paid 40 percent or more of their income for child care, and 41 percent paid at least half. These percentages are all similar to the corresponding percentages in 1995.

For single-parent, poor families not meeting the work standard and paying for child care, child care expenses also consumed an extraordinary percentage of family income for most families. Over 75 percent spent at least half of their income on child care expenses in 2000. Again, because child care expenses have been imputed, results should be viewed with caution.


The share of income devoted to child care for both working poor families and poor families not meeting the work standard and paying for child care is essentially unchanged from 1995.

How are children in working poor families different from those in other more prosperous working families?


The three rightmost columns of Tables 4.6a and 4.7a address this question.
Patterns by family structure
In 2000, nearly half of all children in working poor families lived in a single-parent family. In contrast, two-thirds of children in working families with incomes between 100

Table 4.8a Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by paid child care status and percentage distribution by amount paid for those paying, 2000






Poor Families Not Meeting the Work Standard



Working Poor

Families

Working



100-200% of poverty

Working

200% of poverty and above


Married-couple families

Percent paying for child care

4%

7%

18%



29%

% income paid for those paying













Less than 10%

na

6%

7%

54%

10-19.9%

na

12%

34%

42%

20-29.9%

na

16%

46%

4%

30-39.9%

na

37%

12%

0%

40-49.9%

na

9%

1%

0%

50% or greater




20%

0%

0%

Single-mother families 7% 22% 27% 23%

% income paid for those paying



Percent of income paid

Less than 10%

4%

0%

1%

22%

10-19.9%

0%

7%

18%

50%

20-29.9%

0%

15%

45%

16%

30-39.9%

12%

13%

29%

4%

40-49.9%

9%

25%

5%

2%

50% or greater

76%

41%

2%

6%

Source: March 2001 Current Population Survey.

percent and 200 percent of the poverty line lived in married-couple families, as did 85 percent of those with incomes above 200 percent of the poverty line.


This pattern also prevailed in 1995. However, in 2000, single-mothers headed a larger share (30 percent) of the working families with incomes between 100 percent and 200 percent of the poverty threshold than in 1995 (25 percent). Conversely, married-couples share of this group declined from 70 percent to 65 percent.
Patterns by race/ethnicity
White children make up an increasingly large share of working families as incomes increase. In 2000, non-Hispanic white children account for 38 percent of children in working poor families, 50 percent of children in working families with incomes between

100 percent and 200 percent of the poverty threshold, and 75 percent of children in working families with incomes over 200 percent of the poverty line.


In contrast, black children account for 24 percent of the working poor, 19 percent of

those with incomes between 100 percent and 200 percent of the poverty line, and only 10 percent of those above 200 percent. For Hispanic children, the corresponding

percentages are 32 percent, 25 percent and 10 percent. Asian children make up 4 percent of all three groups.
Patterns by family structure and parental education
There are substantial differences in the education of the better educated parent of children in working poor families compared with more prosperous working families, regardless of family structure. In 2000, 38 percent of the better educated parents of children in

married-couple, working poor families had not completed 12 years of education. In contrast, only 18 percent of the better educated parents in married-couple, working


Table 4.8b Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by paid child care status and percentage distribution by amount paid for those paying, 1995


Married-couple families

Poor Families Not Meeting the Work Standard


Working Poor

Families

Working

100-200% of poverty

Working

200% of poverty and above


Percent paying for child care

Percent of income paid for those paying

Less than 10%



3%
na

9%
27%

17%
6%

30%
45%

10-19.9%

na

4%

39%

50%

20-29.9%

na

27%

40%

4%

30-39.9%

na

25%

13%

0%

40-49.9%

na

27%

1%

0%

50% or greater




17%

1%

0%

Single-mother families

Percent paying for child care

Percent of income paid

Less than 10%


6%
4%


19%
0%




21%
0%

22%
18%



10-19.9%

2%

0%

13%

60%

20-29.9%

7%

16%

51%

12%

30-39.9%

14%

19%

29%

3%

40-49.9%

14%

21%

5%

1%

50% or greater

58%

43%

3%

5%

Source: March 1996 Current Population Survey.

families with incomes between 100 percent and 200 percent of the poverty line had failed to complete high school. For working married-couple families with incomes over 200 percent, only 3 percent had failed to complete a high school education. For children in

single-mother families, the differences were also substantial—35 percent for working poor mothers versus 14 percent for near-poor single working mothers, and 6 percent for more affluent single working mothers.
Patterns by parental wage rate
There are similar differences for average hourly wage rates. Only 10 percent of children in married-couple, working poor families had a parent who was paid more than $10.00 per hour in 2000. In contrast, nearly half of children in married-couple, working families with incomes between 100 percent and 200 percent of the poverty line, had a parent paid at least $10.00 per hour. For children in married-couple, working families with incomes above 200 percent of the poverty line, the percentage earning at least $10.00 per hour

was 87 percent. Findings are similar for single-mother families.


Patterns by transfer receipt, health insurance coverage, home ownership, and paid child care utilization
Transfer receipt. As would be expected, children in working poor families were much more likely than children in other working families to receive public assistance, as shown in Table 4.7a. This held true regardless of family structure.
Health insurance. In 2000, children in married-couple, working poor families were less likely to be covered by health insurance (70 percent) than working families with incomes between 100 percent and 200 percent of the poverty threshold (84 percent), and more affluent working families (95 percent). However, among single-mother families, coverage rates were not significantly different across all three groups (87 percent, 83 percent, and 89 percent, respectively).
Home ownership. Rates of home-ownership were substantially lower among children in working poor families than among children in other working families, regardless of

family structure.


As noted earlier, only 7 percent of children in married-couple working poor families were in paid child care in 2000. As shown in the two right-most columns of Table 4.8a, 18 percent of children in married-couple families with incomes between 100 and 200

percent of the poverty threshold were in paid child care, and 29 percent of those in more affluent married-couple families were in paid child care.


Child care. Among children in single-mother families, the pattern was less straightforward. Twenty-two percent of children in single-mother working poor families were in paid child care, compared with 27 percent of children in single-mother, near-poor working families, and 23 percent in more affluent single-mother working families. Thus, the children in near-poor working single-mother families are the most likely to be in paid child care.
As noted earlier, among children in working poor families whose parents paid for child care, child care expenses tended to consume a large percentage of family income.

Among children in married-couple, working poor families that paid for child care, around two-thirds paid at least 30 percent of their total family income for child care expenses,

and 20 percent reported paying over half of their income. For single-parent, working

poor families, the financial burden was even greater. Nearly two-thirds paid 40 percent or more of their income for child care, and 41 percent paid at least half their income.


As incomes rise, child care expenses consume a shrinking percentage of family income for both married-couple and single-mother working families. Among married-couple working families with incomes between 100 percent and 200 percent of the poverty threshold, 13 percent paid at least 30 percent of their income for child care. For those with incomes at least twice the poverty line, none did so. Among single-mother working families with incomes above the poverty threshold, only 7-8 percent paid at least 40 percent of their income for child care. However, 36 percent of single-mother working

families with incomes between 100 percent and 200 percent of the poverty threshold paid at least 30 percent—a substantial share.


Summary
Prior to federal welfare reform, in 1995 and 1996, about 4.5 million children lived in working poor families. During the first two years of welfare reform, this number increased—to 4.7 million in 1997 and 5.0 million in 1998. However, the number of children in working poor families decreased to 4.7 million in both 1999 and 2000. The number of children in poor families not meeting the work standard fell steadily over the entire period, from 9.5 million in 1995 to 6.4 million in 2000. This pattern presumably reflects the combined effects of a strong economy and federal welfare reform—both of which encouraged and facilitated greater work effort by persons with relatively low earning potential.
Throughout the 1995-2000 period, being in a working family dramatically reduced the likelihood of poverty for a child. Among children in working families only about 8-9 percent were poor. In contrast, among children in families not meeting the work standard, 63 percent were poor in 1995-1997, with a substantial decline to 54 percent in

2000—presumably because they also worked more hours. Nonetheless, even in 2000, children in families not meeting the work standard were nearly 7 times as likely as children in working families to be poor.


Poverty rates for children varied substantially by family structure, race/ethnicity, and parental education. Nonetheless, meeting the work standard was associated with a substantial reduction in the risk of child poverty across all demographic and educational groups.
Even though living in a working family substantially reduces the likelihood of a child being poor, many children live in poor families that meet the working standard. The enactment of federal welfare reform appears to have substantially increased the percentage of poor children who live in working families. In 2000, about 43 percent of all poor children lived in families that met the working standard—a sharp increase from

1995 in which 32 percent met the work standard, and a pattern which generally holds across racial/ethnic groups.


Compared with children in poor families not meeting the work standard in 2000, children in working poor families are:

• more likely to live with two parents,



about as likely have a parent who has completed 12 years of education,13

less likely to be receiving AFDC/TANF or Food Stamps,

about as likely to be covered by health insurance (an important change since 1995

when the working poor were less likely to be covered),

• more likely to live in a family that owns a home (married-couple families only),14 and

more likely to be in paid child care.

Among children in both working poor families and poor families not meeting the work standard and whose parents paid for child care, child care expenses consumed a much large share of family income than among children in more affluent families.


Compared with children in two categories of more prosperous working families in 2000, children in working poor families are:

• less likely to live with two parents,



less likely to have a parent who has completed 12 years of education,

less likely to have a parent earnings more than $10.00 per hour,

more likely to be receiving AFDC/TANF or Food Stamps,

less likely to be covered by health insurance (married-couple families only),

less likely to live in a family that owns a house, and

less likely to be in paid child care.

Moreover, among children in working poor families with paid child care, child care expenses consumed a much larger share of family income than among children in more affluent working families.



13 This is a change from 1995 when single mothers in working poor families were more likely to have completed 12 years of education.

14 This pattern is the opposite of 1995 when single-mother working poor families were more likely to own their own homes but married-couple working poor families were not.

Chapter 5. The dynamics of poverty and working, 1987-1997
The statistics and discussion presented in Chapter 4 provided snapshots of children in a single year. However, the population of children in poverty is not static. A substantial percentage of one year’s poor children were not poor in the previous year. Conversely, a substantial percentage of children who are poor in the current year will not be in the next.
As stated earlier, an explicit goal of the 1996 welfare reform legislation was to move families and children out of poverty by encouraging work. This chapter has two main purposes. The first is to update our previous analysis of the association for children between having parent(s) who meet the work standard15 and the likelihood of moving into or out of poverty. (For example, we can calculate the proportion of poor children living

in families not meeting the work standard in one year who are no longer poor in the next year and whose families are meeting the work standard in the next year.)


The second purpose is to examine the 1996-1997 changes in work behavior and family economic resources of children living in families who were (1) on TANF in 1996, (2) poor and not meeting the work standard, (3) working poor families, (4) working families with incomes between 100 percent and 200 percent of the poverty threshold, and (5)

more affluent working families. We would expect the families in the first three groups to be likely to change their behavior in response to both welfare reform and the improving economy, while families in the last two groups to change their behavior only in response to the improving economy.
Patterns of movement into and out of poverty
In our previous report, we analyzed the movements of families into and out of poverty by their work status and found that over the period between 1987 and 1994 there was a

rough balance between the numbers of children entering and leaving poverty each year. In this section we update this analysis by including the 1996-1997 transitions.


Recognizing that the poverty transition process is not a one-way street (even for those whose parents work), in addition to calculating the proportion of poor children in working families leaving poverty, we can also calculate the proportion of nonpoor

children living in families meeting the work standard in one year who become poor in the next year in spite meeting the work standard in the second year as well. This

counterflow is a phenomenon that does not receive much popular attention. However, it is just as important as the flow out of poverty in determining the percentage of children living in working poor families in any given year.
The last year presented in this analysis—1996—is especially interesting because it coincides with the implementation of federal welfare reform. For this year’s data, we are observing children who were in poverty in 1996 but were not in 1997 and vice versa.
Movements of children into and out of poverty using data from SIPP are summarized in Table 5.1. There is a rough balance between the number of children entering and leaving poverty in each year. In other words, the poor children moving out of poverty are mostly

15 Full-time work for married-couple families and half-time work for single-parent families.

Table 5.1. Number (thousands) and percentage of poor children leaving poverty next year and number



(thousands) and percentage of nonpoor children becoming poor next year, 1987, 1990-1996



Leaving poverty

1987-88 1988-89 1989-90 1990-91 1991-92 1992-93 1993-94 1996-97



Number 2,876 na na 2,337 2,647 2,619 3,214 2,594

Percentage 24.9% na na 21.0% 20.6% 19.9% 21.0% 26.0% Entering poverty



Number 2,068 na na 2,778 2,598 2,335 2,511 2,283

Percentage 4.0% na na 5.2% 4.8% 4.3% 4.7% 4.9%

Source: Survey of Income of Program Participation.

Comparable data for 1988 and 1989 were not available (na) from SIPP.



offset by nonpoor children moving into poverty. In some years, however, there is a significant imbalance. For example, in 1990-1991, a recession period, the number of children entering poverty was significantly higher than the number leaving poverty. In contrast, in 1993-1994 and 1996-1997, both part of the 1990s period of rapid economic growth, the number of children leaving poverty was significantly higher than the number entering poverty. There is no indication that the 1996-1997 transitions were out of the ordinary when compared with previous periods.
In contrast, the proportions of children moving into and out of poverty are quite different—reflecting the difference in the size of the underlying populations. There are many more nonpoor children than there are poor children. For example, about 26 percent of children who were poor in 1996 were no longer poor in 1997. However, about 5 percent of children who were not poor in 1996 became poor in 1997. Again there is no indication of differences during the 1996-1997 transition compared with previous

periods.
Next, we consider how these movements of children into and out of poverty are affected by the work behavior of their parent(s). We have classified each child into one of four categories:


1. Parent(s) met the working standard neither year;
2. Parent(s) met the working standard in the current year but not the next year;
3. Parent(s) met the working standard in the next year but not the current year; and
4. Parent(s) met the working standard in both years.
Tables 5.2 through 5.5 are divided into two panels. The first panel provides information on poor children leaving poverty the next year. The second panel provides information on nonpoor children entering poverty the next year.
As shown in the top panel of Table 5.2a, among children who were poor in 1996 and whose parents failed to meet the working standard in either 1996 or 1997, 88 percent remained in poverty in 1997, and only 12 percent left poverty. These percentages are similar to those prevailing prior to welfare reform. Clearly, if a family’s goal is to move out of poverty, keeping parental work effort below the work standard is not a good strategy.

Table 5.2a. Number (thousands) and percentage of poor children leaving poverty next

year, by work status of parents in current and next year, 1987, 1990-1993, 1996



Number

1987-88 1990-91 1991-92 1992-93 1993-94 1996-97

Met working standard neither year 1,204 629 861 1,021 937 482

Met working standard in current year but not next 180 243 307 173 313 120

Met working standard next year but not current 619 398 503 438 788 828



Met working standard both years 873 1,067 975 988 1,175 1,165

Percentage



Met working standard neither year 18% 10% 12% 12% 10% 12% Met working standard in current year but not next 20% 23% 22% 17% 26% 13% Met working standard next year but not current 46% 42% 46% 48% 47% 45% Met working standard both years 36% 38% 33% 33% 38% 37%
Table 5.2b. Number (thousands) and percentage of nonpoor children entering poverty next

year, by work status of parents in current and next year, 1987, 1990-1993, 1996



Number

1987-88 1990-91 1991-92 1992-93 1993-94 1996-97

Met working standard neither year 676 578 708 671 511 279

Met working standard in current year but not next 517 957 799 880 900 696

Met working standard next year but not current 54 104 90 59 117 159

Met working standard both years 821 1,139 1,001 725 984 1,149

Percentage

Met working standard neither year 23% 19% 22% 18% 16% 21% Met working standard in current year but not next 9% 15% 14% 14% 15% 20% Met working standard next year but not current 4% 10% 7% 4% 7% 10% Met working standard both years 2% 3% 2% 2% 2% 3%

Source: Survey of Income of Program Participation

In contrast, among children who were poor in 1996 and whose parents did not meet the work standard in 1993 but did meet the work standard in 1997, 45 percent moved out of poverty—a significantly higher percentage. While increasing parental work effort to meet the work standard is a significantly better strategy, it is worth noting that in every year tabulated the strategy was successful less than half the time. For the other half of the children, increasing work effort led the family into the ranks of the working poor.


Among children who were in poor families in 1996 and whose parents met the working standard in both 1996 and 1997, 37 percent moved out of poverty by 1997. While working in both years yields a significantly higher exit rate than working neither year, the fact that fewer than four in ten children exit poverty if their parents meet the working standard for two consecutive years suggests that working poverty can be a persistent

state. Again there is no indication of a shift in this pattern from previous years.


Some children exited poverty between 1996 and 1997 despite the fact that their parent(s) met the work standard in 1996 but not in 1997. This occurred because parental hours worked are not the only determinant of whether a child’s family income falls above or below the poverty threshold. Other factors that could have led to this result include a change in family structure (e.g., marriage), a wage increase, and an increase in unearned income (e.g., SSI, Social Security Disability payments, public assistance, income from assets).
Looking at the likelihood of a nonpoor child entering poverty (Table 5.2b) produces results which are a mirror image of the likelihood of exiting poverty. Children whose

parents work in neither year or who meet the working standard in one year but not the next have a relatively high likelihood of moving into poverty from one year to the next. For example, children whose family income exceeded the poverty threshold in 1996 and whose parents failed to meet the working standard in both years had a 21 percent likelihood of falling into poverty by 1997. Similarly, children whose family income exceeded the poverty threshold in 1996 and whose parents met the working standard in

1996 but did not meet the working standard in 1997 had a 20 percent likelihood of falling into poverty by 1997. For this latter group, the likelihood of falling into poverty is significantly higher than in 1987 and 1992 and bordering on significantly higher when compared with 1991 and 1993. This could possibly reflect reluctance on the part of

needy families to apply for TANF benefits under the newly enacted legislation.
In contrast, nonpoor children whose parent(s) did not meet the working standard in 1996 but did meet the working standard in 1997 had only a 10 percent likelihood of moving into poverty. For nonpoor children whose parents met the working standard in both years, the likelihood was only 3 percent—significantly lower than for those nonpoor children whose parents met the working threshold in 1997 but not in 1996.16
Patterns by family structure
As shown in Table 5.3.a, the pattern of association between parental work behavior and the likelihood of a poor child exiting poverty held for both married-couple and single- mother families. For example, among children in married-couple families who were poor in 1996 and whose parents did not meet the working standard in either 1996 or 1997,

only 13 percent escaped poverty in 1997. In contrast, among children in married-couple families who were poor in 1996 and whose parents did not meet the working standard in



1996 but did meet the working standard in 1997, 50 percent escaped poverty—a significantly higher proportion. The corresponding percentages for children in single- mother families were 11 percent and 47 percent, respectively. This pattern is not substantially different from periods prior to welfare reform.
Children in both poor married-couple and single-mother families also had a significantly higher chance of escaping poverty in the next year if their parent(s) worked both years than if they worked in neither year. This result is also the same as was obtained when all poor children were analyzed together.
It is interesting to note that there was no significant difference by family structure in the likelihood of poor children leaving poverty as a consequence of their parent(s) meeting the working standard in 1997, given that they did not meet the work standard in 1996. This is surprising in light of the fact that married-couple families have the opportunity to work many more hours per year than single-mother families. However, the finding is no different from the 1993-1994 period.

16 This difference was not consistently significant in the other pairs of consecutive years.

Table 5.3a. Number (thousands) and percentage of poor children leaving poverty next year, by



family structure and work status of parents in current and next year, 1987, 1990-1996



Married couple families

Number

1987-88 1990-91 1991-92 1992-93 1993-94 1996-97

Met working standard neither year 505 234 291 471 325 110

Met working standard in current year but not next 115 119 137 105 131 67

Met working standard next year but not current 307 187 183 206 321 333

Met working standard both years 511 709 627 579 715 618

Percentage

Met working standard neither year 24% 15% 15% 20% 12% 13% Met working standard in current year but not next 24% 20% 20% 20% 21% 15% Met working standard next year but not current 46% 40% 36% 53% 49% 50% Met working standard both years 30% 39% 35% 34% 37% 39% Single-mother families

Number



Met working standard neither year 541 192 431 411 425 339

Met working standard in current year but not next 65 123 162 44 141 43

Met working standard next year but not current 226 185 296 144 403 481

Met working standard both years 362 351 286 359 406 420

Percentage

Met working standard neither year 12% 5% 9% 8% 7% 11% Met working standard in current year but not next 17% 29% 25% 9% 27% 11% Met working standard next year but not current 42% 43% 54% 36% 46% 46% Met working standard both years 49% 38% 27% 32% 38% 35% Source: Survey of Income of Program Participation.

As shown in Table 5.3.b, the pattern of association between parental work behavior and the likelihood of a nonpoor child entering poverty also appeared to hold for both married- couple and single-mother families. For example, among children in married-couple families above the poverty threshold in 1996 whose parents met the working threshold in neither 1996 nor 1997, 24 percent entered poverty in 1997. Among children in married- couple families above the poverty threshold in 1996 whose parents did not meet the working threshold in 1996 but did meet the working threshold in 1997, only 7 percent entered poverty. This was also true for most periods in the past.
Once again, regardless of family structure, children living in nonpoor families in 1993 had the lowest likelihood of entering poverty if their families met the working threshold in both 1996 and 1997—2 percent for children in married-couple families above the poverty threshold and 6 percent for children in single-mother families above the poverty threshold. The difference in the likelihood of entering poverty for these children versus children in nonpoor families not meeting the working threshold in both years was statistically significant in all years.
With only a few exceptions, nonpoor children in single-mother families were not significantly more likely than nonpoor children in married-couple families to enter poverty in the following year, regardless of their pattern of work behavior in the two years.

Table 5.3b. Number (thousands) and percentage of nonpoor children entering poverty next year, by



family structure and work status of parents in current and next year, 1987, 1990-1996



Married couple families

Number



1987-88 1990-91 1991-92 1992-93 1993-94 1996-97

Met working standard neither year 303 278 332 362 256 116

Met working standard in current year but not next 293 644 591 660 568 500

Met working standard next year but not current 16 40 28 23 79 67

Met working standard both years 603 918 865 519 728 758

Percentage

Met working standard neither year 20% 17% 22% 20% 18% 24% Met working standard in current year but not next 7% 13% 12% 13% 12% 18% Met working standard next year but not current 3% 9% 4% 3% 11% 7% Met working standard both years 2% 2% 2% 1% 2% 2% Single-mother families

Number



Met working standard neither year 322 131 272 178 159 159

Met working standard in current year but not next 199 233 146 176 267 166

Met working standard next year but not current 5 38 49 15 29 67

Met working standard both years 203 209 130 175 212 330

Percentage

Met working standard neither year 29% 16% 22% 14% 15% 22% Met working standard in current year but not next 18% 20% 16% 16% 23% 29% Met working standard next year but not current 2% 21% 21% 5% 8% 10% Met working standard both years 4% 4% 3% 3% 4% 6% Source: Survey of Income of Program Participation.

Patterns by educational attainment of better educated parent


As shown in Table 5.4.a, the pattern of association between parental work behavior and the likelihood of a poor child exiting poverty held regardless of the educational attainment of the better educated parent. For example, among children in families who were poor in 1996, whose better educated parent had fewer than 12 years of education and whose parents did not meet the working standard in either 1996 or 1997, only 9 percent escaped poverty in 1997. In contrast, among children in families who were poor in 1996 with the same parental education level and whose parents did not meet the working standard in 1996 but did meet the working standard in 1997, 34 percent escaped poverty—a significantly higher proportion. The corresponding percentages for children in families with parental education of 12 years were 12 percent and 49 percent, respectively; for children in families with parental education of at least 13 years, they were 18 percent and 54 percent respectively.
Regardless of parental education, children in both types of families also had a significantly higher chance of escaping poverty in the next year if their parent(s) worked both years than if they worked in neither year. This result is also the same as was obtained when all poor children were analyzed together.
There are indications that parental educational attainment may play a role in the likelihood of a poor child exiting poverty, after controlling for the dynamics of parental work behavior. For example, in 1996-1997, 54 percent of poor children whose parents moved from below to above the working threshold left poverty if their better educated parent had at least 13 years of education, compared with 34 percent of children whose

Table 5.4a. Number (thousands) and percentage of poor children leaving poverty next year, by

parental education and work status of parents in current and next year, 1987, 1990-1996

Less than 12 years



Number

1987-88 1990-91 1991-92 1992-93 1993-94 1996-97

Met working standard neither year 695 349 336 316 373 179

Met working standard in current year but not next 35 45 129 58 112 291

Met working standard next year but not current 232 97 162 117 246 204

Met working standard both years 209 263 221 268 334 275

Percentage

Met working standard neither year 17% 11% 9% 9% 8% 9% Met working standard in current year but not next 11% 10% 27% 16% 27% 9% Met working standard next year but not current 43% 33% 46% 37% 48% 34% Met working standard both years 26% 38% 21% 33% 38% 26%

12 years

Number



Met working standard neither year 302 197 291 450 306 157

Met working standard in current year but not next 87 169 122 54 82 44

Met working standard next year but not current 132 134 192 186 292 326

Met working standard both years 336 440 439 508 413 434

Percentage

Met working standard neither year 14% 9% 13% 15% 10% 12% Met working standard in current year but not next 19% 38% 19% 15% 19% 15% Met working standard next year but not current 31% 36% 39% 47% 44% 49% Met working standard both years 30% 36% 39% 35% 31% 36% At least 13 years

Number



Met working standard neither year 207 83 234 255 259 145

Met working standard in current year but not next nc 29 56 62 120 48

Met working standard next year but not current 255 167 149 134 250 297

Met working standard both years 328 364 315 211 428 456

Percentage

Met working standard neither year 27% 9% 16% 16% 14% 18% Met working standard in current year but not next nc 15% 19% 22% 31% 15% Met working standard next year but not current 69% 60% 60% 67% 50% 54% Met working standard both years 63% 42% 42% 31% 50% 51% Source: Survey of Income of Program Participation.

better-educated parent had than fewer than 12 years. However, this difference was statistically significant in only two of the five years. Results are similar if children whose parents were above the working threshold in both years are compared based on their parents’ educational attainment.


As shown in Table 5.4.b, the pattern of association between parental work behavior and the likelihood of a nonpoor child entering poverty also appeared to hold regardless of parental education. However, because of relatively small sample sizes, standard errors were large enough that even large estimated differences were not necessarily statistically significant.
For example, among children in families above the poverty threshold in 1996 with parental education of 12 years and whose parents met the working threshold in neither

1996 nor 1997, 29 percent entered poverty in 1994. Among children in families above

the poverty threshold in 1996 with the same parental education and whose parents did not meet the working threshold in 1996 but did meet the working threshold in 1997, only 6


Table 5.4b. Number (thousands) and percentage of nonpoor children entering poverty next year, by

parental education and work status of parents in current and next year, 1987, 1990-1996

Less than 12 years

Number

1987-88 1990-91 1991-92 1992-93 1993-94 1996-97


Met working standard neither year 242 215 325 197 142 75

Met working standard in current year but not next 109 256 180 284 100 155

Met working standard next year but not current 7 44 75 27 23 201



Met working standard both years 194 346 79 146 270 1,864

Percentage

Met working standard neither year 26% 22% 30% 20% 16% 24% Met working standard in current year but not next 14% 31% 24% 24% 17% 45% Met working standard next year but not current 2% 12% 14% 7% 6% 19% Met working standard both years 6% 11% 3% 6% 10% 13%

12 years

Number

Met working standard neither year 226 200 182 246 198 136



Met working standard in current year but not next 230 343 375 345 374 236

Met working standard next year but not current 13 6 15 23 49 26



Met working standard both years 401 373 514 326 381 440

Percentage

Met working standard neither year 17% 32% 16% 17% 18% 29% Met working standard in current year but not next 11% 15% 18% 16% 19% 24% Met working standard next year but not current 3% 2% 2% 7% 10% 6% Met working standard both years 3% 3% 4% 3% 3% 5% At least 13 years

Number

Met working standard neither year 208 163 201 228 171 68

Met working standard in current year but not next 177 358 244 251 426 336

Met working standard next year but not current 34 53 0 9 45 86



Met working standard both years 226 420 408 254 333 435

Percentage

Met working standard neither year 29% 16% 20% 19% 15% 13% Met working standard in current year but not next nc 11% 8% 8% 12% 15% Met working standard next year but not current 9% 14% 0% 1% 6% 9% Met working standard both years 1% 2% 1% 1% 1% 2%

Source: Survey of Income of Program Participation.

percent entered poverty. There are similar differences for the other two parental education categories, but not all are statistically significant.


Nonpoor children whose parents met the working threshold in both years generally had the lowest likelihood of entering poverty for the two higher education categories—2 percent in 1997 if parental education was at least 13 years and 5 percent in 1997 if parental education was 12 years. However, in the lowest education category, the likelihood of poverty was not consistently or significantly lower than for children whose parents moved from below to above the working threshold.
Indeed, there is evidence that parental education is significantly associated with the likelihood of a nonpoor child entering poverty, regardless of the dynamics of parental work behavior. For example, the likelihood of a nonpoor child entering poverty in spite of parental employment above the working threshold is consistently only 1-2 percent if parental education exceeds 13 years. The corresponding percentage for nonpoor children whose parental education is less than 12 years ranges between 3 percent and 13 percent. The difference is statistically significant in all but one year. Results are similar if

children whose better educated parent has 13 or more years of education are compared with children whose better educated parent has 12 years of education. Comparisons of children whose better educated parent has 12 years of education and children whose

better educated parent has less than 12 years of education yields somewhat less consistent and significant results, but the direction of the difference is generally in the same direction.


Summary
Over time, there is a rough balance between the number of children entering and leaving poverty each year. However, during the 1990-1991 recession, the number of children entering poverty was significantly higher than the number of children leaving poverty; in contrast, during the 1993-94 and 1996-1997 periods of substantial economic growth, the number of children leaving poverty was significantly higher than the number entering poverty.
Consistently over the entire period, increasing parental work effort to meet or exceed the working standard is more effective for moving children out of poverty than not meeting the working standard in either year. However, in every year tabulated, this was successful at removing the child from poverty only about half the time. For the other half, increasing work effort led the family into the ranks of the working poor.
Conversely, meeting the working standard for two consecutive years is a good strategy for avoiding moving into poverty. For nonpoor children whose parents met the working standard in both 1996 and 1997, the likelihood of moving into poverty was only 3 percent. This is significantly lower than the likelihood for (a) those nonpoor children whose parents met the working standard in the second year only (10 percent), and (b) those nonpoor children whose parents failed to meet the working standard in either year (21 percent).
These two findings generally hold regardless of whether the child is in a married-couple family or a single-mother family. In fact, there was no significant difference by family structure in the likelihood of poor children leaving poverty as a consequence of their parent(s) meeting the working standard in 1997, given that they did not meet the working standard in 1996. Moreover, with only a few exceptions, nonpoor children in single- mother families were not significantly more likely than nonpoor children in married- couple families to enter poverty in the following years, regardless of the work behavior

of their parent(s).


Similarly, the pattern of association between parental work behavior and the likelihood of a child exiting or entering poverty held regardless of the educational attainment of the better educated parent. However, there were indications that parental education may play a role in the likelihood of a child exiting or entering poverty, after controlling for the dynamics of parental work behavior. Higher levels of education appear to be associated with a better chance of escaping poverty and a better chance of avoiding poverty. However, these relationships were not consistently statistically significant.

Changes in work behavior and family resources, 1996-1997


In Chapter 5, looking at repeated cross-sections of the U.S. population using the Current Population Survey, we found that the number of children in poor families not meeting the work standard fell steadily over the entire period from 9.5 million in 1995 to 6.4 million in 2000. We also observed a decline in poverty over the same period. This pattern presumably reflects the combined effects of a strong economy and federal welfare reform—both of which encouraged and facilitated greater work effort by persons with relatively low earning potential.
In this section we use SIPP to examine the changes in both work behavior and family resources between 1996 and 1997—the first year of transition to a welfare reform environment. To do this, we have focused on five key groups of children:
1. Children in families who were receiving TANF in 1996;
2. Children in poor families not meeting the work standard in 1996;
3. Children in working poor families in 1996;
4. Children in working families with incomes between 100 percent and 200 percent of the poverty threshold in 1996; and
5. Children in working families with incomes over 200 percent of the poverty threshold in 1996.
Using SIPP’s longitudinal capability, we examine how each group of families changed their work behavior between 1996 and 1997 and how their economic resources changed.
Given the emphasis of welfare reform on encouraging increased work and the context of a strengthening economy, we would expect hours worked per week and weeks worked per year to have increased for children in TANF families, children in poor families not meeting the work standard, and possibly children in working poor families as well. The impact on weeks of unemployment is unclear. On the one hand, the incentives to work embedded in the welfare reform law would likely increase the likelihood of an unemployed person accepting a job (assuming a strong economy) and, thus, reducing unemployment. On the other hand, these same incentives would be likely to cause many people who had previously not been seeking employment to do so and, thus, potentially to increase unemployment.
For the last two groups, children in working families with incomes either between 100 percent and 200 percent of the poverty line or above 200 percent of the poverty line, we would expect that the strong economy might also encourage increased work but that the effect of federal welfare reform would be much less for the near poor and non-existent for the more affluent.
Turning to changes in economic resources, the likely impact of welfare reform on family income for our five groups is less clear. On the one hand, welfare reform, by encouraging work, was likely to increase earnings of 1996 welfare recipients, poor families not meeting the work standard, and possibly working poor families. On the

other hand, cash transfers from AFDC/TANF were likely to decrease or be eliminated for

1996 welfare recipients and poor families not meeting the work standard. These transfers

would be much less important for the other three groups. We also observed in Chapter 4 a substantial drop in receipt of food stamps, so we might expect the same phenomenon for the first three groups.


Changes in work behavior
1996-1997 changes in work behavior are reported in Table 5.5. The top panel reports on changes for fathers (if present), while the bottom panel reports on changes for mothers (if present). Results are reported separately for two-parent and single-parent families.
Families receiving AFDC/TANF in 1996. Single parents who received AFDC/TANF in 1996 worked substantially more in 1997 than in 1996. The percentage of single fathers who usually worked zero hours per week (in other words usually didn’t work at

all) decreased from 52 percent in 1996 to 23 percent in 1997. Conversely, the percentage who usually worked at least 35 hours per week (full-time) increased from 48 percent to

70 percent. Similarly, the percentage of single mothers who usually didn’t work at all decreased from 49 percent to 42 percent, while the percentage who worked full-time increased from 30 percent to 39 percent.
In addition to an increase in hours worked per week, there was also an increase in weeks worked per year. The percentage of single fathers who worked zero weeks per year decreased from 46 percent to 21 percent, while the percentage who worked 50-52 weeks per year (full-year) increased from 39 percent to 51 percent. Similarly, the percentage of single mothers who worked zero weeks per year decreased from 49.5 percent to 42 percent, while the percentage who worked full-year increased from 14 percent to 25 percent.
For mothers in two-parent families receiving AFDC/TANF in 1996, the pattern is the same. The percentage of mothers usually working zero hours per week and/or zero weeks per year decreased significantly, while the percentage of mothers working full-

time and/or full-year increased significantly. For fathers in two-parent families receiving AFDC/TANF in 1996, there was a significant increase in the percentage working full- year but no significant changes in the percentage not working at all.


In short, for AFDC/TANF recipients, the group of parents most directly affected by federal welfare reform, there was a substantial increase in work effort for single fathers and both single and married mothers with a less apparent increase for married fathers.
The pattern is not clear for weeks of unemployment with, for example, a significant increase in the percentage of married mothers unemployed for 14 weeks or more but a decrease in the percentage unemployed for 5-13 weeks. This is not unexpected since there is no a priori reason for expecting either an increase or decrease in weeks of unemployment as a result of federal welfare reform.
Table 5.5. Percentage of children in families by 1996 TANF receipt, work and poverty status, by parents' weekly work hours, weeks worked, and weeks unemployed, 1996 and 1997




Families receiving TANF in 1996



Poor Families Not Meeting the

Work Standard in 1996


Working Poor Families in 1996

Working 100-200% of poverty in

1996

Working 200% poverty and above in 1996

Two-Parent

Single-Parent

Two-Parent

Single-Parent

Two-Parent

Single-Parent

Two-Parent

Single-Parent

Two-Parent

Single-Parent

Fathers



Hours worked per week

0 hours



1-10 hours

11-20 hours

21-34 hours

35+ hours


1996 1997

32.4% 33.5%

0.5% 1.4%

4.1% 2.9%

7.8% 4.9%

55.2% 57.3%

1996 1997

51.8% 23.0%



0.0% 0.0%

0.0% 0.9%

0.0% 6.0%

48.3% 70.0%



1996 1997

45.6% 45.4%

1.0% 1.1%

4.4% 3.6%

15.5% 7.7%

33.5% 42.3%


1996 1997

70.9% 29.5%

0.0% 1.3%

9.5% 4.3%

3.8% 14.3%

15.8% 50.6%

1996 1997

18.1% 18.0%

0.1% 1.0%

1.3% 1.6%

4.2% 4.6%

76.3% 74.7%

1996 1997

21.1% 13.5%



0.0% 2.0%

1.2% 0.9%

2.3% 0.0%

75.4% 83.6%



1996 1997

15.1% 16.5%

0.5% 0.2%

0.9% 1.7%

3.0% 2.4%

80.5% 79.3%



1996 1997

11.4% 13.2%

0.0% 0.0%

0.0% 0.0%

1.7% 4.6%

86.8% 82.2%

1996 1997

12.3% 13.0%

0.3% 0.3%

0.3% 0.5%

1.4% 1.3%

85.7% 84.9%

1996 1997

9.9% 9.4%

0.2% 0.0%

0.3% 0.8%

1.0% 0.9%

88.5% 88.9%

Weeks worked

0 weeks

1-13 weeks

14-26 weeks

27-39 weeks

40-49 weeks

50-52 weeks



26.4% 26.2%

5.9% 5.0%

11.3% 6.2%

8.0% 9.0%

8.5% 7.0%

40.0% 46.7%

45.6% 21.4%



1.0% 3.8%

1.9% 7.1%

2.1% 6.6%

10.3% 10.5%

39.2% 50.7%



37.1% 37.3%

9.1% 6.0%



12.0% 7.8%

14.3% 9.8%

9.5% 10.9%



17.8% 28.1%


67.7% 27.0%

8.6% 0.0%

8.5% 4.5%

7.1% 12.6%

2.6% 6.0%

5.6% 49.9%


2.7% 3.1%

2.3% 2.4%

3.1% 3.2%

7.2% 6.0%

13.1% 6.1%

71.5% 79.1%


0.0% 2.1%

1.3% 0.0%

4.2% 1.5%

10.5% 11.0%

20.9% 10.4%

63.1% 75.1%


2.1% 3.4%

1.1% 0.7%

2.6% 2.1%

6.5% 6.2%

7.8% 6.2%

79.9% 81.4%



0.9% 1.6%

0.6% 0.7%

0.0% 3.5%

8.9% 4.8%

10.8% 8.1%

78.8% 81.5%


0.8% 1.1%

0.3% 0.4%

0.6% 0.8%

1.9% 2.2%

3.1% 3.7%

93.3% 91.8%


0.9% 2.5%

0.0% 1.1%

0.0% 1.4%

2.0% 3.1%

1.6% 4.3%

95.5% 87.6%

Weeks unemployed

0 weeks



2-4 weeks

5-13 weeks

14+ weeks



63.5% 69.3%

2.6% 2.3%

5.1% 4.3%

28.9% 24.2%

69.8% 71.0%



6.9% 0.0%

6.7% 6.4%

16.6% 22.6%




53.3% 62.0%

3.2% 2.4%

3.9% 8.2%

39.6% 27.4%



67.9% 83.1%

16.2% 0.0%

2.2% 4.0%

13.8% 12.9%


73.0% 81.3%

1.9% 3.3%

7.9% 3.6%

17.1% 11.8%

70.0% 72.0%



0.0% 0.0%

5.4% 13.0%

24.6% 14.9%


82.4% 85.6%

2.1% 1.7%

3.5% 3.4%

12.0% 9.4%



83.1% 85.2%

2.9% 1.0%

5.5% 2.4%

8.5% 11.3%


93.9% 94.7%

1.0% 0.6%

1.7% 1.2%

3.5% 3.5%


95.2% 90.7%

1.0% 0.9%

2.7% 0.9%

1.1% 7.5%

Mothers

Hours worked per week

0 hours



1-10 hours

11-20 hours

21-34 hours

35+ hours




53.6% 47.4%

3.3% 1.4%

5.6% 4.7%

10.3% 11.2%

27.2% 35.4%

49.2% 42.0%



1.9% 1.4%

6.3% 4.7%

12.8% 12.9%

29.8% 39.1%




69.1% 61.3%

3.2% 0.3%

8.0% 3.9%

9.4% 9.6%



10.2% 25.0%



64.4% 54.6%

2.8% 1.7%

7.9% 5.5%

11.1% 13.3%

13.9% 25.0%



45.8% 44.6%

6.6% 5.5%

4.5% 5.6%

14.0% 12.1%

29.1% 32.1%



7.1% 8.0%

0.3% 0.0%

0.8% 3.1%

20.0% 16.0%

71.8% 73.0%


33.8% 34.6%

4.0% 3.2%

6.9% 6.6%

15.8% 14.6%

39.6% 41.0%





4.4% 7.1%

0.2% 0.3%

1.8% 1.4%

9.1% 9.8%

84.5% 81.5%



22.3% 23.2%

3.2% 2.9%

6.9% 6.5%

13.8% 13.3%

53.8% 54.1%



3.0% 4.5%

0.0% 0.3%

0.5% 0.9%

7.5% 7.4%

89.0% 87.0%

Weeks worked

0 weeks

1-13 weeks

14-26 weeks

27-39 weeks

40-49 weeks

50-52 weeks




49.0% 43.4%

10.3% 10.1%

10.2% 9.0%

7.0% 7.8%

4.0% 6.3%

19.6% 23.4%

49.5% 42.0%

10.3% 8.7%

9.3% 9.2%

10.4% 7.9%



6.9% 7.2%

13.6% 25.1%




68.3% 60.9%

12.1% 12.8%

4.9% 6.5%

7.3% 5.4%

1.2% 3.4%

6.2% 11.0%



65.8% 53.9%

14.5% 11.8%

10.2% 9.1%

5.5% 7.3%

1.7% 5.6%

2.3% 12.4%


42.6% 42.6%

10.8% 8.2%

11.5% 8.1%

5.4% 8.6%

7.5% 6.0%

22.3% 26.5%


5.6% 8.2%

1.2% 2.9%

3.4% 7.6%

19.0% 9.2%

18.0% 11.2%

52.9% 60.9%


28.0% 30.1%

6.8% 5.1%

8.1% 5.9%

7.9% 6.3%

8.3% 7.0%

40.8% 45.7%



2.3% 4.9%

0.7% 2.0%

2.8% 4.0%

6.8% 5.5%

12.2% 6.0%

75.2% 77.7%


15.6% 16.6%

3.0% 3.0%

3.9% 3.4%

5.2% 4.6%

4.4% 4.8%

67.9% 67.6%


1.0% 2.1%

0.4% 0.7%

0.5% 2.4%

4.5% 3.6%

4.7% 5.0%

89.0% 86.3%

Weeks unemployed

0 weeks



2-4 weeks

5-13 weeks

14+ weeks



69.6% 67.1%

3.4% 5.1%

10.4% 6.6%

16.6% 21.3%

58.9% 62.6%



6.5% 5.2%

9.4% 9.8%

25.2% 22.4%




71.2% 77.3%

3.8% 3.7%



10.4% 4.5%

14.6% 14.6%


57.3% 58.9%

6.2% 5.5%

9.6% 9.1%

27.0% 26.5%


76.5% 75.9%

3.3% 3.8%

6.4% 7.1%

13.8% 13.2%

64.3% 70.7%



6.0% 3.8%

7.2% 9.8%

22.5% 15.7%


81.1% 84.4%

3.6% 2.9%

5.5% 4.1%

9.8% 8.7%



81.9% 84.5%

3.7% 1.9%

4.3% 5.6%

10.1% 8.0%


90.1% 91.5%

2.1% 1.6%

2.7% 2.6%

5.2% 4.3%


91.7% 91.5%

2.0% 1.0%

2.8% 2.0%

3.5% 5.4%

Source: 1996 Survey of Income and Program Participation

49
Families not meeting the work standard and poor in 1996. For parents of children in poor families not meeting the work standard in 1996, a group which overlaps extensively but not exactly with families receiving AFDC/TANF in 1996,17 the results are similar to those for families receiving AFDC/TANF. The percentage of single fathers who report working zero hours per week dropped from 71 percent in 1996 to 29 percent in 1997. Conversely, the percentage who usually worked full-time increased from 16 percent to 51 percent. Similarly, the percentage of single mothers who usually didn’t work at all decreased from 64 percent to 55 percent, while the percentage who worked full-time increased from 14 percent to 25 percent.


In addition to an increase in hours worked per week, there was also an increase in weeks worked per year. The percentage of single fathers who worked zero weeks per year decreased from 68 percent to 27 percent, while the percentage who worked 50-52 weeks per year (full-year) increased from 6 percent to 50 percent. Similarly, the percentage of single mothers who worked zero weeks per year decreased from 66 percent to 54 percent, while the percentage who worked full-year increased from 2 percent to 12 percent.
For mothers in two-parent poor families not meeting the work standard in 1996, the pattern is the same. The percentage of mothers usually working zero hours per week and/or zero weeks per year decreased significantly, while the percentage of mothers working full-time and/or full-year increased significantly. For fathers in two-parent families not meeting the work standard in 1996, there was a significant increase in the percentage working full-time and full-year but no significant changes in the percentage not working at all.
Unlike families receiving AFDC/TANF, there is a clear pattern to the changes in weeks unemployed for poor two-parent families not meeting the work standard. For fathers in two-parent families, there was a significant increase in the percentage that had zero weeks of unemployment (from 53 percent to 62 percent) and a significant decrease in the percentage that had at least 14 weeks of unemployment (from 40 percent to 27 percent).

Similarly, for mothers in two-parent families, there was an increase in the percentage that had zero weeks of unemployment (from 71 percent to 77 percent) and a decrease in the percentage that had 5-13 weeks of unemployment.


On the other hand for poor single-parent families not meeting the work standard, there were virtually no significant changes in weeks of unemployment.
Families meeting the work standard but poor in 1996. Parents who were in working poor families in 1996 are much less likely to have received AFDC/TANF in 1996 than parents in poor families not meeting the work standard.18 Consequently, they were less likely to be directly affected by federal welfare reform. Nonetheless, the few significant changes apparent in the work behavior of working poor parents are in the positive direction. For example, the percentage of married fathers who report usually working 50-


17 In 1996, 64 percent of children in poor single-mother families not meeting the work standard received AFDC

benefits, and 34 percent of children in poor, married-couple families did so.


18 In 1996, 25 percent of children in single-mother, working poor families and 7 percent of children in married-couple, working poor families received AFDC.

52 weeks per year increased from 71 percent to 79 percent, and the percentage of single mothers who report working full-time increased from 53 percent to 61 percent. Changes in weeks of unemployment were not significant.


Working families with incomes between 100 percent and 200 percent of the poverty threshold in 1996. For parents in working families with incomes between 100 percent and 200 percent of the poverty threshold, there is little change in work behavior that would be consistent with a response to federal welfare reform. There is a significant increase in the percentage of married mothers working full-year from 41 percent to 46 percent, but most changes in work behavior are statistically insignificant. There was also a decrease in weeks unemployed for married fathers.
The fact that the parents in these families show little or no change in work behavior while the parents in families receiving AFDC/TANF and the parents in poor families not meeting the work standard (and, to a lesser extent, parents in working poor families) do show significant increases may be consistent with the increases being due to welfare reform rather than to the strong economy. However, an alternative hypothesis is that the strong economy was disproportionately benefiting disadvantaged workers.
Working families with incomes above 200 percent of the poverty threshold in 1996.

Significant changes in work behavior for parents in working families with incomes above

200 percent of the poverty threshold are limited to fathers, and the direction of change is generally negative, although the changes are small. For example, the percentage of both married and single fathers working full-year decreased somewhat. This divergence in behavior between more affluent parents and the less advantaged parents in TANF families and families not meeting the work standard again may be consistent with the

latter groups of parents reacting to welfare reform (and possibly also the strong economy)

by working more.
Changes in family resources
1996-1997 changes in family resources are reported in Table 5.6.
Families receiving AFDC/TANF in 1996. For children in families receiving TANF in

1996, mean family income rose from $17,289 in 1996 to $19,072 in 1997. This is consistent with our findings above that parents increased their work effort and also, possibly, with increased wage rates as well. As a consequence the percentage of children in these families that were living in extreme poverty (less than 50 percent of the poverty threshold) decreased from 29 percent to 26 percent, and the percentage living between 50 percent and 100 percent of the poverty threshold decreased from 39 percent to 27

percent. These improvements took place in spite of a decrease in receipt of AFDC/TANF (which counts as income since it’s cash) from 100 percent to 73 percent.
However, somewhat offsetting these positive changes, the percentage of children in these families who had health insurance coverage dropped from 96 percent to 89 percent, and the percentage receiving food stamps (which doesn’t count as income since it’s not cash) dropped from 93 percent to 80 percent.
Families not meeting the work standard and poor in 1996. For the overlapping group of children living in poor families not meeting the work standard in 1996, mean income

rose from $9,234 in 1996 to $12,783—against consistent with the increase in work effort reported earlier and possibly increased wage rates. Consistent with this increase in income, the percentage in extreme poverty dropped from 46 percent in 1996 to 38 percent in 1997, while the percentage living between 50 percent and 100 percent of the poverty threshold decreased from 54 percent to 41 percent. These improvements took place in spite of a decrease in AFDC/TANF receipt from 68 percent to 60 percent.


The increase in mean income for this group seems at first glance to be implausibly large. However, the group whose income is being tracked includes a substantial number of families whose income increased enough to push them above the poverty line. However, by definition, it does not include any families for whom a drop in income pulled them below the poverty line. These are included in the groups that were not poor in 1996. We have seen earlier in this chapter that there are substantial year-to-year movements both into and out of poverty, but for this group, we are measuring income change only for those who stayed in poverty and those who moved out.
Both health insurance coverage and food stamp coverage dropped for this group—from

88 percent to 82 percent for health insurance coverage and from 79 percent to 58 percent for food stamp coverage.


Families meeting the work standard but poor in 1996. For children in families whose parents already met the work standard in 1996, family income rose from $12,794 to

$17,479.19 The percentage in extreme poverty did not significantly change (between 15 percent and 16 percent), but the percentage in families with incomes between 50 percent and 100 percent of the poverty threshold decreased from 84 percent to 50 percent. This improvement occurred in the face of a decrease in TANF receipt from 26 percent to 16 percent.


The percentage of these families with health insurance decreased between 1996 and 1997 from 70 percent to 66 percent, while the percentage receiving food stamps fell from 58 percent to 45 percent.
Working families with incomes between 100 percent and 200 percent of the poverty threshold in 1996. Family income also increased (from $26,631 to $30,120) for children in families who met the work standard and whose income was between 100 percent and

200 percent of the poverty threshold in 1996. Since we didn’t observe substantial increases in hours or weeks worked for parents in this category, it seems most likely that the increase is due to increased wage rates. TANF receipt fell slightly from 9 percent to

7 percent.


19 The reasoning for why this increase is not implausibly large is the same as for the poor families not meeting the work standard.
Table 5.6. Percentage of children in families by 1996 TANF receipt, work status, and poverty status, by mean family income, parental education, family poverty status, health insurance coverage, and

food stamp receipt, 1996 and 1997


Families Receiving



TANF

Poor Families Not

Meeting the Work

Standard
Working Poor

Families


Working 100-

200% of Poverty

Working 200%

of Poverty and

Above







1996

1997

1996

1997

1996

1997

1996

1997

1996 1997

Mean Family Income


$17,289


$19,072


$9,234




$12,783


$12,794

$17,479




$26,631


$30,120

$69,135 $70,023




Poverty Status

Extreme Poverty



29.3%





26.3%

45.8%



38.4%





15.9%

15.2%





0.0%



2.3%

0.0% 0.4%



Poor

39.0%

26.8%

54.2%

41.1%

84.1%

50.4%

0.0%

11.4%

0.0% 0.8%

100-199%

23.7%

27.0%

0.0%

17.7%

0.0%

30.0%

100.0%

62.8%

0.0% 8.9%

200% plus poverty

8.1%

9.9%

0.0%

2.9%

0.0%

4.3%

0.0%

23.6%

100.0% 89.9%

Health Insurance coverage


96.4%




89.0%

87.9%


82.3%




69.8%

65.7%


83.6%




80.7%

97.0% 96.2%


Food stamp receipt


92.8%




80.4%

84.9%


78.7%




57.7%

44.9%


21.7%




16.0%

3.2% 2.5%




AFDC/TANF receipt


100.0%


72.7%

68.4%


60.3%




26.1%

16.2%




8.7%


6.5%

1.5% 1.2%


Source: 1996 Survey of Income and Program Participation



Health insurance coverage did not change significantly for these families, but, as with the preceding group, food stamp coverage decreased from 22 percent to 16 percent.
Working families with incomes above 200 percent of the poverty threshold in 1996. Unlike the other groups, mean family income did not increase significantly between 1996 and 1997 for children in working families with incomes above 200 percent of the poverty threshold in 1996. This may be consistent with the strengthening economy primarily benefiting the earnings of more disadvantaged workers.
Health insurance coverage was nearly universal for this group and did not change significantly. Conversely, almost none of these families reported receiving AFDC/TANF or food stamps in either year.
Summary
For families that received AFDC/TANF in 1996 and the overlapping group of poor families not meeting the work standard, parents generally increased their work effort between 1996 and 1997—typically both hours worked per week and weeks worked per year. To a somewhat lesser extent, this was also true of parents in working poor families. There was little substantial change in work for working families above the poverty line.
Family income rose significantly for children in families receiving TANF in 1996, families that were poor and did not meet the work standard in 1996, families that were poor and did meet the work standard in 1996, and working families with incomes between 100 percent and 200 percent of the poverty threshold in 1996. They did not change significantly for more affluent working families.

Food Stamp receipt and health insurance coverage fell between 1996 and 1997 for families that received AFDC/TANF in 1996 and also both categories of families that were poor in 1996. Food stamp but not health insurance coverage fell for working families with incomes between 100 percent and 200 percent of the poverty threshold. Health insurance coverage for the more affluent working families was nearly universal and did not change year-to-year, while Food Stamp receipt was virtually non-existent among these families in both years.

Chapter 6. Analysis of the sensitivity of results to the definition of poverty


The results presented in Chapter 4 used the official definition of poverty. This definition counts as economic resources only pre-tax money income. Thus, it does not include in- kind transfers including Food Stamps, housing assistance, energy assistance, etc. It does not subtract income or payroll taxes, even though these taxes may reduce or increase (via the Earned Income Tax Credit) the amount of money that families have to purchase

goods and services. The official poverty definition also fails to take into account work- related expenses such as paid child-care expenses and commuting expenses.


Addressing all or most of the problems of the official poverty definition, as was attempted by the National Research Council (National Research Council, 1995), was beyond the scope of this project. Instead, we focused on making several key adjustments that would have the effect of measuring the economic wellbeing of children in working families and children in families not meeting the work standard on a more equal footing.
The alternative poverty definition
Our alternative poverty definition is different in two key respects from the official definition:
1. We add the dollar value of the Earned Income Tax Credit (EITC), food stamps, energy assistance, housing assistance, and school lunch program to each family’s income to obtain a more comprehensive measure of gross economic resources.
2. We subtract from gross economic resources three key work-related expenses including the family’s federal income and payroll tax liability and the family’s estimated child-care expenses to obtain a measure of the family’s net economic resources. For married-couple families, child care expenses are considered deductible only up to the value of the earnings of the lesser earning spouse.
Under this alternative definition, we define a family as poor if its net economic resources

are less than the official poverty threshold.


As should be clear from the preceding paragraphs, this definition does not take into account all of the issues raised by critics of the official poverty definition. However, it addresses some of the major shortcomings of the official measure, and, for purposes of a study of children in working poor families, it puts children in working families and children in families not meeting the work standard on a more equal footing. 20
As shown in Chapter 4, working families are less likely than families not meeting the work standard to receive Food Stamps. Moreover, working families are more likely than families not meeting the work standard to pay a substantial amount of federal payroll taxes and child-care expenses. On the other hand, working families are more likely to be

20 Other problems with the official poverty definition have also been identified (National Research Council, 1996). These include the lack of adjustments for geographic differences in housing costs and the rental value of owner- occupied housing, the absence of a connection between the poverty threshold and the overall standard of living,, and its failure to take into account publicly provided health insurance (especially Medicare and Medicaid).

eligible for substantial refundable tax credits under the Earned Income Tax Credit.21

Thus, taking into account just Food Stamps, federal payroll taxes and child care expenses, working families are likely to have lower net economic resources with the alternative poverty definition than under the official definition. However, when the effect of the Earned Income Tax credit is taken into account as well, net economic resources might be either higher or lower than under the official definition of poverty.
Families not meeting the work standard are likely to have higher net economic resources under the alternative poverty definition. This is because their Food Stamp allotment will be counted as part of net economic resources, their earnings (if any) are likely to generate an earned income tax credit, which will further increase their net economic resources,

and they are less likely to have substantial child care expenses.


If a family’s ability to pay for goods and services for their children is an appropriate yardstick to measure child economic wellbeing, then the alternative poverty definition puts working families and families not meeting the work standard on a more equal footing than does the official poverty definition. This is because it is based on a family’s income after federal payroll and income taxes and after child-care expenses. Thus,

when the impact of welfare reform on children’s economic wellbeing is assessed, the alternative poverty definition does a better job than the official poverty definition of measuring whether children are economically better off when their parents increase their work effort in response to the work requirements of TANF.


Imputing child care expenses
Unfortunately, as noted earlier, the Current Population Survey (CPS) does not collect information on child care expenses. However, the 1996 Survey of Income and Program Participation (SIPP) does—along with many variables which it has in common with the CPS. Thus, in order to be able to take child care expenses into account when calculating a family’s net economic resources, we imputed child care expenses onto each family’s record using statistical relationships estimated using SIPP. The model is based on a specification developed by the U.S. Census Bureau (Short et al., 1999).
Results
In this section, we focus on results using our alternative poverty definition for 1995 and

2000—the same period analyzed for the official poverty definition. As shown in Figure

6.1 (and Figure 4.1), employing the alternative poverty definition instead of the official poverty definition decreases the percentage of children in poverty from 15 percent to 14

21The generosity of the Earned Income Tax Credit has increased dramatically between 1987 and 2002. In 1987, the maximum earned income tax credit was $851, which was paid to families whose earnings were between $6,080 and

$6,920. The credit was phased out gradually as earnings increased beyond $6,920 and fell to zero when earnings reached $15,432.


By 2001, the maximum earned income tax credit was $2,428 for a family with one child and $4,008 for a family with more than one child. The maximum credit was paid within the income range of $7,100 and $13,000 for a family with one child and within the income range of $10,000 and $13,000 for a family with more than one child. The phase-out points for earnings were $28,250 for a family with one child and $32,100 for a family with more than one child.

Figure 6.1. Among all children, percentage living in poverty (using the alternative poverty definition): all families, families meeting the work standard, and families not meeting the work standard, 1995-2000



60%

50%

56% 56% 56%

53%

50%

48%




40%

30%
Percent




20%

10%
18% 18% 17% 16%

15% 14%



7% 7% 7% 7% 7% 7%
0%

1995 1996 1997 1998 1999 2000
All families Working Not meeting work std

percent in 2000 and from 20 percent to 18 percent in 1995. These differences are both statistically significant.


Similarly, among families not meeting the work standard, the percentage of children who were living in poor families is significantly lower (48 percent versus 54 percent in 2000 and 56 percent versus 62 percent in 1995) using the alternative poverty definition. For working families, there was a small but statistically significant difference in the percentage of children who were poor. Family resources for children in working families were probably less affected by the change in definition because their families are less likely to receive Food Stamps and other noncash benefits. Also, for working families

near the poverty line, the Earned Income Tax Credit would tend to reduce the likelihood of poverty, while the deduction for child care expenses would increase it.22


When children in married-couple families are considered separately, the differences in the 2000 and 1995 poverty rates are not significantly different from the rates using the official poverty definition. However, the differences in both 2000 (36 percent poor versus 31 percent poor) and 1995 (46 percent versus 40 percent) for children in single- mother families are statistically significant. Moreover, there are also statistically significant differences in the percentage that are poor in single-mother families for

22 About 80 percent of children living in married-couple, poor families not meeting the work standard receive Food

Stamps, compared with about 50 percent of children in married-couple families who meet the work standard (see Table



6.7). For children in single-parent families, the corresponding statistics are 94 percent and 81 percent.

Table 6.1. Using the alternative poverty definition, the number of children (thousands)

living in poor families and as a percentage of all children by family structure and

whether hours worked by parents met the work standard, 1995-2000

1995 1996 1997 1998 1999 2000

All families

12,590

12,350

12,320

11,370

10,540

9,971

Working

3,917

3,899

4,163

4,142

4,158

4,214

Not meeting work std

8,670

8,452

8,156

7,232

6,382

5,757

Married-couple families

4,197

3,974

4,074

3,719

3,718

3,504

Working

2,307

2,262

2,458

2,363

2,334

2,297

Not meeting work std

1,890

1,712

1,615

1,357

1,384

1,207

Single-mother families 6,725 6,529 6,461 6,110 5,249 4,806

Working

1,375

1,363

1,426

1,597

1,660

1,696

Not meeting work std

5,350

5,167

5,035

4,513

3,588

3,111


1995 1996 1997 1998 1999 2000

All families 18% 18% 17% 16% 15% 14% Working 7% 7% 7% 7% 7% 7% Not meeting work std 56% 56% 56% 53% 50% 48% Married-couple families 9% 8% 8% 8% 8% 7% Working 5% 5% 6% 5% 5% 5% Not meeting work std 46% 43% 46% 41% 44% 44%



Single-mother families 40% 39% 39% 36% 33% 31%
Working 15% 15% 15% 16% 16% 16% Not meeting work std 68% 69% 69% 68% 62% 60% Source: March Current Population Survey, 1996-2001.

Although the percentages of children who are poor are significantly lower for many of the groups discussed in the previous paragraphs, an important finding from previous chapters remains unchanged. Children in working families are far less likely to be poor


Table 6.2. Using the alternative poverty definition, the percentage of children living in poor families, by race/ethnicity and whether hours worked by parent(s) met the work standard, 1995-2000
1995 1996 1997 1998 1999 2000

White, non-Hispanic

Working 4% 4% 5% 4% 4% 4% Not meeting work std 45% 47% 48% 46% 43% 41% Black, non-Hispanic

Working 12% 11% 11% 13% 12% 12% Not meeting work std 66% 65% 63% 57% 59% 58% Asian, non-Hispanic

Working 7% 4% 9% 7% 8% 6% Not meeting work std 49% 49% 47% 51% 36% 40% Hispanic

Working 19% 18% 17% 17% 15% 14% Not meeting work std 64% 64% 67% 62% 53% 51% Source: March Current Population Survey, 1996-2001.

Table 6.3. Using the alternative poverty definition, the percentage of children living in poor families, by whether the hours worked by parent(s) met the work standard, family structure, and the educational attainment of better educated

parent, 1995-2000


Working


than children in families not meeting the work standard, regardless of which definition of poverty is used.
As shown in Table 6.2, when children are considered separately by race/ethnicity, black non-Hispanic and Hispanic children have the highest likelihood of poverty—just as with the official poverty definition—and meeting the work standard reduces the likelihood of poverty regardless of children’s race or ethnicity. As shown in table 6.3, when children are considered separately by parental education and family structure, the same pattern emerges as with the official poverty definition. Meeting the work standard reduces the likelihood of poverty for all parental education/family structure groups.


23 Even though children in poor, working, single-parent families are less likely to receive Food Stamps than children in poor, single-parent families not meeting the work standard, about 80 percent of children in poor, working, single- parent families receive Food Stamps. This probably accounts for the result that the change in the poverty definition significantly lowers the poverty rate for children in poor, working, single-parent families.

59


Married-couple

1995 1996 1997 1998 1999 2000

Less than 12 years 24% 26% 23% 20% 23% 23%

children in both working families and children in families not meeting the work standard.23


During the 1995-2000 period, the poverty rate of all families and the poverty rate for single-mother families dropped substantially—just as with standard definition. Just as with the standard definition, the drop is focused on families not meeting the work standard.


12 years 7% 6% 8% 8% 8% 7%

13-15 years 4% 4% 4% 4% 3% 3% At least 16 years 1% 2% 2% 2% 1% 2% Single-mother

Less than 12 years 30% 27% 32% 34% 36% 35%

12 years 17% 18% 20% 19% 16% 20%

13-15 years 11% 11% 9% 10% 12% 9% At least 16 years 5% 3% 3% 2% 3% 2%

Not meeting work standard

Married-couple

Less than 12 years

61%

57%

62%

55%

54%

62%

12 years

46%

44%

45%

42%

52%

47%

13-15 years

37%

34%

38%

38%

38%

38%

At least 16 years

30%

30%

31%

19%

25%

26%

ingle-mother

Less than 12 years

75%

78%

79%

78%

70%

64%

12 years

70%

68%

69%

68%

63%

62%

13-15 years

56%

58%

57%

53%

56%

55%

At least 16 years

45%

38%

31%

45%

29%

39%




S
Source: March Current Population Survey, 1996-2001.

Table 6.4. Using the adjusted poverty measure, among children living in poor families,



the percentage whose parent(s) met the work standard, by race/ethnicity and family

structure, 1995-2000

All races/ethnicities 1995 1996 1997 1998 1999 2000

All types of families 31% 32% 34% 36% 39% 42% Married couple families 55% 57% 60% 64% 63% 66% Single mother families 20% 21% 22% 26% 32% 35%

White non-Hispanic

All types of families 37% 36% 38% 40% 42% 46% Married couple families 57% 58% 63% 62% 63% 68% Single mother families 24% 26% 24% 29% 34% 36%



Black non-Hispanic

All types of families 20% 22% 23% 27% 30% 34% Married couple families 37% 52% 50% 69% 54% 52% Single mother families 20% 19% 22% 24% 32% 38%



Asian non-Hispanic

All types of families 30% 21% 41% 34% 54% 46% Hispanic

All types of families 36% 38% 39% 42% 44% 46% Married couple families 61% 63% 63% 68% 66% 70% Single mother families 16% 17% 20% 25% 27% 28%

Source: March Current Population Survey, 1996-2001.



As shown in Tables 4.4 and 6.4, use of the alternative poverty definition also has little effect on the share of children living in poor families accounted for by working families. Looking at poor children as a whole, there is no statistically significant difference between the standard and the alternative definitions. In 2000, 42-43 percent of poor
Table 6.5. Using the alternative poverty measure, among children living in poor

families, the percentage whose parent(s) met the work standard, by family structure and parental education, 1995-2000


1995 1996 1997 1998 1999 2000

Married-couple

Less than 12 years 50% 57% 53% 58% 64% 68%

12 years 56% 53% 63% 65% 64% 65%

13-15 years 61% 59% 65% 64% 58% 59% At least 16 years 58% 60% 63% 76% 64% 71%

Single-mother

Less than 12 years 14% 11% 15% 20% 27% 32%

12 years 23% 26% 29% 30% 33% 40%

13-15 years 27% 30% 23% 33% 38% 33% At least 16 years 27% 26% 35% 19% 31% 22%



Source: March Current Population Survey, 1996-2001.

Table 6.6. Using the alternative poverty definition, children in poor families not meeting

the work standard, working poor families, and other working families, by family structure,

by race/ethnicity, parental education, and average hourly wage rate, 2000


Family structure

Poor Families Not Meeting the Work Standard
Working Poor

Families
100-200%

poverty


200% and above

Married-couple 21% 55% 69% 85% Single-mother 55% 40% 26% 11% Single-father 6% 5% 5% 4% Other 17% 0% 0% 0% Race/ethnicity

White non-Hispanic 36% 42% 53% 76% Black non-Hispanic 32% 22% 18% 10% Asian non-Hispanic 3% 4% 4% 5% Hispanic 26% 30% 23% 9% Family structure and parental education



Married and education less than 12 years 31% 34% 15% 2% Single mother and education less than 12 years 37% 32% 15% 7% Family structure and average hourly wage rate

Married

Less than $5 per hour

17%

41%

9%

2%

$5 - $6.99 per hour

23%

22%

13%

3%

$7 - $9.99 per hour

29%

27%

24%

6%

At least $10 per hour

31%

10%

54%

89%

ingle mother

Less than $5 per hour

30%

45%

7%

3%

$5 - $6.99 per hour

28%

24%

17%

6%

$7 - $9.99 per hour

20%

24%

37%

12%

At least $10 per hour

21%

8%

39%

78%




S
Source: March 2001 Current Population Survey.

children live in families meeting the work standard. Similarly, there is no significant difference using the two different definitions in 1995. Children in working families account for just over 30 percent of poor families in 1995. The apparent difference in the percentage of children in single-mother families meeting the work standard under the alternative definitions is not statistically significant.


Just as in our previous analysis, children in poor, married-couple families are significantly more likely than children in poor, single-parent families to have a parent(s) meeting the work standard, as shown in Table 6.4. This finding holds across all four racial/ethnic groups. In fact, over two-thirds of white non-Hispanic and Hispanic

children living in poor married-couple families have parents that meet the work standard.


Other findings reported earlier (e.g, educational differences, differences in health insurance coverage, receipt of public assistance, home ownership, auto ownership, etc., between children in working poor families and children in families not meeting the work standard) are essentially unchanged as a result of using the alternative poverty definition. Consequently, they are not reported in detail here. However, the results are presented in Tables 6.6 and 6.7 for completeness.

Table 6.7. Children in poor families not meeting the work standard, working poor families, and other working families, by family structure, by receipt of AFDC and Food Stamps,

health insurance coverage, home ownership, 2000



Married-couple families

Poor Families Not Meeting the Work Standard


Working Poor

Families


100-200%

poverty


200% and above

Received AFDC 22% 6% 2% 0% Received Food Stamps 45% 18% 7% 0% Health insurance coverage 71% 69% 86% 96% Home ownershipa 27% 54% 64% 88%

Single-mother families



Received AFDC

40%

12%

6%

2%

Received Food Stamps

60%

39%

18%

4%

Health insurance coverage

84%

86%

85%

88%

Home ownershipa

21%

23%

34%

63%

Source: March 2001 Current Population Survey.

aAs of March 2001

Thus, although use of the alternative poverty definition changes the percentage of children in poverty—in some cases significantly—the patterns described in earlier chapters remain intact.


Summary
In this chapter, we have defined poverty differently and tested the sensitivity of our analysis of children in working poor families to using this alternative definition instead of the official poverty definition. More specifically, we have defined economic resources to include not just pre-tax money income but also the value of Food Stamps and other non- cash benefits and any refundable earned income tax credit received by the family. We have also excluded from economic resources any positive federal payroll or income tax liability and child care expenses. However, we have continued to use the poverty thresholds of the official poverty definition.
Using this definition results in 18 percent of children being classified as poor in 1996 and

14 percent in 2000. These percentages are slightly lower than the percentage classified as poor using the official definition (20 percent in 1995 and 15 percent in 2000). Moreover, the 1993-1994 decline in the alternative poverty rate is statistically significant.


However, use of the alternative poverty definition has little effect on the share of poor children living in working families. Moreover, findings reported in earlier chapters comparing working poor families with other poor families and other working families (e.g., differences in parental education, receipt of public assistance, health insurance coverage, etc.) are essentially unchanged.
In other words, use of the alternative poverty definition has noticeable effects on the level

of poverty but not on the patterns of poverty.



Chapter 7. The well being of children in working poor families and other families
In addition to measuring income and poverty status, the 1996 Survey of Income and Program Participation (SIPP) contains (1) variables that measure how well a child is developing and (2) variables that have been shown to be associated with how well a child is likely to develop. These variables are some of those developed by Child Trends for SIPP and the Survey of Program Dynamics to study the implications of welfare reform

for children.


The first section of this chapter uses these measures to contrast how well children in working poor families are developing compared with (1) children in poor families not meeting the work standard, (2) children in working families with incomes between 100 percent and 200 percent of the poverty line, and (3) children in more affluent working families (over 200 percent of the poverty line). This analysis permits us to assess whether children in working poor families are likely to develop differently from other children.
The second section of the chapter uses these same measures to contrast how well children are developing in two key groups—those whose families received AFDC in 1996 but did not receive TANF in1997 (termed “leavers”) and those who received AFDC/TANF in both years (termed “stayers”). This period encompasses 1996, the year prior to federal welfare reform, and 1997, the year in which. federal welfare reform was implemented.24

If we assume that families that succeed in leaving welfare are less likely than families staying on welfare for both years to have substantial pre-existing problems that could slow down child development, then we would expect to see more favorable child- wellbeing results for the children in leaver families than for children in stayer families. If, in fact, we observe that children in leaver families have less favorable results than children in stayer families, it is possible that the transition off of welfare may be having adverse effects on the children in leaver families. However, since the two groups are

self-selected, this analysis can’t establish any causal relationship between leaving welfare and child wellbeing.
SIPP variables that measure how well a child is developing include a health measure, two cognitive measures, and a behavior measure:

• Overweight status of the child;



Whether the child has been identified as a gifted student;

Whether child has ever repeated a grade; and

Whether child has ever been suspended or expelled from school.

24 More precisely, the periods we are contrasting are the first 12 months of observations for each of the four SIPP rotation groups in 1996 with the corresponding period one year later. For three of the rotation groups, the last 1-3 months of the first year’s data extend into the early part of 1997. Unfortunately, the scheduling of the 1996 SIPP (unlike the 1993 SIPP) precluding obtaining data for all 12 months of 1996 for all four rotation groups.

SIPP variables that have been shown to be associated with how well a child is likely to develop include both family-level variables and variables reflecting the child’s interaction with the broader community. Family variables include:

• Rules on watching television;

Whether child eats meals together with mother and father;

Parental aggravation (e.g., feeling angry, child is hard to care fore, etc.);

Father’s involvement with child;

Whether child has ever lived apart from parents; and

Parents’ expectations for the child’s education.

Variables indicating involvement in non-family activities include:

• Child’s participation in extra-curricular activities;

Child’s engagement in school;

Child’s receipt of child care services;

Child’s enrollment in private school;

Child’s enrollment in in religious school; and

Parents’ attitude towards their community.

Each of these groups of measures will be discussed in turn along with results.


Variables measuring how well the child is developing
Overweight status. Overweight children are more likely than other children to develop type 2 diabetes, cardiovascular problems, orthopedic abnormalities, gout, arthritis, and skin problems (Gidding et al., 1996). Moreover, overweight children are more likely than other children to become overweight adults, who face such problems as reduced productivity, social stigmatization, high costs for health care, and early death (U.S. Department of Health and Human Services, 2001b).

Children are considered to be overweight if their body-mass index is at the 95th percentile or greater.


Gifted status. Analysis of white, black, and Hispanic, elementary school students found that minority students in gifted programs scored significantly higher on achievement measures than minority students in regular classrooms (Cornell, 1995). Our measure the percentage of children participating in educational programs for gifted and talented students.
Repeated a grade. Children who have repeated kindergarten or a primary grade fall behind both academically and socially compared with students who have been promoted. Children who have repeated a grade in secondary school have lower cognitive achievement and much higher rates of school dropout (Heubert & Hauser, 1999).

School suspension or expulsion. Students who feel “cared for by people at their school and feel like part of a school” are less likely than other students to use illegal drugs, engage in violent activities, or become sexually active at an early age. Such “school connectedness” is lower in schools that suspend students for infractions such as alcohol possession and lower still in schools that expel students permanently for the first infraction (McNeely, Nonnemaker, & Blum, 2002).


Variables associated with how well a child will develop
In this section, we focus on variables associated with how well a child will develop.

They are divided into two major categories—those measuring the home environment and those measuring how the child and the child’s family are interacting with the broader community.


Variables measuring the home environment
Television rules. Excessive television viewing has been associated with lower scores on reading tests, reduced readiness for school, and a higher incidence of aggressive acts (Zaslow, Halle, Zaff, Calkins, & Margie, 2000). We have measured the degree to which television viewing is under parental control through the use of a scale that measures the number of rules which parents employ. The scale varies from zero (no rules at all) to three (three different types of rules).
Meals with parents. The amount of time a child spends with their parents is important. For example, time spent together by parents and their children is positively associated with the amount of education a child receives later in life (Haveman & Wolfe, 1994). Moreover, frequent father-child interactions have been associated with the child experiencing a greater sense of self-control, social competence, and ability to empathize (Hofferth, Pleck, Stueve, Bianchi, & Sayer, 2001; Lamb, 1997). We have constructed two scales varying from 0-14, which measure the total number of breakfasts and dinners a child eats with each of the child’s parents.
Parental aggravation. Child Trends has constructed an index varying from 0-12 based on responses to four questions about whether the child is perceived by the parents as “harder to care for than other children,” doing “things that really bother me,” requiring the parent to “giv[e] up more of my life to meet my child’s needs than I ever expected,” and making them feel angry with the child. Children whose parents are highly

aggravated are more likely than other children to have both cognitive and socioemotional problems (McGroder, 2000). Higher values for the index indicate lower levels of aggravation.


Father involvement. As stated earlier, frequent father-child interactions have been associated with the child experiencing a greater sense of self-control, social competence, and ability to empathize. We have constructed an index of father involvement which varies from 0-12 based on responses to three questions concerning the degree to which the father interacts with, has expectations for, and praises his child. Higher index values indicate a higher level of involvement.

Living apart from parents. Any separation of a child from his or her primary caregiver can traumatize a child and has been linked with poor academic performance and teenage pregnancy (Aquilino, 1996; Bowlby, 1973; Wu, 1996).


Variables measuring interaction with the community
Participation in extracurricular activities. Involvement in activities outside of school has been associated with a reduced risk of children being involved in delinquent

activities, as well as greater academic success, higher self-esteem, and greater community involvement as adults (Eccles & Barber, 1999; Mahoney, 1997). We have constructed an index of extracurricular involvement based on responses to questions about participation in sports teams, music, dance, language, computers or religion, and clubs or other organizations. The index ranges from 0-3 with a higher value indicated greater involvement.


School engagement. Children and youth with high levels of school engagement score higher on tests, have better attendance records, and more likely to advance from one grade to the next (Ehrle, Moore, & Brown, 1999). Moreover, teenagers with a high level of school engagement are less likely to get pregnant (Manlove, 1998).
We have constructed an index of the child’s engagement in school based on the parent’s view of whether the child likes to go to school, is interested in school work, and works hard in school. The index ranges from 0-6 with higher values indicated higher levels of involvement.
Parental attitudes towards the community. Parents were asked questions in two categories—one category reflecting positive views and one category reflecting negative views. We have constructed an index of positive views towards ones community based on responses to questions concerned with whether people in the neighborhood “help each other out,” whether they “watch out for each other’s children,” whether there are “people I can count on,” whether there are adults a parent could count on to “help my child,” and whether there are “safe places in this neighborhood for children to play outside.” The

1997 index can range from 0 to 20 with higher scores indicating a more positive attitude. The 1994 index is not comparable to the 1997 index even though it is based on the same set of questions25 and varies between 0 and 10.


We have constructed an index of negative view towards one’s community based on responses to questions concerned with whether the parent “keep[s] my child inside as much as possible because of the dangers in the neighborhood,” and whether “there are people in the neighborhood who might be a bad influence on my child.” The 1997 index can range from 0 to 8 with higher scores indicating a more positive view of the neighborhood. Like the positive index, the negative attitude index for 1994 is not comparable to the 1997 index and varies between 0 and 10.
Use of child care services. The associations between use of child care services and child development are complex. There is evidence that participation in quality preschool


25 The reason for the noncomparability is that the 1994 index is based on questions to which the possible responses ranged from zero to 10, while the 1997 is based on responses that ranged from zero to 4.

programs at ages 3 and 4 promotes cognitive development and gets children off to a good start in school (National Institute of Child Health and Human Development, 2002). However, there is also evidence that spending substantial amounts of time away from parents during infancy may be associated with behavior problems at age 2 (although

these problems disappeared by age 3) (National Institute of Child Health and Human

Development, 2002).


Attendance at kindergarten and private and religious schools. In a study of students in the Philadelphia public schools, research found that students who attended full-day kindergarten were 2.2 times as likely to be on grade level by third grade as students who did not attend kindergarten, after controlling for age, sex, and neighborhood poverty. Similarly, students who attended half-day kindergarten were 70 percent more likely to on grade level (Del Gaudio Weiss & Offenberg, 2001). Full-day kindergarten was also associated with higher achievement in reading, mathematics, and science, as well as

grade point average and attendance.


Results
We present our results first by work and poverty status and then by whether the family remained on TANF or left.
Results by work and poverty status
The analysis for 1997 presented here permits us to assess whether children in working poor families are likely to develop differently from other children. Where available, we also provide the same or similar measures for 1994.

Table 7.1. Percentage of children by overweight status, gifted student status, grade repetition status, and school suspension/expulsion status of children by family work and poverty status, 1997




Poor families not meeting

work standard
Working poor families
Working families,

100-199% poverty
Working familie

>200% povert

Overweight 6% 8% 6% 5% Gifted student 12% 9% 13% 20% Repeated a grade 13% 11% 10% 6% Suspended/expelled from school 19% 16% 12% 9%
Source: 1996 Survey of Income and Program Participation

Overweight status. As shown in Table 7.1, children in working poor families were significantly more likely than other children to be overweight. In 1997 8 percent of children in working poor families were overweight. This is higher than the



corresponding percentages for children in poor families not meeting the work standard (6 percent), working families with incomes between 100 percent and 200 percent of the poverty threshold (6 percent), and more affluent working families (5 percent).

Gifted status. As shown in Table 7.1, in 1997 children in working poor families have a significantly lower likelihood of being identified as gifted than either children in poor families not meeting the work standard or children in working families with incomes above the poverty threshold. Only 9 percent of children in working poor families have been identified as gifted. In contrast, 12 percent of children in poor families not meeting the work standard, 13 percent of children in working families with incomes between 100 percent and 200 percent of the poverty threshold and 20 percent of children in more affluent working families have been identified as gifted.
Repeated a grade. As shown in Table 7.1, in 1997 the percentage of children who have repeated a grade is similar (10-13 percent) for children in working poor families, children in poor families not meeting the work standard, and children in working families. However, only 6 percent of children in more affluent working families have repeated a grade.
School suspension or expulsion. As shown in Table 7.1, in 1997 the likelihood of a school suspension or expulsion decreases as income increases. Sixteen percent of children in working poor families have either been suspended or expelled from school. This percentage is not significantly different than for children in poor families not meeting the work standard (19 percent), but significantly higher than for children in working families with incomes between 100 percent and 200 percent of the poverty line (12 percent), and children in more affluent working families (9 percent).
Summary for variables measuring how well the child is developing. According to the limited number of indicators measuring how well a child is developing, children in working poor families generally appear to fare worse than children in more affluent working families in all three domains—health, cognitive development, and behavior. Moreover, children in working poor families are less likely than children in poor families not meeting the work standard to be identified as gifted and more likely to be overweight.
Television rules. As shown in Table 7.2, in 1997 the mean value of the television rules scale was in the range of 2.14 to 2.20 for all four groups of children. The slight difference in the score for children in working poor families and the scores for the other three groups is not statistically significant. In 1994 the range of variation was similarly small and not significantly different from 1997.
Meals with parents. As shown in Table 7.2, in 1997, the weekly number of meals with mother was lowest (9.61) for children in working poor families. This value is significantly lower than the corresponding values for children in poor families not meeting the work standard (10.04) and children in working families with incomes above twice the poverty threshold (9.83).
However, the pattern is somewhat different for meals with father. Similar to results for eating with mother, children in working poor families eat fewer meals per week with father (8.42) than children in poor families not meeting the work standard (8.97). However, children in working poor families eat significantly more meals with father than children in working families with incomes at least twice the poverty threshold (7.97).

This suggests a tradeoff between fathers’ eating meals with their children and success in

income-earning activities—although the difference in meals eaten together is relatively small.
Parental aggravation. As shown in Table 7.2, the mean value of the parental aggravation index for parents of children in working poor families is 9.32, which is slightly but significantly lower than for working families with incomes between 100 percent and 200 percent of the poverty threshold (9.49) and more affluent working

Table 7.2. Selected family measures related to how well a child is likely to develop, 1997 and 1994


1997

Index

(range)

Poor families not meeting work standard

Working poor families



Working families,

100-199% poverty



Working famili

>200% povert



Television rules (0-3) 2.16 2.14 2.20 2.16

Meals with mother (0-14) 10.04 9.61 9.79 9.83

Meals with father (0-14) 8.97 8.42 8.30 7.97

Parental aggravation (0-12) 9.19 9.32 9.49 9.48

Father involvement (0-12) 7.85 7.61 7.85 8.15

Maternal involvement (0-12) 8.56 8.61 8.79 9.03



Living apart from parents


Has lived apart from parents

5%

6%

6%

5%

Father's educational expectations for child













Leave School Before Graduation

18%

13%

7%

3%

Graduate from high school

13%

13%

10%

5%

Get some college experience

52%

57%

62%

61%

Graduate from College

18%

18%

21%

31%




1994
Index

(range)

Television rules

Poor families not meeting work standard

Working poor families



Working families,

100-199% poverty



Working famili

>200% povert



(0-3) 2.13 2.12 2.15 2.18

Living apart from parents

Has lived apart from parents 3% 2% 2% 2%

Sources: 1996 and 1993 Surveys of Income and Program Participation

families (9.48). This means a higher level of parental aggravation for working poor parents.


The small difference between parents of children in working poor families and parent of children in poor families not meeting the work standard is not statistically significant. Thus, parental aggravation seems to be somewhat more pronounced among poor families than among non-poor families.

Parental involvement. As shown in Table 7.2, the father involvement score for children in working poor families (7.61) was lower than each of the other three groups—showing children in working poor families at a distinct disadvantage for this measure. The mother involvement score for children in working poor families (8.61) was lower than for children in working families with incomes between 100 percent and 200 percent of the poverty threshold (8.79) and children in more affluent working families (9.03).
Living apart from parents. As shown in Table 7.2, the percentage of children who have lived apart from their parents ranges between 5 percent and 6 percent with no significant differences across the four groups.
Father’s educational expectations for child. As shown in Table 7.2, fathers of children in poor families (both working and not meeting the work standard) had substantially lower expectations for their children’s education. Nearly one in five fathers of children in poor families not meeting the work standard expected their child to drop out before graduating from high school. For fathers of children in working poor families, the percentage was somewhat lower—13 percent. However, only 7 percent of fathers of children in working families with incomes between 100 percent and 200 percent of the poverty threshold and 3 percent of fathers in more affluent working families held these expectations. Conversely, fathers in more affluent families were much more likely than their counterparts in poor families to expect their children to graduate from college.
Summary for variables measuring the home environment. For meals with father, parental involvement, parental aggravation, and father’s educational expectations for his child, children in working poor families appear to be at a disadvantage compared with at least one other group. On other measures of the home environment, we have observed little difference between children in working poor families and other children.
Participation in extracurricular activities. Children in poor families are less likely than families with higher incomes to participate in extracurricular activities. As shown in Table 7.3, the mean values of the index are 0.57 and 0.58 for children in poor families not meeting the work standard and children in working poor families, respectively.

However, the value of the index is substantially higher for children in working families with incomes between 100 percent and 200 percent of the poverty threshold (.77) and higher still (1.18) for children in working families with incomes above 200 percent of the poverty threshold. These differences are statistically significant.


Results for 1994 follow the same pattern across the different types of families. However, the level of participation for children in working families with incomes at least twice the poverty line was higher in 1994 than in 1997. Participation did not change significantly for the other three groups.
School engagement. As shown in Table 7.3, the mean value for the school engagement index varies insignificantly between 4.70 and 4.79 across the poor and near-poor groups. However, children in working families with incomes above twice the poverty line have significantly higher school engagement (4.96).
Parental attitudes towards the community. As shown in Table 7.3, the parents of children in working poor families scored significantly higher on the positive index of attitudes towards the community (13.44) than poor families not meeting the work

standard (12.55). However, they scored significantly lower than children in working families with incomes higher than 200 percent of the poverty threshold (14.73). The difference between children in working poor and working near-poor families was marginally significant.26


The non-comparable 1994 version of this index follows the same pattern across the four groups.
As shown in Table 7.3, a similar pattern emerged for the negative index of attitudes towards the community. The mean value was 3.79 for children in poor families not
Table 7.3. Selected community measures related to how well a child is likely to develop, 1997 and 1994


1997

Index

(range)

Poor families not meeting work standard

Working poor families




Working families,

100-199% poverty




Working famili

>200% povert



Extracurricular activities (0-3)

0.57

0.58

0.77

1.18

School engagement (0-6)

4.70

4.75

4.79

4.96

Parent's positive attitude toward













community (0-20)

12.55

13.44

13.66

14.73

Parent's negative attitude toward

community (0-8)




3.79


4.18


4.30


4.96

Child care services













Ever received services

29%

31%

35%

47%

Age child care first provided













0-3 months

14%

11%

18%

27%

4-6 months

9%

11%

10%

15%

7-11 months

4%

3%

5%

6%

12-17 months

8%

6%

8%

7%

18-23 months

4%

4%

5%

5%

24 or more months

61%

65%

54%

41%

School attendance













Ever attended kindergarten

83%

83%

85%

89%

Enrolled in private school

4%

4%

7%

13%

Religious affiliation

2%

3%

5%

10%


1994

Index

(range)

Extracurricular activities (0-3)

0.54

0.58

0.81

1.28

Parent's positive attitude toward community (0-10)


5.38


6.10


6.53


7.19

Parent's negative attitude toward

community (0-10)




4.72


5.64


6.52


7.52


Sources: 1996 and 1993 Surveys of Income and Program Participation

26 By marginally significant, we mean the difference was significant at the .10 level but not at the .05 level.

meeting the work standard, 4.18 for children in working poor families, 4.30 for children in near-poor working families, and 4.96 for children in more affluent working families. The value of the index for working poor families is significantly higher than for poor families not meeting the work standard and significantly lower than for both groups of more affluent working families.
The non-comparable 1994 version of this index follows the same pattern across the four groups.
Use of child care services. When compared with more affluent working families, children in working poor families and poor families not meeting the work standard are less likely to ever receive child care services, and, among those that do, likely to receive

it at a later age. As shown in Table 7.3, children in poor families—both those in working families and those in families not meeting the work standard—are much less likely (29-

31 percent) than children in working families with incomes above twice the poverty threshold (47 percent) to ever receive child care services. Among children in working poor families who do receive child care, the percentage who first receive child care services during the first and second three months of life (when receipt of child care may be problematic) is substantially lower than for children in working families with incomes above two times the poverty threshold. Only 11 percent of such children in working poor families first received child care during the first three months, compared with 27 percent of children in working families with incomes above twice the poverty line. During the

next three months 11 percent of children ever receiving care and in working poor families first received child care, compared with 15 percent of children ever receiving care and in working families with incomes above twice the poverty line. By age 9 months, only one- quarter of such children in working poor families had received child care expenses, but nearly one-half of such children in more affluent families had done so.


Conversely, 65 percent of working poor children ever receiving care first received child care at ages 24 months or older, compared with 41 percent of children in families with income at least twice the poverty line.
Although there are some noticeable differences in the precise times at which children first received child care in 1994, the pattern is the same. Children in poor families begin to receive child care services significantly later than children in families with incomes at least twice the poverty line.
Attendance at kindergarten and private and religious schools. Children in working families with incomes above the poverty threshold are more likely than children in working poor families or children in poor families not meeting the work standard to have attended kindergarten, private school, or a school with a religious affiliation. As shown in Table 7.3, 83 percent of children in working poor families and poor families not

meeting the work standard ever attended kindergarten. In contrast, 85 percent of children in near-poor families and 89 percent of children in working families with incomes over twice the poverty threshold have done so. The corresponding statistics for enrollment in private school are 4 percent, 7 percent, and 13 percent, respectively. Only 2 percent of children in poor families not meeting the work standard and 3 percent of children in working poor families are enrolled in schools with a religious affiliation. In contrast, 5 percent of children in near-poor working families, and 10 percent of children in working

families with incomes over twice the poverty line are enrolled in schools with a religious affiliation.


Summary for variables measuring interaction with the community. We have observed several differences in the degree of interaction with their community among children in working poor families versus other children. Children in working poor families were less likely than more affluent working families to participate in extracurricular activities. However, differences in school engagement were not statistically significant. Parents in working poor families were less likely than parents in more affluent working families to hold positive attitudes towards their community and more likely to hold negative views. Finally, children in working poor families were less likely than children in more affluent working families to receive child care services at an early age, to attend kindergarten, private school, and religious school. Thus, with the exception of school engagement, the degree of interaction of working poor families (and other poor families) was less than for working families with incomes at twice the poverty threshold.
Results by TANF leaver/stayer status
Overweight status. As shown in Table 7.4, in 1997 among children whose families left welfare 7 percent were overweight. This is not significantly different from the corresponding percentage for children whose families stayed on welfare (5 percent).
Gifted status. As shown in Table 7.4, in 1997 10 percent of children whose families left welfare and 10 percent of children whose families stayed on welfare were identified as gifted. Thus, there is no difference for this measure.

Table 7.4. Percentage of children by overweight status, gifted student status, grade repetition status, and school suspension/expulsion status of children by whether family has left welfare or stayed on welfare, 1997




Families leaving welfare between

1996 and 1997

Families on welfare in both 1996 and

1997

Overweight 7% 5% Gifted student 10% 10% Repeated a grade 13% 12% Suspended/expelled from school 21% 23%
Source: 1996 Survey of Income and Program Participation

Repeated a grade. As shown in Table 7.4, in 1997 13 percent of children whose families left welfare had repeated a grade. This is not significantly different from the 12 percent who repeated a grade from among children in families who stayed on welfare.


School suspension or expulsion. As shown in Table 7.4, in 1997 21 percent of children in families leaving welfare had been suspended or expelled from school. This is not significantly different from the 23 percent for children from families who stated on welfare.

Summary for variables measuring how well the child is developing. According to the limited number of indicators measuring how well a child is developing, we found no statistically significant difference between welfare leavers and welfare stayers.
Television rules. As shown in Table 7.5, in 1997 the mean value of the television rules index was in the range of 2.04 to 2.11 for both groups of children. The slight difference in scores for children in stayer and leaver families is not statistically significant.
Meals with parents. As shown in Table 7.5, in 1997, the weekly number of meals with mother was significantly lower (9.10) for children in welfare leaver families than for children in welfare stayer families (10.11)—not surprising in view of the additional work responsibilities of those leaving welfare.
The difference in the number of weekly meals with father was not statistically significant—in the range of 8.68-9.03.
Parental aggravation. As shown in Table 7.5, the mean value of the parental aggravation index for parents of children in welfare leaver families is 8.84, which is slightly but significantly lower than for welfare stayer families (9.15). Thus, parental aggravation seems to be somewhat more pronounced among leaver families than among stayer families.

Table 7.5. Selected family measures related to how well a child is likely to develop, by whether family left welfare or stayed on welfare, 1997




Families leaving Index welfare between (range) 1996 and 1997



Families on welfare in both 1996 and

1997

Television rules (0-3) 2.04

2.11

Meals with mother (0-14) 9.10

10.11

Meals with father (0-14) 8.68

9.03

Parental aggravation (0-12) 8.84

9.15

Father involvement (0-12) 7.97

7.24

Maternal involvement (0-12) 8.58

8.33

Living apart from parents

Has lived apart from parents

5%

4%

Father's educational expectations for child







Leave School Before Graduation

10%

21%

Graduate from high school

13%

12%

Get some college experience

54%

50%

Graduate from College

23%

17%


Sources: 1996 Survey of Income and Program Participation

Father involvement. As shown in Table 7.5, the father involvement score for children in welfare leaver families (7.97) was higher than for welfare stayer families (7.24)— showing children in leaver families at an advantage for this measure.


Living apart from parents. As shown in Table 7.5, the percentage of children who have lived apart from their parents ranges between 4 percent and 5 percent with no significant differences between the leaver and stayer groups.
Father’s educational expectations for child. Fathers in families leaving welfare are only half as likely (10 percent) as fathers in families staying on welfare (21 percent) to expect that their child will leave school before graduation. The apparently substantial difference between fathers in leaver and stayer families who expect that their child will graduate from college is not statistically significant.
Summary for variables measuring the home environment. For meals with mother, father involvement, and father’s educational expectations for his child, children in welfare leaver families appear to be at an advantage compared with children in stayer
Table 7.6. Selected community measures related to how well a child is likely to develop, by whether family left welfare or stayed on welfare, 1997



Index


(range)

Families leaving welfare between

1996 and 1997

Families on welfare in both 1996 and



1997




Extracurricular activities (0-3)

0.57

0.54

School engagement (0-6)

4.59

4.57

Parent's positive attitude toward







community (0-20)

12.59

12.77

Parent's negative attitude toward community (0-8)


3.87


3.48

Child care services







Ever received services

39%

35%

Age child care first provided







0-3 months

13%

7%

4-6 months

11%

12%

7-11 months

4%

3%

12-17 months

8%

8%

18-23 months

4%

4%

24 or more months

60%

67%

School attendance







Ever attended kindergarten

85%

82%

Enrolled in private school

5%

6%

Religious affiliation

2%

6%

Source: 1996 Survey of Income and Program Participation



families. In contrast, they appear to be at a slight disadvantage relative to stayer families according to the parental aggravation index. On other measures of the home

environment, we observed little difference between children in leaver and stayer families.


Participation in extracurricular activities. Children in welfare leaver and welfare stayer families are about equally likely to participate in extracurricular activities. As shown in Table 7.6, the mean values of the index are 0.54 and 0.57, respectively, which is not a significant difference.
School engagement. As shown in Table 7.6, the mean value for the school engagement index varies insignificantly between 4.57 and 4.59 across the leaver and stayer groups.
Parental attitudes towards the community. As shown in Table 7.6, the parents of children in leaver and stayer families scored in the range of 12.59-12.77 on the positive index of attitudes towards the community. The difference is not statistically significant. However, parents of children in leaver families scored significantly higher (3.87) than parents of children in stayer families (3.48) on the negative index of attitudes towards the community. This indicates that the parents in the leaver families feel significantly less negative towards their community than parents in the stayer families.
Use of child care services. As shown in Table 7.6, the percentage of children in leaver and stayer families that ever received child care services is in the range of 35-39 percent.27 Not surprisingly, children in families leaving welfare were significantly more likely (13 percent) to use child care services during the first three months of life than children in families staying of welfare (7 percent). Among those that ever received child care, between 60 percent and 67 percent of children in leaver and stayer families receive it at 24 months of age or greater. The apparent difference in this statistic between the leaver and stayer families is not statistically significant.
Attendance at kindergarten and private and religious schools. As shown in Table 7.6, in 1997, over 80 percent of both children in families leaving welfare and children in families staying on welfare had ever attended kindergarten. The apparent difference between the percentages attending in the two groups is not statistically significant.
Only 5-6 percent of both children in families leaving welfare and children in families staying on welfare attended private school in 1997. Again the difference between the two groups is not statistically significant.
However, children in leaver families were significantly less likely (2 percent) than children in stayer families (6 percent) to have attended religious school.
Summary for variables measuring interaction with the community. We have observed only a few differences in the degree of interaction with their community among children in leaver families versus children in stayer families. Children in leaver families were less likely than children in stayer families to have parents with negative attitudes towards the community and less likely to attend religious schools. They were more likely than children in stayer families to receive child care services in the first three months of life.

27 The difference between the percentages for stayer and leaver families is not statistically significant.

Summary
Our results for indicators measuring child wellbeing or factors related to child wellbeing are summarized in Table 7.7 and described below.


Children in working poor families versus other children
According to the limited number of variables directly measuring how well a child is developing, children in working poor families appear to fare worse than children in more affluent working families across all three domains—health, cognitive development, and behavior. They also are less likely than children in poor families to be identified as

gifted and are more likely to be overweight.


In the area of home environment, children in working poor families also seem to be at a disadvantage compared to at least one other group on the following measures: meals with father, parental involvement, parental aggravation, and father’s educational expectations for his child.
Finally, in the area of interaction with the community, children in working poor families were less active in extracurricular activities than more affluent children and less likely to attend kindergarten, private school, or religious school, while their parents were less likely to hold positive views of their community and more likely to hold negative views.
Children in TANF leaver families versus children in stayer families
We found no difference between TANF leaver and stayer families in variables measuring how well a child is developing.
However, there is a mixed picture in the area of home environment. For meals with mother, father involvement, and father’s educational expectations for his child, children in leaver families appear to be at an advantage compared with children in stayer families. On the other hand, appear to be at a slight disadvantage according to the parental aggravation index.
Finally, there are only a few differences between TANF leaver and stayer families for variable measuring interaction with the community. Parents in leaver families were less likely than parents in stayer families to hold negative attitudes towards the community and were less likely to attend religious schools. They were also more likely to receive child care services during the first three months of life.


Table 7.7. Significant differences in selected measures between children in working poor families versus children in three other family types; and between children in families leaving welfare

versus families staying on welfare, 1997






Working poor families versus

Welfare leavers versus


Poor families not meeting work standard

Working families:

Income 100%-200% of poverty threshold

Working families:

Income over 200% of poverty threshold


Families staying on welfare

How well is child developing?

Overweight status Gifted status Repeated a grade

School suspension/expulsion

-

-


-

-



-

-

-



-

-





Home environment Meals with mother Meals with father Parental aggravation Father involvement Mother involvement

Living apart from parents Father expects child to leave school before graduation

-

-



-

+


-

-

-



-

-

+



-

-

-



-

-

-



+

+


Interaction with community Extracurricular activities School engagement

Parental attitudes (positive) Parental attitudes (negative) Ever use child care services Kindergarten attendance Private school attendance

Religious school attendance



+

+

-

-

-



-

-

-



-

-

-



-

-

-



-

-

-



+


Source: Tables 7.1 - 7.6

78

Chapter 8. Discussion

Summary
Discussion
In Chapter 1, we noted that welfare reform explicitly includes two major purposes. The first is to increase the amount of work performed by adults currently on welfare. The second is to decrease child poverty. In our discussion of the findings of this report, we focus on these two purposes in turn.
Negotiating the transition from welfare to work
In recent years, welfare caseloads have plummeted. In 1996, the monthly average AFDC caseload was 12.6 million recipients in 4.6 million families (U.S. Department of Health and Human Services, 1998). In fiscal year 2001, there were only 5.5 million recipients in

2.1 million families (U.S. Department of Health and Human Services, 2002c). It is especially striking to note that nationwide the welfare caseload decline has continued even as the economy has weakened.


Our SIPP analysis shows that, even during 1997, the first year of federal welfare reform, work activity of parents that were TANF recipients in 1996 increased substantially—both in terms of hours worked per week and weeks worked per year. Moreover, adult TANF recipients were substantially more likely to be employed in 2000 than in 1996. In 1996 only 11 percent of recipients were employed; by 2000, 26 percent were employed (U.S. Department of Health and Human Services, 2002a). This provides evidence that welfare reform is promoting increased work activity on the part of welfare recipients, in spite of their disadvantages in competing in the labor market.
It should be noted, however, that the 1996-2000 increase in work occurred in the context of an unusually robust economy with unemployment rates lower than they have been since 1969 (during the Vietnam War). The average monthly unemployment rate in 1996

was 5.4 percent, while by 2000 the rate dropped to 4.0 percent. However, since 2000, the unemployment rate has risen to as high as 6.0 percent in April 2002 (U.S. Department of Labor, 2002). Thus, it isn’t clear if the increase in employment by welfare recipients continued past 2000.


From an economic standpoint, the increase in work activity is good news since, as we have seen, regardless of family structure, race/ethnicity, and parental education, living in a working family dramatically reduces the likelihood of a child being in poverty. Moreover, we also observed that families receiving TANF in 1996 not only increased their work activity but their incomes as well.
However, maternal employment does not necessarily have a benign impact on at least some groups of children. According to the only experimental studies which have been performed to date (and prior to federal welfare reform, many adolescent children were negatively affected by welfare-to-work programs (Brooks, Hair, & Zaslow, 2001). Negative impacts included increased consumption of alcohol and cigarettes and decreased academic performance (Duncan & Chase-Lansdale, 2000). However,

excluding adolescents, other studies have found neutral or positive effects on children in families receiving welfare, unless a family’s economic situation deteriorated. For example, The academic performance, health, and behavior of children whose parent participated in welfare-to-work programs changed little (Haskins, Sawhill, & Weaver,

2001; Zaslow, Moore, Brooks et al., 2002). Researchers found some positive impacts for programs that either increased the mother’s education or the family’s income (Zaslow, Moore, Brooks et al., 2002).
Non-experimental studies focused on federal welfare reform also paint a mixed picture. In Chapter 7, we found that children whose families left welfare between 1996 and 1997 fared somewhat worse on parental aggravation than children whose families stayed on welfare both years. On the other hand, children in welfare leaver families ate meals with their mother more frequently, and father involvement was higher. These differences, however, may be due to selection rather than an effect of leaving welfare. Moreover, there were no significant differences for several other measures.
A similar study using the National Survey of American Families (NSAF) found that, with two exceptions, there was no significant difference between children in welfare leaver

and welfare stayer families in social behavior, engagement in school activities and health. The two exceptions were adolescents likelihood of suspension or expulsion from school (leavers more likely) and the presence of a health condition that limits activity (stayers more likely) (Zaslow, Moore, Tout, Scarpa, & Vandivere, 2002).


The mixed findings regarding maternal employment patterns and child wellbeing suggest that additional research is urgently needed to better understand when employment is associated with negative outcomes for children in some circumstances. It also suggests that continuing collections of data on child well-being in national surveys and also at the state level could be extremely useful in monitoring the continuing progress of welfare reform.
Although meeting the work standard is strongly associated with a decreased risk of poverty for children, we found that, among poor children, 64 percent of children in married-couple families had parents who met the work standard (the equivalent of one full-time worker) in 2000, while 38 percent of children in single-mother, families had a parent who met the work standard (the equivalent of half-time work). Thus, meeting the work standard provides no guarantee of a child’s escaping poverty.
Assuming that jobs were available, virtually all of the children in the married-couple, poor families could have moved above the poverty line if the parents had worked more hours. An obvious question is why the wives in these married couples did not increase their hours worked. Our analysis provides a partial answer to this question—they did, but perhaps not enough for the family to escape poverty.
Among poor married-couple families not meeting the work standard in 1996, there were substantial 1996-1997 decreases in the percentage of mothers who did not work at all and increases in the percentage of mothers who worked full-time and full-year. Nonetheless, among these families, over half of these mothers did not work at all in 1997, only one- quarter worked full-time, and only 11 percent worked all year.

The low hourly wage rates earned by a large percentage of the better paid of these parents may explain why they didn’t work more. One of the parents may feel that the value of their uncompensated time at home exceeds the value of the extra income they would earn if they worked for pay--especially if that work requires expenditures on child care. We noted in Chapter 4 that, for children in married-couple, working poor families whose parents paid for child care, around two-thirds paid 30 percent or more of their family income for care. Another possibility is that some of these married parents might be working “off the books,” an apparently common practice among poor single mothers (Edin & Lein, 1997).


This same tradeoff applies to single parents. However, with one earner and relatively low wage rates, it is much more difficult for a single parent family to rise above the poverty threshold than it is for a married-couple family.
The problem of low wage rates deterring people from working enough to rise out of poverty is likely to apply even more strongly to welfare recipients who go to work. A summary of studies of welfare recipients who found jobs in twelve states28 found that “employed former recipients and recipients combining work and welfare typically are paid less than $8 per hour and a substantial portion earn less than $6 per hour” (Parrot,

1998). Although they tend to work more than 30 hours per week, recipients who find jobs typically earn between $8,000 and $10,800 annually, due, presumably, to periods of joblessness. Similar findings are reported for a group of employed welfare recipients in Boston, Chicago, and San Antonio (Moffitt & Cherlin, 2000).


We found no consistent pattern that educational attainment played a significant role in determining whether poor, single-mother or married-couple families worked enough to meet the work standard. However, a recent analysis of experimental programs offered to welfare recipients prior to federal welfare reform suggests that human capital development programs can increase the educational attainment of welfare recipients, which, in turn, increases the cognitive achievement of their children (Zaslow et al.,

2001). This suggests that policy makers should continue to explore the payoff from adult education. 29


Current welfare policy strongly favors moving welfare recipients into work-related activities (“Work First”) rather than remedial education. The Administration’s current welfare reauthorization proposal would strengthen the focus on work by increasing the number of hours per week that welfare recipients are expected to work. Given the already heavy demands on the time of working single mothers, it would seem unlikely that many of them would be able to devote time to obtaining a high school diploma or a post-secondary education. Thus, changing the economic incentives to work (such as

providing more subsidized day care or providing an even more generous EITC) represent additional potential strategies to move couples to become economically independent. Work requirements might also be made to vary by age, given the greateer costs for chld care for young children and young children’s greater need for parental time.



28 California, Delaware, Florida, Georgia, Indiana, Maryland, Minnesota, Michigan Ohio, Oregon, South Carolina, and

Wisconsin.


29 Research, mainly done for men, suggests that the payoff for GEDs is low (Murnane, 1995).

Child care subsidies can also encourage families with low incomes to increase their work effort. The welfare reform legislation increased federal spending on child care assistance in 1997 by an estimated 27 percent over prior law (Long & Clark, 1997). However, it did so by consolidating several key federal child care assistance programs for low income families into a single block grant entitled the Child Care and Development Fund (CCDF). CCDF gives states much more autonomy in both setting total child care spending and in how both the federal and state money is spent. Since federal welfare reform was enacted, combined state and federal spending on CCDF expanded substantially from $5.3 billion

in fiscal year (FY) 1998 to $9.7 billion in FY 2001 (U.S. Department of Health and Human Services, 2001a, 2002b). However, the emergence of budget deficits at both the federal and state levels may force freezes or cuts in the future.
The Earned Income Tax Credit (EITC) also encourages increased work effort for parents that with low family income. For families with one child and incomes below $7,100 or families with two or more children and incomes below $10,000, the EITC acts as a wage rate subsidy, with the federal government paying a percentage refund for each additional dollar of earnings. However, for one-child families with incomes in the range of $13,000 to $28,250 or two-or-more-children families with incomes in the range of $13,000 to

$32,000, the benefit is phased out. Thus, the EITC discourages additional work effort by, in effect, imposing a tax surcharge on earnings. Although this phase-out problem is probably inevitable, the point at which the EITC begins to act as a disincentive to work could be raised to a higher level of income (thus, reducing the effective rate of taxation



on the earnings of parents with very low earnings). However, this would increase the cost of the program.
Until recently, increasing parental work effort and leaving welfare could lead to loss of health insurance eligibility. We have seen in Chapters 4 and 5 that, as recently as 1996, children in working poor families were less likely to have health insurance coverage than either children in poor families not meeting the work standard or children in other working families. However, federal and state policies on health insurance have been rapidly changing over the past few years, and there is now a complex patchwork of eligibility criteria for publicly subsidized insurance for children in working poor families. Under present law, low-income children qualify for Medicaid coverage in all states if: (1) they are children less than age 6 with family incomes below 133 percent of the official poverty line, (2) they are children ages 6 to 18 with income below the official poverty line, or (3) they are Title IV-E foster care children or adoption assistance children (U.S. House of Representatives Committee on Ways and Means, 2000). States can also

provide (but are not required to provide) Medicaid coverage for children whose families meet more generous income cutoffs.


The State Child Health Insurance Program (SCHIP) provides another way for states to obtain a federal subsidy to provide health insurance coverage for children. As of January

1, 2000, the Department of Health and Human Services had approved SCHIP programs for all 50 states and the District of Columbia (U.S. House of Representatives Committee on Ways and Means, 2000). In 24 of these states, federal SCHIP funds are being used to finance eligibility expansions for Medicaid; 15 run their own programs; and 17 have combined a Medicaid expansion and a separate state program. SCHIP allows states to cover children in families not eligible for either Medicaid or a group health plan provided



their incomes are either (1) above the state’s 1997 Medicaid eligibility standard as of March 1997 but under 200 percent of the poverty threshold or (2) within 50 percentage points of the 1997 Medicaid standard if the Medicaid standard is already at or above 200 percent of the poverty threshold (U.S. House of Representatives Committee on Ways and Means, 2000). Within this framework states can set their own rules.
With this patchwork of coverage, as we saw, in 2000, health insurance coverage rates were comparable for working poor families and poor families not meeting the work standard—although still substantially lower than for more affluent working families.
Negotiating the transition out of poverty
Federal and state welfare reform legislation seems to be meeting with success in moving welfare recipients into work activities in spite of their competitive disadvantage in the labor market. However, the next step of moving working poor families out of poverty seems to be much more difficult. There is a vast gulf between poor families (both working and not meeting the work standard) and working nonpoor families in both educational attainment and hourly wage rates.
A high school diploma is much more common for at least one parent in working families with incomes between 100 percent and 200 percent of the poverty threshold and nearly universal for at least one parent in working families with incomes above 200 percent of the poverty line. Indeed a majority of the latter group of parents have at least some college. Similarly 87 percent of the better paid parents in married-couple families and 81 percent of the mothers in single-mother families with incomes at least twice the poverty line are earning more than $10.00 per hour.
This suggests that moving children from the ranks of both working poor families and

poor families not meeting the work standard (especially single-parent families) above the poverty line and higher may require a substantial investment in the human capital of their parents. This investment may be very difficult to undertake if we are expecting these parents to be making a very substantial work effort at the same time (which is the case under the “Work First” philosophy of the welfare reform legislation). Moreover, as noted earlier in this chapter, policy-makers should note the results of the JOBS experiment focused on remedial adult education which suggests that a public investment in the education of parents in poor families may be worth considering.


Children in working poor families are at a substantial disadvantage compared with children in non-poor working families, and this disadvantage is not just economic. Their parents are substantially less educated; they are less likely to be covered by health insurance; and they are less likely to live in owner-occupied housing. With respect to health insurance, as of 2000, they were less likely than children in more affluent families to be covered than children in poor families not meeting the working standard. When children in working poor families are enrolled in paid child care, the cost of this child care often consumes a substantial percentage of family income. Finally, we found that

children in working poor families tended to be worse off in all three domains examined— health, cognitive development, and behavior—when compared with families with higher incomes.



There are several ways in which adults in working poor families with children could increase their income to the point that they escape poverty. One way is to work more hours. However, even assuming that jobs are available, families are likely to face increased child care expenses, which may largely offset the extra income—especially if the extra hours earn low wages. In addition, as explained below, these extra wages will be taxed at a high marginal rate if income is in the “phase-out” interval of the earned income tax credit.
A second way is to obtain jobs paying a higher wage rate. Sometimes, obtaining and holding on to a job may generate a higher wage rate over the long run. This is especially likely to be true if the employee receives on-the-job training, which increases work productivity and may possibly lead to promotions, and we found that the parents of children in working poor families did substantially increase their incomes between 1996 and 1997. However, low-wage jobs do not necessarily lead to an upward career path. Often additional formal education is required. However, as noted earlier, it may be difficult for persons already making a substantial work commitment to find the time to become better educated.
A third plausible way to help the children in these families escape poverty is to encourage marriage for single parents. Marriage provides a family with at least the potential for two earners, and two parents working full-time can generally escape poverty as measured by the official standard. As we have seen, however, there are many married- couple families meeting the work standard that are nonetheless in poverty. Moreover, there is general agreement across the political spectrum that a healthy marriage requires more than an ample family income.
However, it is worth noting that many of the programs that have been set up to provide assistance to low-income families with children weaken the economic incentive for a single parent to get married. One of the most important programs causing incentive problems is the earned income tax credit (EITC), but the phase-out of Food Stamps and Medicaid cause similar problems. For example, a single mother earning the maximum EITC is likely to lose much or even all of the family’s EITC if she marries a working husband and continues to work herself. This same feature of the EITC also reduces the incentive for parents in the “phase-out” interval to either increase their hours worked or invest in education or training to increase their wage rates. Consider a single parent considering whether to increase her earnings by increasing her hours worked.
The EITC phaseout “tax” is imposed on top of a federal payroll tax, a federal income tax, and (in most states) a state income tax. All told she is likely to keep only about half of every additional dollar she earns. Out of those extra net earnings, she may also have to pay for extra hours of child care services.
The discussion above only addresses the effects of the tax system. The needs-tested transfer system, including TANF, Food Stamps, and Medicaid, have similar phase-out problems. As income increases, benefits either gradually or abruptly are terminated. Thus, in certain circumstances, families may actually be made worse off by earning more
Although it is impossible to eliminate altogether the “phase-out” problems discussed above, it is possible (at a cost to the federal treasury) to adjust the ranges over which they

take place and/or reduce the “tax rates” that they implicitly impose. For example, providing two-earner couples with a $5,000 increase in their federal income tax standard deduction (Wheaton, 1998) would move the phase-out range for the EITC up the income scale by $5,000. Thus, its disincentive effects wouldn’t apply to families still at or just above the poverty threshold. Raising the threshold for a child’s eligibility for Medicaid (which has already been done, although not uniformly across the states or across children’s ages) is another way of avoiding imposing penalties on low-income families for increasing their work effort and/or wages.


In short, while policy-makers’ attention has been focused on getting parents in welfare families to increase their work effort, they must also focus their attention on how to help both welfare families and working poor families out of poverty.

Appendix A. Imputing child care costs
Our goal is to impute annual child care expenses onto the Current Population Survey (CPS) using a four-equation model estimated using the 1996 panel of the Survey of Income and Program Participation (SIPP). As explanatory variables, the model uses variables that are common to both SIPP and the CPS. The model is based on a specification developed by the U.S. Census Bureau (Short et al., 1999).
The first two models (see Table 1) predict whether a family incurred child care expenses separately for single-parent families and married-couple families. The second two models (see Table 2) predict child care expenses for those families that incurred child care expenses—again, separately for single-parent and married-couple families.
Imputation takes place in two steps. Using the corresponding explanatory variables on the1995-2000 CPS files and using the models displayed in Table 1, we first predict the probability pi that family i incurs child care expenses for each CPS family with children. Then we draw a random number ri from a uniform distribution ranging between 0.0 and

1.0. If ri is less than or equal to pi, then the family is imputed to have child care expenditures. Otherwise, the family is imputed not to have child care expenditures.


For those families that are imputed to have child care expenditures, using the models displayed in Table 2, we next predict the amount of child care expenditures by calculating the predicted value of child care expenditures.
More formally:
Let yi = predicted value of SIPP logit child care equation for ith family on CPS file
Let pi = probability of having child care associated with yi
e yi

pi = 1 + e yi

ri = a random number drawn from a uniform distribution with range {0,1}



TABLE 1. LOGIT COEFFICIENTS PREDICTING WHETHER A FAMILY INCURRED CHILD CARE EXPENSES


Explanatory variable

Single-parent family

Married-couple family

Intercept

-1.8745

-7.0582

Black family head

-0.3360

-0.3034

Hispanic family head

-0.6423

0.0935

Number of children

0-5 years old

6-11 years old

12-15 years old

16-18 years old



0.2300

0.0370

-0.6067

-1.0659


0.4727

-0.0612

-0.5881

-0.6477

Census region Midwest South

West


0.3450

0.1032

0.4261


0.1603

0.2770

0.00661

Log of family income

0.1316

0.5760

Percent of family’s income earned by mother

-----



2.3844

Likelihood ratio Chi square

149.9086

919.4380

Number of observations

2301

5,156

Source for weighted regression: 1996 SIPP: SAS run entitled “Weekly child care expenses for 1-parent family,” and



“Weekly child care expenses for 2-parent family,” by Melissa Long, 11:17, December 14, 2001, pp. 7-8.
TABLE 2. ORDINARY LEAST SQUARES REGRESSION COEFFICIENTS FOR CHILD CAR E EXPENDITURES

Explanatory variable

Single-parent family

Married-couple family

Intercept

-68.82054

-246.88382

Black family head

-4.35448

-6.47595

Hispanic family head

-2.70484

-4.02282

Number of children

0-5 years old

6-11 years old

12-15 years old

16-18 years old



20.57217

9.90966

-11.04150

-19.75778


34.02771

4.51251

-6.57538

-9.86443

Census region Midwest South

West


-7.60990

-6.72456

11.11121


-14.35430

-10.17148

-10.53331

Log of family income

13.33587

28.70627

Percent of family’s income earned by mother

---------




52.33844

Adjusted R2

.0870

.2128

Number of observations

757

1,854

Source for weighted regression: 1996 SIPP: SAS run entitled “Weekly child care expenses for 1-parent family,” and



“Weekly child care expenses for 2-parent family,” by Melissa Long, 11:17, December 14, 2001, pp. 7-8.

Let Ci = 1 if family is imputed to have child care expenditures and Ci = 0 if family is imputed not to have child care expenditures. Then



Ci = 1 if ri ≤ pi
and Ci = 0 if ri f pi

If Ci = 1, let zi = predicted value of SIPP OLS child care expenditure regression for ith



family on CPS file.
Let Xi = imputed child care expenditures of the ith family.

X i = zi


Adjustment of imputation equations to replicate SIPP findings
When the equations described above were used to impute child care expenses onto the CPS, neither the percentage of working families with child care expenses nor mean expenditures for the working families exactly agreed with the original findings from SIPP (see attached spreadsheet for summary). The CPS percentage of working families with child care expenses was about 1.7% lower than the corresponding percentage on SIPP; and the mean CPS expenditures for those families with expenditures was about 20

percent lower than on the SIPP.


To fix this problem we use the following adjustment process based on the initial findings. Let p*i = the adjusted probability of a family having child care expenses, and

Xi* = adjusted child care expenditures.


p *i

= pi ⋅1.017



X *i = X i ⋅1.218
Imputations in years other than 1997
We want our equation to work in constant dollars and have adjustments for changes in the cost of living to be taken care of outside the equations. We accomplish this as follows:
Let cpit be the value of the consumer price index-urban (CPI-U) in year t.
Let yit be the value of personal income for the ith family in year t in current dollars.

Instead of using yit in the two equations, use


yˆ it (personal income in 1997 dollars), where


yˆ it =
y cpi1997

it cpi

t
Then, in the second equation, after imputing child care expenditures in 1997 dollars (Ĉit), convert child care expenditures to current dollars as follows:


Cit
= Cˆ
it
cpit

cpi1997

REFERENCES
Aquilino, W. S. (1996). The life course of children born to unmarried mothers: Childhood living arrangements and young adult outcomes. Journal of Marriage and the Family, 58, 293-310.

Bowlby, J. (1973). Separation. London: Hogarth Press and the Institute of

Psychoanalysis.

Brooks, J. L., Hair, E. C., & Zaslow, M. J. (2001). Welfare reform's impact on adolescents: Early warning signs (Research Brief). Washington, DC: Child Trends.

Citro, C. F., & Michael, R. T. (Eds.). (1995). Measuring poverty: A new approach.

Washington, DC: National Academy Press.

Cornell, D. G. (1995). Achievement and self-concept of minority students in elementary school gifted programs. Journal for the education of the gifted, 18(2), 189-209.

Croan, T., Hatcher, J., Long, M., & Wertheimer, R. F. (2002). Children in working poor families: a review of the literature. Washington, DC: Child Trends.

Dalaker, J. (2001). Poverty in the United States: 2000, Current Population Reports, Series P60-214. Washington, DC: U.S. Government Printing Office.

Del Gaudio Weiss, A. M., & Offenberg, R. m. (2001). Enhancing urban children's early success in school: the power of full-day kindergarten. Philadelphia: School District of Philadelphia.

Duncan, G., & Brooks-Gunn, J. (1997). The consequences of growing up poor. New

York: Russell-Sage.

Duncan, G. J., & Chase-Lansdale, P. L. (2000). Welfare reform and child well-being.

Chicago, IL: Northwestern University.

Eccles, J. S., & Barber, B. L. (1999). Student council, volunteering, basketball, or marching band: What kind of extracurricular involvement matters? Journal of Adolescent Research, 14, 10-43.

Edin, K., & Lein, L. (1997). Making ends meet. New York: Russell Sage Foundation. Ehrle, J., Moore, K. A., & Brown, B. (1999). Children's environment and behavior: high

engagement in school. 1997 snapshots of America's families. Washington, DC: The Urban Institute.

Gidding, S., Leibel, R., Daniels, S., Rosenbaum, M., Van Horn, L., & Marx, G. (1996).

Understanding obesity in youth. American Heart Association Medical/Scientific

Statement, http://circ.ahajournals.org/chi/content/full/94/12/3383.

Greenberg, M. H., & Savner, S. (1996). A detailed summary of key provisions of the Temporary Assistance for Needy Families block grant of H.R. 3734. Washington, DC: Center for Law and Social Policy.

Haskins, R., Sawhill, I., & Weaver, K. (2001). Welfare reform: An overview of effects to date. Washington, DC: The Brookings Institution.

Haveman, R., & Wolfe, B. (1994). Suceeding generations. On the effects of investments in children. New York, NY: Russell Sage Foundation.

Heubert, J. P., & Hauser, R. M. (1999). High stakes: testing for tracking, promotion, and graduation. Washington, DC: National Academy Press.

Hofferth, S., Pleck, J., Stueve, J., Bianchi, S., & Sayer, L. (2001). The demography of fathers: What fathers do. In C. Tamis-Lamonda (Ed.), The Handbook of Father Involvement. Mahwah, NJ: Erlbaum Associates.

Huggins, V. J., & Winters, F. F. (1995). Income and poverty time series data from the Survey of Income and Program Participation. Paper presented at the Annual meeting of the American Statistical Association.

Lamb, M. (1997). The role of the father in child development. New York, NY: John



Wiley and Sons.

Long, S. K., & Clark, S. J. (1997). The new child care block grant: state funding choices and their implications. New federalism: issues and options for states, The Urban Institute, A-12.

Mahoney, J. L. (1997). From companions to convictions: Peer groups, school engagement, and the development of criminality. Paper presented at the The Society for Research on Child Development, Washington, DC.

Manlove, J. (1998). The influence of high school dropout and school disengagement on the risk of school-age pregnancy. Journal of Research on Adolescence, 8, 187-

220.

McGroder, S. (2000). Parenting among low-income, African-American single mothers with preschool-age children: patterns, predictors, and developmental correlates. Child Development, 71(3), 752-771.



McNeely, C. A., Nonnemaker, J. M., & Blum, R. W. (2002). Promoting school connectedness: evidence from the National Longitudinal Study of Adolescent Health. Journal Of School Health, 72(4), 138-146.

Moffitt, R. A., & Cherlin, A. (2000). Disadvantage among families remaining on welfare.

JCPR policy briefs, 3(12).

National Institute of Child Health and Human Development. (2002). The NICHD study of early child care. Bethesda, MD: National Institute of Child Health and Human Development, http://www.nichd.nih.gov/publications/pubs/early_child_care.htm.

National Research Council. (1995). Measuring poverty: a new approach. Washington, DC: National Academy Press.

Parrot, S. (1998). Welfare recipients who find jobs: what do we know about their earnings and employment? Washington, DC: Center on Budget and Policy Priorities.

Short, K., Garner, T., Johnson, D., & Doyle, P. (1999). Experimental poverty measures:

1990 to 1997. (Current Population Reports, P60-205). Washington, DC: U.S. Census Bureau.

U.S. Census Bureau. (2001a, 12/11/01). Detailed Poverty Statistics, Table 2, http://ferret.bls.census.gov/macro/032001/pov/new02_001.htm. Retrieved

7/30/02, from the World Wide Web:



http://ferret.bls.census.gov/macro/032001/pov/new02_001.htm

U.S. Census Bureau. (2001b). Detailed Poverty Statistics, Table 3, http://ferret.bls.census.gov/macro/032001/pov/new02_001.htm. Retrieved, from the World Wide Web:

U.S. Census Bureau. (2001c). Detailed Poverty Statistics, Table 11, http://ferret.bls.census.gov/macro/03/pov/new11_001.htm. Retrieved, from the World Wide Web:

U.S. Department of Health and Human Services. (1998). Temporary Assistance for Needy Families (TANF), 1936-1998. Washington, DC: Administration for Children and Families.

U.S. Department of Health and Human Services. (2001a). FFY 2000 CCDF state expenditures. Washington, DC: Child Care Bureau.

U.S. Department of Health and Human Services. (2001b). The Surgeon General's call to action to prevent and decrease overweight and obesity. Washington, DC: U.S. Government Printing Office, http://www.surgeongeneral.gov/library.

U.S. Department of Health and Human Services. (2002a). Characteristics and financial circumstances of TANF recipients, October 1999 - September 2000. Washington, DC: Administration of Children and Families.

U.S. Department of Health and Human Services. (2002b). Child care expenditures during

FFY 2001. Washington, DC: Child Care Bureau.

U.S. Department of Health and Human Services. (2002c). U.S. welfare caseloads information (Website): Administration on Children and Families at http://www.acf.dhhs.gov/news/stats/newstat2/shtml.

U.S. Department of Labor. (2002). Labor force statistics from the Current Population Survey. Washington, DC: Bureau of Labor Statistics at http://146.142.4.24/cgi- bin/surveymost?bls.

U.S. House of Representatives Committee on Ways and Means. (2000). 2000 Green

Book. Washington, DC.

U.S. House of Representatives, C. o. W. a. M. (2000). 2000 Green Book. Washington, DC.



Wertheimer, R. F. (1999). Children in working poor families. (Final report to the

Foundation for Child Development). Washington, D.C.: Child Trends. Wheaton, L. (1998). Low-income families and the marriage tax. Washington, DC: The

Urban Institute.

Wu, L. L. (1996). Effects of family instability, income, and income Instability on the

Risk of a Premarital Birth. American Sociological Review, 61, 386-406.

Zaslow, M., Brooks, J. L., Moore, K. A., Morris, P. A., Tout, K., & Redd, Z. A. (2001).

Impacts on children in experimental studies of welfare to work programs. Washington, DC: Child Trends.

Zaslow, M., Halle, T., Zaff, J., Calkins, J., & Margie, N. (2000). Background for community-level work on school readiness: A Review of definitions, assessments, and investment strategies. Washington, DC: Child Trends.

Zaslow, M., Moore, K. A., Brooks, J. L., Morris, P. A., Tout, K., Redd, Z. A., & Emig, C.



A. (2002). Experimental studies of welfare reform and children. The future of children, 12(1), 79-95.

Zaslow, M., Moore, K. A., Tout, K., Scarpa, J., & Vandivere, S. (2002). How are children faring under welfare reform? In K. Finegod (Ed.), Welfare reform: the next act. Washington, DC: The Urban Institute Press.

Download 0.69 Mb.

Share with your friends:




The database is protected by copyright ©essaydocs.org 2022
send message

    Main page