Strengths of the Safety Net:
How the EITC, Social Security, and Other Government Programs Affect Poverty

I.  Introduction
II.  Why Did Poverty Not Decline More in 1996?
III.  The Earned Income Tax Credit and Poverty Among Children
IV.  Social Security and Poverty Among the Elderly
V.  Impact of the Safety Net by Race, Ethnicity, and Age
VI.  Impact of the Safety Net Across Regions
Appendix:  Description of Data, Methodology and Definitions Used in the Analysis
End Notes

Executive Summary

The safety net of government benefit programs lifts millions of people, mostly elderly people and children, out of poverty. Using data collected by the Census Bureau, this report describes the impact of these programs — including Social Security, cash assistance, food and housing benefits, and the Earned Income Tax Credit — on poverty.

This analysis compares the number of people who would be poor if government benefits were not counted as part of their income to the number who are poor after counting government benefits. The analysis uses two measures of poverty that differ from the official poverty definition. One measure is poverty before government benefits are counted, which is determined using the official poverty line but excluding government benefits from people's incomes when those incomes are compared to the poverty line. The other measure is poverty after government benefits are counted, with government cash and non-cash benefits included as income and federal income and payroll taxes subtracted from income. The difference between the poverty counts under these two measures is the number of people lifted out of poverty by government safety net programs.

The purpose of this analysis is to identify the impact of the safety net on the extent and depth of poverty and to examine the relationship between changes in safety net programs and the number of people moved out of poverty.

In 1996, some 57.5 million people — or 21.6 percent of Americans — would have been poor if government benefits were not counted as part of their income. When government benefits are counted as income, these figures drop by nearly half, to 30.5 million people, or 11.5 percent of Americans. As these figures indicate, government benefit programs lifted 27 million people out of poverty in 1996.

Safety Net Weakens Slightly in 1996

Between 1995 and 1996, the poverty rate before counting government benefits fell. After counting government benefits, however, there was no real change in the poverty rate. One reason poverty after counting government benefits failed to decline was that the effectiveness of the safety net in reducing poverty decreased slightly in 1996, especially among children.

• In 1995, some 33.1 percent of all children who would have been poor without government benefits were lifted from poverty by those benefits.

• In 1996, this percentage slipped to 31.9 percent.

Most of the reduction between 1995 and 1996 in the number of children lifted out of poverty by safety net programs was due to a decline in the impact of the food and housing programs, primarily food assistance. The number of children moved out of poverty by these programs dropped from 2.2 million in 1995 to 1.9 million in 1996. The impact of cash assistance programs, primarily AFDC, on the depth of poverty among children also declined.

These changes correspond to reductions in the number of poor children receiving AFDC and food stamps between 1995 and 1996.

• Between 1995 and 1996, the number of children receiving AFDC benefits declined from 9 million to 8.4 million, a drop of 600,000 children. While an improving economy contributed to this decline, it can explain only part of it. The drop in the number of children receiving AFDC assistance exceeded the decline in the number of children who were poor. In 1995, some 52.7 children received AFDC benefits for every 100 children who were below the poverty line before government benefits were counted. In 1996, some 50.2 children were receiving AFDC benefits for every 100 children in poverty before government benefits were counted.

• The number of children receiving food stamps also dropped between 1995 and 1996, with about 700,000 fewer children receiving food stamps in 1996. Here, too, the improvement in the economy can explain only part of this change. In 1995, some 81.1 children received food stamps for every 100 children who were poor before government benefits were counted. In 1996, some 79.3 children received food stamps for every 100 children in poverty before government benefits were counted.

The welfare law enacted in 1996 was not a major reason for the drop in the number of children receiving AFDC or food stamps that year, because few of the legislation's impacts were felt in 1996. Many states, however, already had efforts underway to reduce the number of families receiving AFDC benefits. Some of the families who became ineligible for AFDC also appear to have been dropped from, or to have left, the food stamp program, although they may have continued to be eligible for food stamp benefits. The impacts of the federal welfare law will be felt in future years, as families begin to become ineligible for assistance due to the time limits and as other welfare policy changes take effect.

The ratio of the number of children who receive cash assistance and food stamps to the number of children who are poor before receipt of government benefits almost certainly will continue to decline. Preliminary data for 1997 show sharper declines in the number of children receiving cash assistance and food stamps between 1996 and 1997 than occurred between 1995 and 1996.

• Between mid-year (May to August) 1996 and mid-year 1997, the average monthly number of recipients of means-tested cash assistance provided to families with children dropped by two million, from 12.2 million recipients to 10.2 million. This is a drop of 16 percent, larger than any single-year reduction in recent years.

• Between 1996 and 1997, average monthly participation in the food stamp program fell by more than three million people — an unprecedented drop of 12.6 percent. Of that number, between one million and 1.5 million are likely to have been children.

Such large declines in the numbers of children receiving cash assistance and food stamps, far in excess of any likely decline in the number of children who are poor before government benefits are counted, would indicate a significant weakening of the safety net in 1997.

The one program that was more effective in 1996 than in 1995 in reducing poverty among children was the Earned Income Tax Credit. The increase in the impact of the EITC offset some of the decline in the effectiveness of the cash and non-cash means-tested programs in reducing poverty.

EITC Now Lifts More Children out of Poverty than Any Other Program

In recent years, the impact of the Earned Income Tax Credit on poverty among children has increased substantially. The EITC now lifts more than two million children out of poverty. Since its creation in 1975, the EITC has been expanded several times and has become more effective in moving children out of poverty.

• In 1989, federal income and payroll taxes pushed more families with children into poverty than the EITC lifted out of poverty. The net effect of federal income and payroll taxes and the EITC that year was to push 400,000 children from low-income working families into poverty.

• By 1993, the impact of the EITC was large enough to offset fully the negative impact of income and payroll taxes on poverty among families with children and to move a small number of children out of poverty.

• In 1994, the final phase of the EITC expansion enacted in 1990 took effect and a further expansion approved in 1993 began to phase in. With a larger EITC, the net effect of federal income and payroll taxes and the EITC was to lift 600,000 children out of poverty.

• The number of children moved out of poverty due to the combined effect of federal income and payroll taxes and the EITC increased to 1 million in 1995 and 1.2 million in 1996.

Separating out the impact of the EITC from the rest of the federal tax system shows that the EITC now lifts more children out of poverty than any other government benefit program.

• The EITC lifted 2.4 million children out of poverty in 1996. This constituted 37.3 percent of all children moved out of poverty by government programs that year.

• No other category of government benefit programs lifted as many children out of poverty as the EITC.

• Although the EITC lifts more children out of poverty, the means-tested cash and non-cash programs have a larger impact than the EITC in reducing the depth of poverty among children.

Because the EITC is designed to benefit low-income working families, it is no surprise that the EITC is most effective in reducing poverty among children from families with a parent who works at least half-time.

• Among children in families with at least one half-time worker, the EITC moves 30.2 percent of those who would otherwise be poor out of poverty.

The EITC is especially effective for Hispanic children.

• The EITC lifted 800,000 Hispanic children out of poverty in 1996. This was nearly one-fifth — 18 percent — of Hispanic children who were poor before counting government benefits.

• Nearly half of the Hispanic children who were lifted out of poverty by government benefit programs in 1996 were lifted out by the EITC.

• The EITC has such a large effect in reducing poverty among Hispanic children because the proportion of poor children living in families with a full-time worker is larger for Hispanics than for either whites or blacks.

The effectiveness of the EITC in lifting children out of poverty varies by region, with its greatest impact in the South.

• In 1996, the EITC lifted from poverty 17.6 percent of children in the South who were poor before government benefits were counted — more than one in every six such children.

• The South has a larger proportion of families that have earnings but are still poor than any other region.

Social Security Dramatically Reduces Poverty Among the Elderly

Of all age groups in the population, it is the elderly who depend on government programs to the greatest degree. Half of the population aged 65 and older — 50.1 percent in 1996 — would be poor if not for the benefits of government programs. After counting government benefits, the poverty rate among elderly people dropped to only 9.2 percent in 1996.

The primary reason government benefit programs have such a large impact on poverty among the elderly is the powerful effect of Social Security.

• Social Security lifted 11.7 million elderly people out of poverty in 1996.

• Altogether, government benefit programs lifted 13 million elderly people out of poverty in 1996. Nine out of every ten elderly people lifted out of poverty by government benefit programs were lifted out by Social Security.

Poverty rates have dropped consistently over the past two decades among people 65 and older, largely because of the increased effect of Social Security.

• In each year since 1979 for which data are available, about half of the elderly population was poor before counting government benefits.

• After counting social insurance benefits (mainly Social Security), but not other government benefits, the poverty rate among elderly people was 17.4 percent in 1979. By 1996, the elderly poverty rate after social insurance benefits are counted had dropped to 12.3 percent. This means that the proportion of elderly people lifted out of poverty by Social Security has increased in recent decades.

• In 1979, two-thirds of the elderly people who would otherwise be poor — 68 percent — were lifted from poverty by social insurance programs, mainly Social Security. In 1996, three-quarters of those who would otherwise be poor — 75.6 percent — were moved out of poverty by these programs.

Social Security is also important to low-income elderly people because it contributes a large portion of their income. In 1995, Social Security made up two-thirds of the total incomes of elderly people who were poor under the official measure of poverty.

Effect of Safety Net Varies by Race and by Region

In 1996, the safety net significantly reduced poverty for whites, blacks and Hispanics of all age categories. The number of people who remained poor was substantially lower for each of these groups than it would have been in the absence of government programs. (In this report, the terms "white" and "black" mean non-Hispanic white and non-Hispanic black.)

• In 1996, safety net programs lifted out of poverty more than half — 56 percent — of whites who would otherwise have been poor, along with 35.9 percent of blacks and 33.5 percent of Hispanics.

These racial differences stem in large part from the fact that government programs lift from poverty a much greater share of the elderly than of the non-elderly and a substantially higher proportion of the white poor than of the black and Hispanic poor are elderly. Some 40.7 percent of whites who are poor before receipt of government benefits were elderly in 1996, compared with only 12.7 percent of the black poor and 8.7 percent of the Hispanic poor.

The impact of the safety net in moving Hispanics out of poverty is affected also by the fact that a significant portion of the Hispanic poor are immigrants who do not have citizenship status. The safety net lifts from poverty a much smaller proportion of non-citizens than of citizens. The proportion of Hispanic non-citizens lifted from poverty by government benefit programs is expected to decline significantly after 1996, as new restrictions on the eligibility of legal immigrants for major government benefit programs take effect.

Data on the four Census Bureau regions — the Northeast, Midwest, South, and West — show that the safety net reduces the number of people who otherwise would have been poor in all regions, with the magnitude of the impact varying by region. One reason for the regional variation is the difference in the effectiveness of cash assistance programs across regions.

The percentage of those poor in the absence of government benefits who were lifted from poverty by cash assistance benefits ranged from a low of 4.1 percent in the South to a high of 6.8 percent in the West.

This variation is largely due to substantial differences in benefit levels for government programs in those regions. In the South, AFDC maximum benefits for a family of three with no other income averaged $251.57 in 1996, the lowest level of all regions. The two regions with the highest average maximum AFDC benefit levels — the Northeast at $553.58 and the West at $566.64 — also had the largest percentages of people lifted out of poverty by cash assistance programs.

I. Introduction

This report examines the effect of government safety net programs on poverty. The effect of these programs on poverty is determined by comparing the number of people who would be poor if government benefits were not counted as part of their incomes to the number who are poor when government benefits are counted. The difference is the effect of government programs in lifting people out of poverty.

The analysis in this report uses data collected by the Census Bureau each year on the incomes of the American people — the same data on which the official poverty measure is based. These Census data include information on cash income from earnings, as well as from Social Security payments, Unemployment Compensation, Supplemental Security Income, welfare payments, and many other government cash benefits. These Census data also include certain government benefits not in the form of cash, such as food stamps, school lunches, and housing assistance. Federal income and payroll taxes paid and any Earned Income Tax Credit received are also included in these data.

The official poverty measure is based on a definition of income that includes all cash payments received by an individual or family, whether from earnings, government benefits, or any other source. Benefits not in the form of cash are not counted as income in the official measure. To determine whether an individual or family is poor, that person's or family's cash income is compared to the official poverty line.

This analysis uses two measures of poverty that differ from the official poverty definition. One is poverty before government benefits are counted, or "pre-transfer" poverty. This measure is determined using the official poverty line but excluding from income all government benefits. The number of people counted as poor under the pre-transfer poverty measure is higher than under the official poverty count because, unlike the official poverty definition, the pre-transfer measure excludes from income all government benefits provided in cash.

The pre-transfer poverty measure is also useful in determining the impact of the private-sector economy on poverty. If government benefits are excluded, the remaining income is that produced by the private sector, primarily as earnings from employment. Tracking the pre-transfer measure of poverty over time shows trends in the impact of the economy on poverty. Pre-transfer poverty rises during economic recessions and falls during economic recoveries.

The other measure used in this report is poverty after government benefits are counted, or "post-transfer" poverty. It, too, is based on the official poverty line, but with government benefits, both cash and non-cash, included as income.⁽¹⁾ The number of post-transfer poor is lower than the official poverty count because this measure includes non-cash government benefits, which are not counted in the official poverty measure.

The difference between the pre-transfer poverty rate and the post-transfer poverty rate is the impact of government benefits. In 1996, some 57.5 million people, or 21.6 percent of the U.S. population, would have been poor if government benefits were not counted as part of their incomes. After counting government benefits, the number of poor dropped to 30.5 million, or 11.5 percent of the population. Some 27 million people were lifted out of poverty by government benefit programs in 1996. Thus, almost half of those who were poor before receipt of government benefits — 46.9 percent — were lifted from poverty by those benefits.

In addition to examining the impact of government benefits on the number of people in poverty, it is also useful to analyze the impact of government benefits on the depth of poverty. While some people are lifted out of poverty by government programs, others have incomes so low that they remain in poverty even when government benefits are counted as part of their incomes. These individuals are less poor, however, than they would have been without these benefits.

A measure that researchers call the "poverty gap" is used to determine the depth of poverty. The poverty gap is the total amount by which the incomes of all poor people fall below the poverty level. In other words, the poverty gap represents the amount of money that would be needed to lift every poor person exactly to the poverty line. In 1996, the poverty gap before counting government benefits was $200 billion; after counting government benefits, the poverty gap was reduced to $61 billion. This means that government benefit programs reduced the depth of poverty by two-thirds, or $139 billion, in 1996.

Measuring what the extent of poverty would be without the benefits of government programs is not the same as saying that if these programs did not exist, this many more people would be poor. Without these programs, other institutional and behavioral changes would occur that would affect the extent of poverty. The purpose of this analysis is to identify the impact of the safety net on the extent and depth of poverty and to examine the relationship between changes in safety net programs and the number of people moved out of poverty.

Organization of Report

Chapter two of this report discusses the lack of progress in reducing post-transfer poverty between 1995 and 1996. The post-transfer poverty rate failed to improve in spite of a robust economy. The chapter examines changes in the safety net that contributed to this result.

In chapter three, the relationship between the Earned Income Tax Credit and poverty among children is explored. The analysis in this chapter indicates that the EITC has a substantial impact on poverty among children.

Chapter four presents a similar analysis of the relationship between Social Security and poverty among the elderly. This chapter shows that the elderly are more dependent on government programs than any other age group and that Social Security is the most effective government program in lifting elderly people out of poverty.

Chapter five examines differences in the impact of government safety net programs on poverty among whites, blacks and Hispanics. This chapter also includes data on differences in the effectiveness of the safety net for Hispanic citizens and non-citizens.

Chapter six compares the impact of the safety net programs across geographic regions and examines the relationship between variations in state benefit levels and regional differences in the impacts of the safety net on poverty.

The data used in this report are displayed in tables included in an appendix.

II. Why Did Poverty Not Decline More in 1996?

In 1996, the official poverty rate was 13.7 percent, statistically unchanged from the previous year. The official poverty rate for children was 20.5 percent, also statistically the same as 1995.

The poverty rate stalled in 1996 in spite of an improving U.S. economy. Between 1995 and 1996, the gross domestic product — the basic measure of the size of the U.S. economy — grew 2.8 percent, after adjusting for inflation. The average unemployment rate fell from 5.6 percent in 1995 to 5.4 percent in 1996. In the two previous years of 1994 and 1995, the official poverty rate had dropped significantly as the economy improved and the unemployment rate fell.

When the Census Bureau released the data showing that the official poverty rate failed to decline in 1996, some commentators assumed the lack of progress against poverty must be primarily due to eroding wages. An analysis of changes in poverty rates before and after government benefits are counted sheds some light on this issue and suggests other factors were also at work.

The data show that the improving economy did in fact reduce poverty a small amount, at least for children. This can be seen by examining the child poverty rate as measured before government benefits are counted; this is the measure of poverty that largely reflects the impact of private-market income. Between 1995 and 1996, this measure — also called the pre-transfer child poverty rate — dropped from 24.2 percent to 23.6 percent. (See Table 1) This continued a decline in the pre-transfer child poverty rate from its peak of 26.3 percent in 1993.

After counting government benefits, however, the decline in the poverty rate among children between 1995 and 1996 practically disappears. The post-transfer child poverty rate remained at nearly the same level in both years rather than declining as it did in the previous two years.

These data indicate that one of the principal reasons the poverty rate among children failed to improve between 1995 and 1996 was that the safety net programs had less effect in 1996 in lifting children out of poverty. In 1995, government benefit programs lifted 5.7 million children from poverty, reducing the number of children who otherwise would have been poor by 33.1 percent. (See Table 2.) In 1996, government benefit programs lifted 5.3 million children from poverty — 400,000 fewer than in 1995 — reducing the number of poor children by 31.9 percent.

An analysis of child poverty rates before and after counting government benefits thus shows that in 1996, economic growth had some effect in modestly reducing poverty among children, but a weakening of the impact of the safety net programs in reducing poverty offset that gain. The net effect was no real change in child poverty between 1995 and 1996 after counting government benefits.

The lack of a change in the post-transfer child poverty rate is especially notable because an expansion in the Earned Income Tax Credit reduced the post-transfer child poverty rate by 0.2 percentage points. This indicates that the effectiveness of the other safety net programs in reducing child poverty declined by 0.7 percentage points between 1995 and 1996, a large decline for a single year.

This stands in contrast to a trend of increasing effectiveness of the safety net programs in reducing child poverty in the years before 1996. The number of children lifted from poverty in 1995 — 5.7 million — was substantially greater than the 4.3 million moved out of poverty in 1993 or the 3.1 million moved out of poverty in 1989. Because of this substantial progress, the number and percentage of children lifted from poverty by government benefit programs in 1996 still exceeded the number and percentage lifted from poverty in the late 1980s and early 1990s.

Impacts of Different Types of Safety Net Programs

The safety net programs can be divided into four groups: social insurance programs, mainly Social Security but also including federal pensions and unemployment insurance; means-tested cash assistance programs such as Aid to Families with Dependent Children and Supplemental Security Income; means-tested non-cash programs like food stamps, school lunches, and housing assistance; and federal income and payroll taxes, including the Earned Income Tax Credit.⁽²⁾

In 1996, as in previous years, social insurance programs had the greatest impact of any of the safety net programs in reducing poverty. Of the 27 million people moved out of poverty by safety net programs in 1996, social insurance programs were responsible for 17.9 million, or two-thirds. (See Table 3.) This is primarily a reflection of the powerful effect of Social Security on poverty among the elderly and disabled. Of the 17.9 million people moved out of poverty by social insurance programs, two-thirds — 12.1 million — were elderly, while most of the rest were non-elderly adults receiving Social Security disability or survivors benefits. Relatively few of the people lifted out of poverty by social insurance programs were children.

The means-tested programs that provide non-cash food and housing benefits reduced the number of people in poverty more than means-tested programs that provide benefits in cash. In 1996, some 4.3 million people were lifted from poverty by non-cash programs. By comparison, means-tested cash assistance programs lifted 3 million people out of poverty.

On the other hand, the means-tested cash programs had a greater impact than the non-cash programs on the depth of poverty. As explained in the first chapter, the measure used to determine the depth of poverty is the poverty gap, or the amount of money needed to move every poor person to the poverty line. In 1996, the cash programs eliminated 12.8 percent of the total poverty gap, or $25.6 billion of that gap. The non-cash programs reduced the poverty gap by 9.2 percent, eliminating $18.3 billion of the gap.

The non-cash means-tested programs had nearly twice the impact of the cash means-tested programs in reducing the number of poor children. The non-cash programs lifted 1.9 million children out of poverty in 1996. The cash programs moved about 1 million children out of poverty. Both cash and non-cash means-tested programs reduced the poverty gap among children by about the same amount.

The different effects of cash and non-cash means-tested benefit programs on poverty should not be surprising. Cash assistance benefits for poor families with children are very low; they leave families well below the poverty level in most states. Cash assistance benefits also go primarily to families with no other income. In these circumstances, cash assistance is effective in reducing the depth of poverty among these families but moves few of them out of poverty altogether. Food stamps, school lunches, and housing assistance, by contrast, go to working poor families as well as to families receiving cash welfare benefits. Added to earnings or other benefits, these non-cash benefits are enough to bring some families' incomes above the poverty line.⁽³⁾

Federal taxes and the EITC brought 1.7 million people, including 1.2 million children, out of poverty in 1996. What is noteworthy here is not the size of the impact so much as the fact that the combination of federal taxes and the EITC has a positive impact at all. Federal income and payroll taxes generally reduce income and thereby increase poverty. But the impact of the EITC among certain groups, particularly families with children, is sufficiently positive to outweigh the negative impact of other federal taxes and to cause the net effect of the tax system to be a reduction in the number of children in poverty.

Changes Between 1995 and 1996 in the Impacts of Different Types of Programs

Between 1995 and 1996, the impact of the safety net on poverty decreased. The reduction in the impact of government benefits on poverty was sharpest among poor children.

Changes in the effectiveness of the means-tested non-cash programs accounted for the largest portion of this drop. In 1995, these programs lifted 2.2 million children out of poverty. In 1996, they moved 1.9 million children out of poverty. (See Table 4.)

The means-tested cash programs lifted slightly fewer children out of poverty in 1996 than in 1995. In 1995, these programs moved 1.1 million children above the poverty level, compared with 1 million in 1996. The social insurance programs also accounted for some of the decline in the effectiveness of the safety net for children between 1995 and 1996.

The only category that moved more children out of poverty in 1996 than in 1995 was federal taxes and the EITC. Between 1995 and 1996, the number of children moved out of poverty by the tax system, including the EITC, rose from 1 million to 1.2 million.

Although the change between 1995 and 1996 in the number of children lifted out of poverty by means-tested cash assistance programs was small, there was a larger change in the impact of these programs on the depth of child poverty. In 1995, the cash assistance programs reduced the poverty gap among children by $10.5 billion; in 1996 they reduced the child poverty gap by $9.1 billion, or 13 percent less.

Effectiveness of Safety Net Programs Has Been Increasing

In spite of a generally improving economy, pre-transfer poverty rates have risen over the past decade. The 21.5 percent poverty rate in 1996 before government benefits are counted was nearly a full percentage point higher than the corresponding poverty rate of 20.6 percent for 1987.⁽⁴⁾

The 1996 rate was higher despite the fact that both years came at a similar point in the economic cycle, five years after the trough of the most recent recessions. Moreover, economic conditions were more favorable in 1996 than in 1987, with lower unemployment. The unemployment rate was 5.4 percent in 1996, compared to 6.2 percent in 1987. The proportion of the non-elderly adult population that was employed was also larger in 1996 than in 1987. (See Table 5.)

This disconcerting increase in the poverty rate before government benefits are counted between 1987 and 1996 despite the economic improvement is probably attributable to an unequal distribution of income gains from the economic growth. During this period, a disproportionate share of the economic gains accrued to those on the upper rungs of the income ladder.⁽⁵⁾

Yet, while the pre-transfer poverty rate was higher in 1996 than in 1987, the post-transfer poverty rate was lower. The post-transfer poverty rate declined from 12.6 percent in 1987 to 11.5 percent in 1996.

The reason for this lower post-transfer poverty rate is the considerable strengthening of the safety net between 1987 and 1996. Even though the safety net programs weakened somewhat in 1996, these programs still lifted more people out of poverty than they had in 1987. In 1987, government programs lifted 19.3 million people out of poverty. By 1996 that number had risen to 26.7 million. The proportion of those poor before counting government benefits who were poor moved out of poverty by safety net programs rose from 38.8 percent in 1987 to 46.6 percent in 1996. This shows that when the safety net is strengthened, poverty is reduced.

The strengthening of certain means-tested programs during this period played an important role in this bolstering of the safety net. The number of people lifted out of poverty by the means-tested cash programs increased from 1.8 million in 1987 to 2.9 million in 1996. (See Table 6.) The proportion of the pre-transfer poor who were moved out of poverty by means-tested cash programs rose from 3.7 percent in 1987 to 5.1 percent in 1996. Similarly, the number of people moved out of poverty by the non-cash programs increased from 3.5 million in 1987 to 4.3 million in 1996. This was an increase from 7.1 percent to 7.5 percent of the pre-transfer poor.

Most significant was the increased effect of the EITC on poverty. In 1987, the combination of taxes and the EITC pushed 1.4 million people into poverty. By 1996, taxes and the EITC were lifting 1.7 million people out of poverty.⁽⁶⁾

A number of changes in the safety net programs during these years contributed to their greater effectiveness in lifting out of poverty those who would otherwise be poor. In the early 1990s, more low-income disabled children became eligible for SSI. Food stamp benefits were improved and participation in the AFDC and food stamp programs increased. The number of families receiving housing assistance rose. And the EITC was substantially expanded by legislation enacted in 1990 and again by legislation approved in 1993.

Although the number and percentage of pre-transfer poor individuals lifted out of poverty by the safety net programs were higher in 1996 than in 1987, the impact of the programs was not at a historically high level in 1995 or 1996. In 1979, the earliest year for which we have these data, government benefit programs lifted 48.1 percent of the pre-transfer poor out of poverty. This exceeds both the 47.2 percent level for 1995 and the 46.6 percent level for 1996. The impact of government benefit programs in reducing poverty fell sharply in the early 1980s and, while it has increased in recent years, it remains below the 1979 level.

Lack of Progress Against Poverty in 1996 Linked to Reductions in Safety Net

After a period of steadily increasing effectiveness, the impact of government safety net programs in reducing poverty lessened somewhat between 1995 and 1996. The principal reductions were in the impacts of cash and non-cash means-tested programs on poverty among children. The impact of the EITC in reducing child poverty increased, however, and offset some of the decline in the impact of the other programs.

The principal means-tested cash assistance program for families with children in 1996 was AFDC. (The AFDC program has now been replaced by a block grant to the states through the Temporary Assistance for Needy Families program.) The largest non-cash means-tested program affecting families with children is the food stamp program. Participation by children dropped in both programs in 1996.

Between 1995 and 1996, the number of children receiving AFDC benefits in an average month declined from 9 million to 8.4 million, a drop of about 600,000 children. (See Table 7.) During the same period, the total number of children poor before government benefits were counted fell by 500,000. The decrease in the number of children receiving AFDC thus exceeded the drop in child poverty.

This means that a smaller proportion of pre-transfer poor children received AFDC in 1996 than in 1995. In 1995, some 52.7 children received AFDC benefits for every 100 children who were poor before government benefits were counted. In 1996, only 50.2 children received AFDC benefits for every 100 children in poverty before counting government benefits.

Between 1995 and 1996, the number of children receiving food stamps in an average month dropped by an estimated 700,000, a drop that exceeded the reduction in the number of children in need. In 1995, some 81.1 children received food stamps for every 100 children who were poor before receiving government benefits. In 1996, for every 100 such children, 79.3 children received food stamps.

This followed a period in which the number of children receiving AFDC and food stamps had increased substantially. Between 1989 and 1994, the number of children receiving AFDC increased by 2.1 million. This increase was partly the result of expanding need; the number of children who were poor before receipt of government benefits was 2.8 million greater in 1994 than in 1989. But this increase also resulted from growth in the proportion of eligible children who participated in the programs. In 1989, some 48.7 children received AFDC for every 100 children who were poor before receipt of government benefits. By 1994, some 53 children were receiving AFDC for every 100 children who were poor before counting government benefits.

This pattern was even more marked in the food stamp program. Between 1989 and 1994, the number of children receiving food stamps increased by 5 million. In 1989, some 63.1 children received food stamps for every 100 pre-transfer poor children. By 1995, the number of children receiving food stamps for every 100 children in pre-transfer poverty had grown to 80.7. This remarkable increase in the proportion of pre-transfer poor children receiving food stamps indicates that a large part of the impact of the food stamp program on post-transfer child poverty rates was due to an increase in the participation rate among eligible children.

The last row in Table 7 shows the number of poor children lifted out of poverty by the means-tested programs. In years when participation by children in AFDC and food stamps increased, the number of children moved out of poverty by means-tested programs rose. Between 1989 and 1994, the number of children lifted out of poverty by the means-tested programs grew from 2.4 million to 3.1 million. Between 1995 and 1996, by contrast, the number of children participating in AFDC and food stamps fell, and fewer children were lifted out of poverty by these programs.

The ratio of the number of children who receive cash assistance and food stamps to the number of children poor before receipt of government benefits almost certainly will continue to decline. Preliminary data for 1997 show sharper declines in the number of children receiving cash assistance and food stamps between 1996 and 1997 than occurred between 1995 and 1996. Between mid-year (May to August) 1996 and mid-year 1997, the average monthly number of recipients of means-tested cash assistance provided to families with children dropped from 12.2 million to 10.2 million. This is a drop of 16 percent, larger than any single-year reduction in recent years.

To be sure, the number of children who are poor before receipt of government benefits is likely to have declined again in 1997 due to continued economic growth. But unless the pre-transfer poverty rate for children fell by approximately three percentage points, the ratio of the number of children receiving cash assistance to the number of children who were poor before counting government benefits is likely to have dropped further in 1997. A drop in the pre-transfer poverty rate for children of three percentage points in a single year would be unprecedented. A decline that large would exceed the entire decline in the pre-transfer poverty rate for children for the three years between its peak in 1993 and 1996.

A decline can be expected also in the ratio of the number of children receiving food stamps to the number of children in poverty. Average monthly food stamp participation for 1997 was more than three million lower than average monthly participation for 1996. This was an unprecedented drop of 12.6 percent. Some of this reduction in participation was due to the phasing in of restrictions on the eligibility of 18- to 49-year-old adults without children and the elimination of many non-citizens from the program, fewer than one-fifth of whom were children. Nonetheless, of the decline in food stamp participation, about one million to 1.5 million was likely to have been a reduction in the participation of children.

Such large declines in the numbers of children receiving cash assistance and food stamps, far in excess of any likely drop in the numbers of pre-transfer poor children, would indicate a serious weakening of the safety net in 1997.

The federal welfare law enacted in 1996 was not the major reason for the drop in children participating in cash assistance programs in 1996 because few of the impacts of the law were felt in 1996. Many states, however, already had efforts underway to reduce the number of families receiving AFDC benefits. The impacts of the federal law will be felt in future years, as families are affected the time limits on cash assistance and other changes in welfare policy. Future declines in the ratio of children receiving cash assistance to pre-transfer poor children are likely for this reason.

Similarly, although the welfare law included major reductions in food stamps, most of these reductions did not take effect until 1997. The drop in 1996 in the ratio of the number of children receiving food stamps to the number of pre-transfer poor children was probably a consequence of state efforts to reduce the number of families receiving cash assistance. Many families that became ineligible for cash assistance likely also were dropped from, or otherwise left, the food stamp program, although a substantial number may have continued to be eligible for food stamp benefits. Because this food stamp participation decline is apparently linked to reductions in cash assistance receipt, the ratio of the number of children receiving food stamps to the number of pre-transfer poor children is likely to fall further as the number of children receiving cash assistance continues to decline. In addition, with the principal food stamp reductions in the welfare law now having taken effect, food stamp participation among children — particularly legal immigrant children, most of whom became ineligible for food stamps in 1997 — is likely to have dropped further.

The only program that was more effective in 1996 than in 1995 in reducing poverty among children was the EITC. The increase in the impact of the EITC offset some of the decline in the effectiveness of the cash and non-cash means-tested programs in reducing poverty.

The EITC had a larger effect on poverty in 1996 than in 1995 because EITC benefits for families with two or more children increased between these two years as a result of the EITC expansions enacted as part of the 1993 budget law. These expansions phased in over the three years from 1994 through 1996. With the EITC expansions having been completed in 1996, there are no further EITC expansions to offset the further reductions in the cash and non-cash means-tested programs for poor families with children that have occurred since 1996 or are expected in coming years as families begin to reach time limits in the cash assistance programs.

III. The Earned Income Tax Credit and Poverty Among Children

In recent years, the EITC has had a growing impact on poverty, especially among children. The EITC now moves more than two million children out of poverty, a larger number of children than are lifted out by any other safety net program. The EITC is the most effective safety net program for children in working poor families.

The EITC lifts so many children out of poverty because it is a refundable tax credit; if the EITC owed to a low-income working family is larger than the income tax the family owes, the family receives a check from the Internal Revenue Service for the difference. Working families with children that have incomes modestly below the poverty line receive the largest EITC credit. This targeting contributes to the EITC's effectiveness in lifting children out of poverty.

Recent Changes in EITC Boost Its Effect on Poverty

Since its enactment in 1975, the EITC has been expanded several times. The Tax Reform Act of 1986, an initiative of President Reagan, was the first major expansion; that legislation substantially increased the amount of the credit and indexed it for inflation. In the 1990 bipartisan budget agreement, the credit was increased further and a larger credit was instituted for families with two or more children. Amendments proposed by President Clinton and enacted in 1993 again expanded the EITC for families with children and added a small credit for childless workers. Each of these changes increased the impact of the credit in reducing poverty among children.

In 1989, the last year before the most recent recession, the EITC was much smaller than it is today. In that year, the effect of federal income and payroll taxes in pushing families below the poverty line outweighed the effect of the EITC in lifting families above the poverty line. The net effect was to push 1.3 million people, including 400,000 children, into poverty.

Increases in the EITC enacted in 1990 and 1993 provided a larger credit each year from 1991 through 1996 to qualifying families with children. By 1993, the impact of the EITC was large enough to offset fully the negative impact of federal income and payroll taxes on poverty among families with children and to move a small number of children out of poverty.

In 1994, the final stage of the EITC increases enacted in 1990 took effect while the increases enacted in 1993 began to phase in. The net effect of federal taxes and the EITC was to move 600,000 children out of poverty in 1994. The net number of children lifted from poverty increased to 1 million in 1995 and 1.2 million in 1996.

EITC Has Substantial Effect in Lifting Children Out of Poverty

For 1996, we can disentangle the impact of the EITC from the effects of federal income and payroll taxes.⁽⁷⁾ This analysis shows that in 1996, the EITC lifted 4.6 million people, including 2.4 million children, out of poverty. (See Table 8.)

The children moved out of poverty by the EITC constituted 14.5 percent of the children who would have been poor in the absence of government programs. This means the EITC lifts out of poverty one of every seven children who otherwise would have been poor.

Because the poor are predominantly white, the number of poor whites moved out of poverty by the EITC is larger than the number of blacks and Hispanics lifted out of poverty.⁽⁸⁾ But the proportion of the pre-transfer poor lifted from poverty is higher among blacks and Hispanics than among whites. The EITC lifts from poverty 14.8 percent of the Hispanic pre-transfer poor and 8.1 percent of the black pre-transfer poor, compared with 5.5 percent of the white pre-transfer poor. As these figures indicate, the EITC has a particularly large impact on poverty rates among Hispanics.

The EITC has a greater impact on poverty rates among children than among adults because the credit is targeted to families with children. In 1996, some 14.8 percent of pre-transfer poor white children and 11 percent of pre-transfer poor black children were lifted out of poverty by the EITC. The EITC has its greatest impact on poverty rates among Hispanic children, lifting 18 percent of pre-transfer poor Hispanic children out of poverty.⁽⁹⁾

EITC Now Lifts More Children Out of Poverty Than Any Other Safety Net Program

The EITC now has a larger impact in reducing poverty rates among children than any other government benefit program. The 2.4 million children lifted from poverty by the EITC constituted 37.3 percent of all children moved out of poverty by government programs in 1996. (See Figure 2.) The EITC lifted more children out of poverty than any other category of safety net programs and almost as many children as means-tested cash and non-cash programs combined. Means-tested cash and non-cash programs together lifted 2.9 million children out of poverty.⁽¹⁰⁾

The EITC is very effective in lifting children out of poverty in large part because it is targeted on families that have more income and therefore are closer to the poverty line than the families eligible for means-tested programs. Unlike the very poor families that receive AFDC benefits, many of the families eligible for the EITC do not need as much in assistance to lift them above the poverty line.

In addition, the EITC is designed so that benefits increase with earnings until the family qualifies for the maximum EITC benefit. For many of the workers receiving the maximum EITC benefit, or close to it, the EITC is sufficient to raise their incomes above the poverty line. Benefits under other programs tend to be largest for families with little or no income and to phase down substantially or phase out entirely well below the poverty line.

Although the EITC lifts more children out of poverty, the means-tested cash and non-cash programs have a larger impact than the EITC in reducing the depth of poverty among children. The measure used to determine the depth of poverty is the poverty gap, or the amount of income needed to bring every poor person to the poverty line. Means-tested cash programs and means-tested non-cash programs reduced the pre-transfer poverty gap for children by 21.7 percent and 21.3 percent respectively in 1996. (See Table 9.) The EITC reduced the pre-transfer poverty gap by 8.7 percent.

EITC Benefits Children from Working Poor Families

Because the EITC is designed to benefit low-income working families, it is no surprise that the EITC is most effective in reducing poverty among children from families with a parent who works at least half-time. Among children in families with at least one half-time worker, the EITC moves 30.2 percent of those who would otherwise be poor out of poverty. (See Table 10.) Families with at least one half-time worker are defined as those with earnings of $5,150 or more — the amount a half-time worker earning the minimum wage throughout the year would earn.

Of the safety net programs examined in this report, the EITC is the most effective in reducing poverty among children from families with a parent who works at least half-time. The EITC lifts from poverty twice as many children from these families as does any other safety net program. The EITC also closes more of the poverty gap for children from families with at least half-time earnings than any other safety net program.

EITC Has Larger Impact on Hispanic Children

The EITC is most effective in lifting Hispanic children out of poverty. The EITC moved 800,000 Hispanic children out of poverty in 1996. This was nearly one-fifth — 18 percent — of Hispanic children who were poor before counting government benefits. The EITC moves more pre-transfer poor Hispanic children out of poverty than do means-tested cash and non-cash programs combined. The means-tested programs moved a total of about 700,000 Hispanic children out of poverty in 1996. Nearly half of the Hispanic children who were lifted out of poverty by government benefit programs — 48.3 percent — were lifted out of poverty by the EITC.

The EITC lifted 900,000 white children out of poverty in 1996, or 34.7 percent of all white children moved out of poverty by government programs. It also lifted out of poverty 500,000 black children, or 31.2 percent of all black children moved out of poverty by government programs.

One reason for the EITC's larger impact on poverty rates among Hispanic children is that poor Hispanic children are more likely than other children to live in families where a family member works full time. Census data show that 38.9 percent of Hispanic children classified as poor under the official poverty measure lived in families where someone worked full time throughout the year in 1996. By comparison, 31.9 percent of poor white children and 17.7 percent of poor black children lived in families with a full-time worker.⁽¹¹⁾ This is significant because the largest EITC benefits go to families that have earnings equal to approximately what a full-time worker earning the minimum wage throughout the year would earn.

The anti-poverty effect of the EITC is especially important to Hispanic children, who have one of the highest pre-transfer poverty rates of all groups of children. In 1996, some 43.9 percent of Hispanic children — more than two of every five — were poor before counting benefits from government programs.

EITC Especially Important to Children in South

The effectiveness of the EITC in lifting children out of poverty also varies by region. The EITC moves a larger proportion of pre-transfer poor children out of poverty in the South than in any other region.

In 1996, the EITC lifted from poverty some 17.6 percent of children in the South who were poor before government benefits — more than one in every six such children. (See Table 11.) This was a full three percentage points higher than the proportion of children moved out of poverty in the next highest region, the West, and nearly seven percentage points higher than the proportion of children moved out of poverty in the Northeast.

One reason for this regional difference in the EITC's effectiveness is that wage rates tend to be lower in the South and the South has more people with wages who live modestly below the poverty level than any other region.⁽¹²⁾

In short, the larger effect of the EITC on poverty in the South is due to the same factors that make the EITC particularly effective for Hispanics. The EITC has the largest impact on poverty among populations in which large proportions of families have low-wage workers.

IV. Social Security and Poverty Among the Elderly

Of all groups within the population, it is the elderly who depend on government programs to the greatest degree. Half of the population aged 65 and older — 50.1 percent in 1996 — would be poor if not for the benefits of government programs. After counting government benefits, the poverty rate among elderly people dropped to only 9.2 percent in 1996.

Social Security Is Most Important Safety Net Program for Elderly

In 1996, Social Security lifted 11.7 million elderly people out of poverty. It moved out of poverty nearly three-quarters — 73 percent — of the elderly who would have been poor in the absence of government programs.⁽¹³⁾ (See Table 12.)

In addition to the elderly, 3.5 million non-elderly adults and 800,000 children were lifted out of poverty by Social Security in 1996. Some 14 percent of non-elderly adults and 4.8 percent of children who were poor before receipt of government benefits were lifted out of poverty by Social Security.

Social Security reaches a high proportion of the elderly poor of all races. The program lifts out of poverty more than three-quarters — 77.7 percent — of the white elderly who would otherwise be poor and about half — 51.5 percent — of the black elderly people who would otherwise be poor. Some 46.1 percent of the Hispanic elderly who would otherwise be poor are lifted out of poverty by Social Security. (See Table 13.)

The citizenship status of Hispanic elderly people affects the proportion that Social Security lifts from poverty. Among elderly Hispanic individuals who were poor before counting government benefits, more than twice as high a proportion of citizens as of non-citizens are lifted out of poverty by these benefits. Among pre-transfer poor Hispanic elderly people who are citizens, more than half — 55.1 percent — are moved out of poverty by Social Security. One-quarter of pre-transfer poor Hispanic elderly people who are not citizens — 24 percent — are lifted out of poverty by these benefits.

Social Security alone is responsible for most of the elderly people lifted out of poverty by the safety net programs. Nine of every ten elderly people lifted out of poverty by government programs are lifted out by Social Security. The 11.7 million elderly people moved out of poverty by Social Security in 1996 constituted 89.3 percent of all elderly people lifted out of poverty by government programs. (See Table 14.)

For the elderly of all racial and ethnic groups, Social Security is more effective in reducing poverty than all other safety net programs combined. Social Security is responsible for 91.6 percent of the white elderly people moved out of poverty by government benefit programs and 80.7 percent of the black elderly lifted out of poverty by the programs. It is also responsible for 66.9 percent of the Hispanic elderly moved out of poverty by government benefits. Among all racial, ethnic, and age groups, the Hispanic elderly are the group that are most affected by means-tested cash programs. Yet Social Security lifts nearly four times as many Hispanic elderly people out of poverty as do the means-tested cash programs.⁽¹⁴⁾

Social Security Reduces Depth of Poverty

Social Security also substantially reduces the depth of poverty among the elderly. In 1996, Social Security eliminated 83.2 percent of the poverty gap among the elderly before receipt of government transfers.

Social Security has a large effect among all racial and ethnic groups in reducing the depth of poverty for elderly people. It eliminated 85.6 percent of the poverty gap the white elderly faced in 1996, some 74 percent of the gap for the black elderly, and 76.7 percent of the gap for elderly Hispanic citizens. Social Security eliminated a smaller, but still substantial share — 41.7 percent — of the poverty gap faced by elderly Hispanic non-citizens.

Social Security Constitutes Large Portion of Elderly Income

Social Security constitutes a large share of the income of the elderly poor. ("Poor" in this discussion of the income sources of the elderly refers to those who are poor under the official measure of poverty.) In 1995, Social Security made up two-thirds — 67.3 percent — of the total income of the elderly poor. This is twice as large a share of their income as all other income sources combined. (See Table 15.)

Means-tested government benefit programs provided 19.8 percent of the income of the elderly poor in 1995, while earnings made up 5.7 percent of their income. Pensions, interest and dividends contributed very little to the income of the elderly poor.

One reason that Social Security makes up such a large share of the income of the elderly poor is that so many of them receive Social Security payments — more than eight of every 10 poor elderly individuals and couples received these payments in 1995. (See Table 16.) Social Security is especially important to those elderly people whose income is just above the poverty line; 96.2 percent of the near-poor elderly individuals and couples (as measured under the official definition of poverty) received Social Security in 1994.⁽¹⁵⁾ The overwhelming majority of elderly people who are modestly above the poverty line have been lifted out of poverty by Social Security.

Reductions in Elderly Poverty Rate Over Time Due to Social Security

Among people 65 and older, poverty rates have dropped consistently over the past two decades. A major reason for this decline has been social insurance benefits, nearly all of which come from Social Security.

Since 1979, when the Census Bureau first began collecting these data, social insurance benefits have become increasingly effective in reducing poverty among elderly people. (See Figure 3.) In each year since 1979 for which data are available, about half of the elderly population would have been poor if not for the benefits of government programs. After counting the benefits of social insurance programs but not other government benefits, the poverty rate among elderly people was 17.4 percent in 1979. By 1996, the elderly poverty rate after counting social insurance benefits dropped to 12.3 percent.

After counting means-tested cash and non-cash benefits, the elderly poverty rate was further reduced. The elderly poverty rate, after counting all government benefits, was 9.2 percent in 1996.

The number and percentage of elderly people lifted from poverty by social insurance programs — mainly Social Security — increased each year through 1994.

(See Table 17.) In 1979, some 8.9 million elderly people were lifted out of poverty by social insurance programs. That figure rose to 12.1 million by 1994 and remained at about that level in 1995 and 1996.

In 1979, social insurance programs lifted out of poverty two-thirds of the elderly people who would otherwise have been poor — 68 percent. By 1995, these programs were lifting out of poverty slightly more than three-quarters — 76.5 percent — of the elderly who would otherwise be poor. Social insurance programs lifted a slightly smaller proportion of elderly people out of poverty in 1996 than in 1995.

Largely due to the increasing effectiveness of Social Security in reducing poverty, the number of elderly people lifted out of poverty by all government programs rose from 9.8 million in 1979 to 13 million in 1995. In 1995, government benefits lifted out of poverty more than four-fifths of elderly people who would otherwise be poor, or 82 percent. In other words, government benefit programs reduced poverty among the elderly to less than one-fifth of what it would have been in the absence of the programs.

V. Impact of the Safety Net by Race, Ethnicity, and Age

The safety net significantly reduces poverty for whites, blacks and Hispanics of all ages. In 1996, the number of people in each of these racial or ethnic groups who were poor was far lower than it would have been in the absence of government benefit programs. Safety net programs also greatly reduced the depth of poverty for all groups. There were, however, significant differences in the impact of the programs on poverty by race, ethnic group, citizenship status, and age.

Differences Across Race and Ethnic Group

Although the safety net reduced poverty for all groups in 1996, it moved out of poverty a greater percentage of whites who were poor before receipt of government benefits than of pre-transfer poor blacks and Hispanics.

Much of the reason the safety net was more effective in reducing poverty among whites than among blacks and Hispanics has to do with the demographic composition of the poverty population. Among whites, a much larger proportion of the poor are elderly, and the elderly are the group helped most by the safety net programs. More than 80 percent of elderly people who are poor before receipt of government benefits are lifted from poverty by these benefits. By contrast, only a little over 30 percent of children who are poor before government benefits are received are lifted from poverty by these benefits.

Two of every five white people who were poor in 1996 before government benefits were counted — 40.7 percent — were elderly. By contrast, only 12.7 percent of blacks who were poor before receipt of government benefits were elderly, and only 8.7 percent of pre-transfer poor Hispanics were elderly. (See Table 18.) Conversely, only 19 percent of the white pre-transfer poor were children. But more than 40 percent of pre-transfer poor blacks and Hispanics were children.

In 1996, government benefit programs lifted well over half —56 percent — of pre-transfer poor whites from poverty. These programs moved 35.9 percent of pre-transfer poor blacks and 33.5 percent of pre-transfer poor Hispanics out of poverty. (See Table 19.)

An analysis of the safety net's effect on the depth of poverty shows less disparity by race. Examining the effect of the safety net on the poverty gap enables us to see the degree to which government benefit programs increase the incomes of poor individuals even if their incomes are not raised above the poverty line.

The poverty gap per poor person in 1996 varied somewhat by race and ethnicity.⁽¹⁶⁾ Before counting government benefits, the poverty gap per poor person was greater for whites than for minorities due to the concentration of elderly people among the white poor. The elderly have higher pre-transfer poverty gaps than do other age groups because they have less income from employment.

Not surprisingly, the safety net is most effective in reducing the poverty gap for the elderly. As a result, the reduction in the pove