The effectiveness of government programs in reducing poverty is measured by comparing the number of people who are poor before program benefits are counted to the number who are poor after these benefits are counted. The difference is the number of people moved out of poverty by government benefits.

To make this comparison, we use data collected by the Census Bureau each year on the incomes of the American people. The Census data contain information on cash income from many sources, including government benefits from the Social Security, Unemployment Compensation, Aid to Families with Dependent Children (AFDC), and Supplemental Security Income programs. These data also include government benefits not provided in cash, such as food stamps, school lunches, and housing assistance, as well as federal taxes and the Earned Income Tax Credit.

In this analysis, we use two different measures of poverty. "Pre-transfer" poverty is determined using a definition of income that does not include any government benefits. The number of people counted as poor under this measure is higher than the official poverty count reported by the Census Bureau, because the official count is based on a definition of income that incorporates all cash income, including government benefits provided in cash.

"Post-transfer" poverty, as used here, is based on income that includes benefits from an array of government programs including non-cash benefit programs.¹ Post-transfer poverty also takes into account the effect of federal taxes in lowering or raising income. The post-transfer poverty count is lower than the official poverty count because the official count does not include as income those government benefits, such as food stamps and housing assistance, that are not in the form of cash. The official poverty count also does not take into account the impact of federal taxes or the Earned Income Tax Credit.

In 1995, some 57.6 million people would have been poor if no government benefits had been included as part of their income. After counting cash and non-cash government benefits as income and subtracting taxes, 30.3 million people remained in poverty. The difference — 27.3 million — is the number of people removed from poverty by government benefit programs. 2 This means that nearly half of those who would have been poor without government benefits were moved out of poverty by those benefits. This is the highest percentage moved out of poverty by government benefits for any year since 1980. 3

It should be noted that this is not the same as saying that if none of these government programs existed, nearly twice as many people would be poor. We have no way of observing American society without these programs, so it is impossible to determine what behavioral or institutional changes would occur in such an environment, or how many people would actually be poor. It is possible, however, to correlate the number of people moved out of poverty with changes made in the safety net programs in order to determine whether and how the effectiveness of the safety net in reducing poverty has changed over time.


Children and the Elderly

The Census data also can be used to show the impact of government programs on different groups of people. The data show that the programs have larger impacts on the elderly than on children.

Before any government benefits were counted as income, 17.1 million children were poor in 1995. Counting government benefits, the number of poor children drops to 11.4 million. Some 5.7 million children were lifted out of poverty by government benefit programs. This is one-third of the children who were poor before government benefits are counted as part of their families' incomes.

The impact of government benefit programs on poverty among the elderly is considerably more dramatic. In 1995, some 15.8 million elderly people were poor before counting government benefits. After these benefits are counted, the number of elderly poor plummets to only 2.8 million. In other words, some 13 million elderly people — or 82 percent of those who were poor before government benefits — were lifted from poverty by these benefits.

The safety net programs have a much larger effect in reducing poverty among the elderly than among children because of differences in the types of programs — and in the magnitude of the benefits — for each group. Total expenditures for income maintenance programs are about 10 times greater per elderly individual than per child.

Different Types of Programs

Programs such as Social Security and unemployment insurance are known as social insurance programs. Working Americans and their employers pay taxes that entitle them to benefits from these programs when they retire, become disabled, or lose their jobs. Another category of programs consists of means-tested programs — programs that are available only to people with low incomes. The means-tested programs include AFDC, Supplemental Security Income, food stamps, subsidized school lunches, and housing assistance. Federal taxes, including the Earned Income Tax Credit, constitute a third category.

By comparing the number of people counted as poor when no government benefits are included in their incomes to the number of people who are poor when only social insurance programs are counted, we can determine the impact of the social insurance programs on poverty. In 1995, some 18.2 million people were removed from poverty by the social insurance programs. The means-tested programs removed an additional 7.7 million people from poverty. The impact of federal taxes and the EITC was to remove another 1.4 million people from the ranks of the poor. 4

The social insurance programs have the largest impact in reducing poverty. It also is noteworthy that the effect of the federal tax system is to reduce poverty. Federal income and payroll taxes reduce the income that households have available and thereby increase the number of people who are poor. The Earned Income Tax Credit, however, offsets part or all of these taxes for millions of low-income working households and provides tax credit payments to working households with incomes too low to owe income tax. In 1995, the EITC more than compensated for the impact that federal income and payroll taxes otherwise would have in increasing poverty. As a result, the net effect of the federal tax system was to lower the number of people who were poor.

Although all three categories of government benefit programs thus reduced poverty, different program categories were more effective in alleviating poverty among different groups of people. Social insurance programs were of most assistance in reducing poverty among the elderly. Among children, the means-tested programs were most important.

In 1995, the means-tested programs lifted 3.2 million children out of poverty. In fact, more than half of all children lifted out of poverty by government programs were moved from poverty by the means-tested programs. This should not surprising; many of the means-tested programs serve primarily families with children. Only families with children are eligible for AFDC payments. In addition, half of all food stamp recipients are children, and families with children receive 80 percent of food stamp benefits.

Social insurance programs moved 1.4 million children out of poverty. Federal taxes, including the EITC, lifted another one million children out of poverty. It is striking that taxes were nearly as effective as social insurance programs in moving children out of poverty because taxes alone would be expected to increase rather than reduce poverty. The figures showing that taxes lifted a million children out of poverty reflect the large impact of the EITC, which last year not only offset the effects that federal income and payroll taxes otherwise would have had in increasing child poverty but actually reduced the number of children who were poor.

Among the elderly, social insurance programs — particularly Social Security — accounted for nearly all of the reduction in poverty that government benefit programs caused. In 1995, social insurance programs lifted 12.1 million elderly people out of poverty. This represented three of every four elderly people who were poor before receipt of government benefits.

Other types of programs had smaller effects. Means-tested programs lifted almost 900,000 elderly people out of poverty in 1995, while taxes had a negligible impact on elderly poverty.

Of particular note, safety net programs reduced the elderly poverty rate in 1995 from 50 percent before receipt of government benefits to 9 percent when the benefits are counted, a stunning 41 percentage point reduction in the poverty rate. By comparison, the safety net programs reduced the child poverty rate from 24 percent before benefits are counted to 16 percent when benefits are taken into account, an eight percentage point reduction.

The social insurance programs were far more effective in reducing poverty among the elderly than means-tested programs and taxes together were in reducing child poverty. The social insurance programs, especially Social Security, are much larger than any means-tested program and provide much more generous benefits. In January 1995, the average monthly benefit for a retired worker under the Social Security program was $697, almost twice the AFDC benefit for a family of three in the typical state, which stood at $377. The social insurance programs have a substantially larger effect on poverty among the elderly than among children because the bulk of social insurance benefits go to retired workers.

The Depth of Poverty

In addition to examining the impact of government benefit programs on the number of people in poverty, we also can look at the effect of the programs on the depth of poverty. Here, we look not only at the extent to which people are lifted from poverty but also at the extent to which those who remain poor are made less poor by the programs.

To do this, we use a measure called the "poverty gap." The poverty gap is the amount by which the incomes of all poor people fall below the poverty line. In other words, the poverty gap represents the amount of money that would be needed to lift everyone who is poor to the poverty line.

In 1995, the amount of money needed to lift all poor people to the poverty line, measured before government benefits are counted as part of people's incomes, was $194.5 billion. Government benefits reduced this gap to $59.4 billion. Government programs thus eliminated 70 percent of the poverty gap. That is a very significant reduction in the depth of poverty.

As might be expected, government benefit programs reduce the poverty gap to a far greater extent among the elderly than among children. Before government benefits are counted, the incomes of poor children fell $42.2 billion below the poverty line in 1995.5 When government benefits are taken into account, this gap shrinks to $15.7 billion. In other words, government benefits reduced the depth of poverty among children by $26.5 billion, or 62.8 percent.

Among the elderly, by contrast, government benefits reduced the poverty gap from $68.8 billion to just $5.4 billion in 1995. Government benefits reduced the depth of poverty among the elderly by more than 90 percent. 6

Using Census data, we can determine the effectiveness of government benefit programs in reducing poverty at different points in time. This enables us to compare the effectiveness of the programs in years when the programs were stronger to their effectiveness in years when the safety net programs had been cut back. From this analysis, it is impossible to escape the conclusion that when the safety net is improved, poverty is reduced, and when the safety net is weakened, poverty increases.

One important effect of government benefit programs is to mitigate increases in poverty caused by downturns in the economy. In the recession years of the early 1990s, government benefit programs significantly moderated the effects of a faltering economy. Without these programs, poverty would have risen to a substantially greater degree. By contrast, during the recession of the early 1980s when the safety net was weaker, the programs were much less effective in stemming the increase in poverty.7

During recessions, poverty generally increases. Hence, poverty rose from 1989 — the last year before the recession of the early 1990s — to 1993. Between 1989 and 1993, the number of people in poverty before government benefits are counted rose from 49.9 million to 60.6 million, with more than 10 million people being added to the ranks of the poor. After government benefits are counted, however, the number of poor people increased during this period by only 5.5 million. The effect of government programs was to cut nearly in half the growth of poverty during this recession.

During the recession of the early 1980s, the impact of government programs on poverty was much smaller. Between 1979 — the last year before the onset of that recession — and 1983 when the poverty rate peaked, the number of people who were poor before receipt of government benefits increased almost 10 million. This increase was of a similar magnitude to the increase in the number of people poor before receipt of government benefits during the recession of the early 1990s. But when poverty is measured after government benefits are counted, poverty grew more than 11 million people during the 1979-1983 period. During that period, the increase in the number of people who were poor after receipt of government benefits exceeded the increase in the number who were poor before receipt of government benefits. This occurred because various safety net programs were cut between 1979 to 1983, causing a decline in the number of people lifted from poverty by government benefits.

Another way to examine these trends is to look at changes in poverty rates during this period. The poverty rate before receipt of government benefits went up by an almost identical amount in the two recessions, rising between 3.3 and 3.5 percentage points in both periods. (Thus, from 1979 to 1983,the poverty rate before receipt of government benefits rose from 19.5 percent to 23.0 percent, while it rose from 20.1 percent to 23.4 percent in the period from 1989 to 1993.) But the poverty rate after receipt of government benefits rose nearly three times as much in the recession of the early 1980s as in the recession a decade later. Between 1979 and 1983, the poverty rate after receipt of government benefits climbed 4.5 percentage points. By contrast, between 1989 and 1993, this poverty rate rose 1.6 percentage points, or only about one-third as much. (See Table 1.)


Effect on Children

The difference in the impact that government programs had during these two recessions in lifting people out of poverty is most striking among children. Government benefit programs had a large effect during the recession of the early 1990s in moderating the increase in poverty among children. The number of children who were poor before receipt of government benefits increased by 3.2 million during the 1989-1993 period, rising from 15 million in 1989 to 18.2 million in 1993. After government benefits are counted, however, the number of poor children increased by a substantially smaller amount — two million. Government benefits thus shaved the increase in poverty by 1.2 million children.

Government benefit programs did not have a similar effect during the recession of the early 1980s . The number of children who were poor before receipt of government benefits rose by 3.4 million between 1979 and 1983. But the number who were poor after government benefits are counted rose by an even larger amount, 4.7 million children. Reductions made in government benefit programs in the early 1980s rendered the programs less effective in reducing child poverty in those years.

The most dramatic difference is seen by comparing the changes in child poverty rates during the two recessions. Before government benefits are counted, the child poverty rate increased 5.8 percentage points during the recession of the early 1980s. It climbed 3.5 percentage points during the recession of the early 1990s. 8

The difference between the two recession periods in the extent to which child poverty rates climbed is much greater, however, when government benefits are counted. Between 1979 and 1983, the child poverty rate climbed 7.7 percentage points when government benefits are counted. In contrast, during the 1989-1993 period, it rose two percentage points. In other words, the child poverty rate when government benefits are counted rose only about one-fourth as much during the recession of the early 1990s as it did during the recession of the early 1980s. Safety net benefits affecting families with children were cut in the first recession and expanded in the second recession, with the result that the safety net performed more effectively in stemming increases in poverty during the latter recession.

Changes in Safety Net Programs

Why were government programs so much more effective during the recession of the early 1990s than in the earlier recession in reducing the growth of poverty, especially among children? The answer is that the programs were weakened during the recession of the early 1980s and strengthened during the recession of the early 1990s. By 1993, the programs were substantially stronger than they had been a decade earlier.

Table 2 on the next page shows several key measures of the performance of the safety net. By nearly all measures, the safety net reached fewer people and provided less in benefits in 1983 than in 1979. During the recession of the early 1980s, the safety net contracted.

In the early 1980s, post-secondary student benefits were eliminated in the Social Security program, and the Social Security disability insurance program was made more restrictive, with disability awards per 1,000 insured workers declining sharply. In addition, continuing disability reviews increased significantly, with this resulting in 182,000 cessations of disability payments in 1983. Food stamp benefits also were cut, AFDC benefits eroded badly as a result of inflation, and the combined AFDC and food stamp benefit in the typical state for a family of three with no other income fell from 74 percent of the poverty line in 1979 to 68 percent in 1983. Furthermore, many low-income working mothers were made ineligible for AFDC, and the proportion of poor children receiving AFDC fell from 68 percent to 50 percent during the 1979-1983 period.

In addition, more low-income working families had to begin paying federal income taxes in these years, as the key elements of the tax code that shield working poor families from federal income taxes (the personal exemption, the standard deduction, and the EITC) eroded to inflation. In 1979, federal income and payroll taxes (including the EITC) consumed less than one percent of the income of a family with earnings at the poverty line. By 1983, these taxes ate up 6.5 percent of such a family's income. For a family of four with earnings at the poverty line, these taxes rose from 4 percent of income in 1979 to more than 10 percent in 1983.

The curtailment of AFDC benefits and the increase in federal taxes had an especially sharp impact on single-parent working families with children. In 1979, there were 36 states in which a family of three with earnings equal to 75 percent of the poverty line was lifted out of poverty by the combined effect of AFDC, food stamps, and taxes. By 1983, however, the increases in the federal taxes these families owed and the cutbacks in their eligibility for AFDC had taken a toll. In that year, there was only one state in the nation in which a mother with earnings equal to 75 percent of the poverty line was brought up to the poverty line by benefits and taxes.

By the early 1990s, this picture had changed, and the safety net had grown substantially stronger. During the period from 1989 to 1993, disability awards increased, cessations of disability benefits were at a fraction of their 1983 levels, and the percentage of the unemployed receiving unemployment benefits climbed markedly. Nearly one million fewer unemployed individuals exhausted federal supplemental or emergency unemployment benefits in 1993 than in 1983.

In addition, many of the reductions made in the early 1980s in programs important to poor children had been restored at least partially by the 1990s, and some expansions in these programs had occurred. In the food stamp program, a number of the reductions had been reversed, and basic food stamp benefit levels had been raised. In AFDC, all states were required to extend benefits to two-parent families, and some states secured waivers enabling more low-income working parents to qualify for AFDC. By 1993, the proportion of poor children receiving AFDC had climbed to 61 percent from 50 percent a decade earlier. The proportion of poor children receiving food stamps also increased substantially. On the other hand, the combined value of AFDC and food stamp benefits eroded further as most states declined to keep AFDC benefits even with inflation, and a number of states made large reductions in general cash assistance benefits for poor single adults without children.

Of particular importance, the EITC has been greatly expanded since 1986. The enlargement of the EITC, coupled with improvements in food stamp benefits, increased dramatically the number of states where a mother with earnings equal to 75 percent of the poverty line is lifted from poverty by government benefits and taxes. By 1993, some 49 states were in this category. 10

Finally, because of changes in the eligibility of low-income disabled children for benefits under the Supplemental Security Income program, 771,000 disabled children received SSI payments in 1993. Four years earlier, fewer than 300,000 such children had received SSI assistance.

The Relationship Between Changes in the Safety Net Programs and Changes in Their Effectiveness in Reducing Poverty

Changes in the effectiveness of government benefit programs in reducing poverty closely track the cuts and increases made in these programs over the years examined here. Social insurance programs are a case in point. Some 34 percent of the individuals who were poor before receipt of government benefits were removed from poverty by social insurance benefits in 1979. By 1983, the fraction of people removed from poverty by these programs had fallen to 30 percent. (See Table 3.) As explained in the previous section, Social Security and unemployment insurance — two of the major social insurance programs — were restricted during this period.

In the early 1990s, by contrast, the social insurance programs were not curtailed. And during that period, the proportion of the poor lifted from poverty by these programs did not decline.

Similarly, when sizeable cuts were made in AFDC, food stamps, and several other means-tested benefit programs in the early 1980s (as discussed in the previous section), the percentage of those poor before government benefits who were lifted from poverty by means-tested programs plunged from 15 percent in 1979 to 10 percent in 1983. The number of people removed from poverty by means-tested programs fell by 1.4 million during this period. (See Table 3.)

In the early 1990s, by comparison, the percentage of people lifted from poverty by means-tested programs stayed about the same. Evidently, the expansions in SSI for disabled children and, to a lesser degree, in food stamps offset the effects of continued reductions in the purchasing power of AFDC benefits and sharp cuts in general cash assistance in some states.

Most striking is the change in the role played by the tax system. Between 1979 and 1983, the tax system added 1.4 million people to the ranks of the poor, as the EITC the personal exemption, and the standard deduction lost ground to inflation and payroll taxes climbed. By contrast, between 1989 and 1993, the number of people made poor by the tax system declined by 800,000.

Particularly worthy of note are data showing the effects on children of changes in the means-tested programs. Between 1979 and 1983, the percentage of children removed from poverty by means-tested benefits fell nearly by half, reflecting the large reductions made during this period in means-tested benefits for low-income families with children. Furthermore, the proportion of children pushed into poverty by federal taxes more than tripled during this period. From 1989 to 1993, by contrast, the proportion of children removed from poverty by means-tested programs changed little, and the tax system began to reduce — rather than increase — the number of poor children as large EITC expansions took hold. (See Table 4.)

Changes in the Depth of Poverty During the Two Recessions

Another way to look at these trends is to compare the changes that occurred during the two recessions in the magnitude of the poverty gap. (The poverty gap grows during recessions as more people fall into poverty and as many of those who are poor become poorer.) Government benefit programs moderated the growth in the poverty gap by a substantially smaller amount during the recession of the early 1980s than during the recession of the early 1990s.

Between 1979 and 1983, the poverty gap, as measured before government benefits are counted, increased by $35.2 billion. (All dollar values are adjusted for inflation to be comparable to 1995 dollars.) During the same period, the poverty gap after government benefits are counted increased $19.9 billion. Government programs thus offset a little more than two-fifths of the increase that otherwise occurred in the poverty gap.

During the recession of the early 1990s, the poverty gap before government benefits are counted grew by $38.2 billion. The gap grew only $14 billion, however, after government benefits are counted. During this period, government benefit programs offset nearly two-thirds of the increase in the poverty gap, registering a much stronger performance than in the recession a decade earlier.

Following both the recession of the early 1980s and the recession of the early 1990s, the U.S. economy recovered, and poverty declined. Comparing poverty during the two recovery periods provides further evidence of the role of the safety net programs in reducing poverty.

In the first two years after poverty rates peaked during each downturn — the years from 1983 to 1985 and from 1993 to 1995 — poverty rates before government benefits are counted declined by identical amounts. Poverty rates after government benefits are counted declined twice as much, however, in the 1993-to-1995 period as in 1983 to 1985. During the 1993-1995 period, large increases in the earned income tax credit and improvements in several other safety net programs took effect.

In addition, although the poverty rate before receipt of government benefits was higher in 1995 than in 1989, the poverty rate after receipt of government benefits was lower in 1995 than in 1989. By 1995, the poverty rate after government benefits are counted had fallen below the pre-recession level. In fact, the poverty rate after receipt of government benefits was lower in 1995, both for the overall population and for children, than in any year since 1980.

Comparing the 1983-1985 and 1993-1995 Periods

In both the 1983-1985 period and the 1993-1995 period, the poverty rate before receipt of government benefits declined 1.5 percentage points from its recession peak. Poverty rates after receipt of benefits declined much more sharply in the 1993-1995 period, however, than in 1983-1985.

Between 1983 and 1985, the poverty rate after receipt of government benefits edged down 1.1 percentage points. Between 1993 and 1995, by comparison, the poverty rate after receipt of government benefits fell 2.1 percentage points, or twice as much. 11

Comparing the Peak Recovery Year of 1989 to 1995

In 1989, when the economic recovery of the 1980s reached its high point, 20.1 percent of the population was poor before receipt of government benefits. Six years later, in 1995, the economy was still rebounding from the recession of the early 1990s, and pre-transfer poverty remained above the 1989 level. Some 21.9 percent of the population was poor before receipt of government benefits in 1995.

Yet while the pre-transfer poverty rate was significantly higher in 1995 than in 1989, the post-transfer poverty rate was lower in 1995 than it had been in 1989. At 11.5 percent, the poverty rate after receipt of government benefits was lower in 1995 than in any year since 1980.

As noted earlier in this report, safety net programs expanded substantially during the first half of the 1990s. The earned income tax credit was enlarged, more low-income children were made eligible for SSI, and some improvements were made in food stamp benefits. The result was a decreased poverty rate when government benefits are taken into account.

The effects are particularly striking for children. The child poverty rate before receipt of government benefits climbed from 22.8 percent in 1989 to 24.2 percent in 1995. But the child poverty rate after receipt of government benefits fell from 18.0 percent to 16.2 percent during this period, hitting a 15-year low in 1995. (See Table 5.)

In 1989, a little more than one in every five children who were poor before receipt of government benefits — 22.8 percent — were lifted from poverty by these benefits. By 1995, one in every three such children — 33.1 percent — was lifted from poverty by these benefits.

What Happened During the Early 1990s to the Safety Net for Children?

These results are not surprising given the legislative changes between 1989 and 1995. Most improvements in safety net programs during this period were targeted on low-income families with children.

The earned income tax credit was increased twice during this period by substantial amounts — once in 1990 and again in 1993. More low-income disabled children became eligible for SSI. Food stamp benefits were improved. In addition, participation in AFDC and food stamps increased significantly for children in low-income families. These changes resulted in a significantly stronger safety net for children.

The largest change occurred in the tax area. In 1989, the combined effect of federal income and payroll taxes and the EITC was to add 400,000 children to the ranks of the poor. By 1995, federal taxes and the EITC were decreasing poverty by one million children.

More than half of the increase between 1989 and 1995 in the number of children lifted from poverty by government programs occurred as a result of improvements in the tax system. The only significant changes in federal income and payroll tax rules during this period that aided low-income families with children were the large EITC expansions. It is apparent that those expansions were responsible for the majority of the improvement in the effectiveness of government programs over this period in reducing child poverty.

The enlargement of means-tested benefit programs also helped. These programs removed 800,000 more children from poverty in 1995 than in 1989.

The poverty trends since 1979 are similar for white, black, and Hispanic children and largely parallel the trends for all children. For example, during the recession of the early 1980s, the poverty rate after receipt of government benefits rose more sharply among all three groups of children than did the poverty rate before receipt of government benefits. By contrast, during the recession of the 1990s, the opposite was true among all three groups — the poverty rate after receipt of government benefits rose much less than the poverty rate before receipt of government benefits.

Of particular note are the patterns among white, black, and Hispanic children during the period from 1989, the peak economic recovery year of the 1980s, to 1995. The poverty rate after receipt of government benefits declined for all three groups of children between 1989 and 1995. Among all three groups, this downward trend in poverty occurred largely or entirely because the safety net programs — and especially the earned income tax credit — lifted a substantially larger proportion of children out of poverty in 1995 than in 1989.

White Children

In 1989, government benefit programs lifted out of poverty almost 25 percent of the non-Hispanic white children who were poor before receipt of government benefits. Just six years later, in 1995, the programs lifted nearly 39 percent of such children from poverty, the largest proportion on record.

As a result, in 1995 the poverty rate after government benefits are counted fell to 8.5 percent among non-Hispanic white children, the lowest level on record for these children. (See Table 6.) None of this reduction in poverty can be attributed to improvements in the economy, since the poverty rate before government benefits are counted was no lower in 1995 than it had been in 1989.

Many factors contributed to this progress. Most important was the substantial enlargement of the earned income tax credit. In 1989, the federal tax system, which includes the earned income tax credit, increased the number of non-Hispanic white children who were poor. By 1995, because of the EITC expansions, the federal tax system removed from poverty 8.1 percent of the non-Hispanic white children who otherwise fell below the poverty line, or more than one million such children. (See Table 7.)

The EITC expansions also are the principal reason that the poverty rate after government benefits are counted was lower among non-Hispanic white children in 1995 than in any previous year for which these data exist. This is seen by examining various trends relating to non-Hispanic white children for the period from 1979, the first year for which these data are available, to 1995.

As Table 6 shows, the poverty rate for non-Hispanic white children before government benefits are counted was actually higher in 1995 than in 1979. Furthermore, as indicated in Table 7, social insurance programs and means-tested programs each lifted smaller proportions of non-Hispanic white children out of poverty in 1995 than they did in 1979. Thus, neither improvements in the economy nor changes in social insurance or means-tested programs can explain why the poverty rate among non-Hispanic white children after government benefits are counted was lower in 1995 than in 1979 or any intervening year.

This progress occurred because of changes in the tax system. The tax system lifted many more non-Hispanic white children out of poverty in 1995 than in 1979. The improvement in this area was so large that it more than offset both the rise in the poverty rate before government benefits are counted and the decline in the effectiveness of social insurance and means-tested programs in reducing poverty among white children.

Black Children

Among non-Hispanic black children, the poverty rate after receipt of government benefits fell sharply between 1989 and 1995. By 1995, this poverty rate was lower than in any other year on record except 1979. (See Table 8.)

Here, also, the expansion of the safety net in recent years was the principal factor behind this progress. The poverty rate among black children before receipt of government benefits edged down only modestly between 1989 and 1995. But the poverty rate after receipt of government benefits fell substantially. The decline in poverty among black children during this period occurred primarily because the proportion of black children poor before receipt of government benefits who were lifted from poverty by those benefits climbed from 18.2 percent in 1989 to 31.4 percent in 1995. (See Table 9.)

About half of the improvement between 1989 and 1995 in the effectiveness of the safety net programs in lifting black children from poverty resulted from the strengthening of the EITC. The other half resulted from improvements in means-tested benefit programs. (See Table 9.)

If one examines the longer period from 1979 to 1995, the role of the EITC in reducing poverty among black children looms still larger. Both social insurance programs and means-tested programs lifted smaller proportions of black children out of poverty in 1995 than they did in 1979. But the tax system lifted a much larger proportion of these children from poverty in 1995 than in 1979 or any intervening year.

Hispanic Children

Among Hispanic children, the poverty rate after government benefits are counted was only slightly lower in 1995 than in 1989. The data indicate that without the expansions of the safety net programs, poverty among Hispanic children would have increased substantially over this period.

The poverty rate among Hispanic children before government benefits are counted climbed from 39.8 percent in 1989 to 44.8 percent in 1995. Yet the poverty rate after receipt of benefits edged down slightly, from 33.1 percent in 1989 to 32.5 percent in 1995. This reflected the strengthening of the safety net programs. (See Table 10.)

An increase in poverty among Hispanic children was averted because the proportion of Hispanic children lifted from poverty by government benefits increased sharply during this period. In 1989, some 17 percent of Hispanic children poor before receipt of government benefits were raised out of poverty by those benefits. In 1995, some 27.4 percent of such children were lifted from poverty by the benefits programs. (See Table 11.) Most of this progress is attributable to the EITC expansions. In 1989, the federal tax system added nearly as many Hispanic children to the ranks of the poor as all of the social insurance programs combined lifted from it. By 1995, the tax system not only was no longer adding Hispanic children to the ranks of the poor — it was lifting more Hispanic children from poverty than were the social insurance programs.

Even so, the poverty rate among Hispanic children after government benefits are counted, while slightly lower in 1995 than in 1989, remained well above its 1979 level. This appears to be due in part to immigration; partly as a result of immigration, there were nearly twice as many Hispanic children in 1995 as in 1979, and a substantial proportion of these immigrants had low incomes in 1995. The decline over this period in wages for low-paid jobs also is likely to have had a significant effect; the proportion of poor Hispanics living in working families exceeds the proportions of poor whites and poor blacks in working families. Downward pressure on wages for low-wage jobs consequently has a somewhat more pronounced effect on Hispanic poverty than it has on poverty among other racial/ethnic groups.

Data and Methodology

All data and tabulations used in preparing this analysis are based on unpublished tabulations from the U.S. Bureau of the Census. The poverty thresholds are the same as those used by the Census Bureau in preparing the official definition of poverty. However, the income used in this report has been expanded to include all near-cash benefits including food stamps, housing benefits, school lunch benefits and the earned income tax credit. Federal income and employee payroll taxes are subtracted from income. Valuing near-cash benefits this way is identical to the procedures employed by the Census Bureau in preparing its experimental measures of poverty. It creates a fuller picture of the other possible sources of income available to poor individuals. This expanded definition of family income is then compared against the official poverty thresholds to determine whether a family is poor. All other aspects of the analysis — the definition of a family, weighting procedures, etc. — are the same as under the official poverty definition.

The definition of poverty used in this study is identical to the definition used by the Office of Management and Budget (OMB) in preparing poverty impact estimates of the House and Senate welfare bills in November 1995. The estimates of the number of children removed from poverty under current law are somewhat smaller than the OMB estimates because there is no adjustment for underreporting of transfer income as there was in the Administration study and because the EITC was assumed to be fully implemented in 1993. However, since we use the same procedures in all years, the errors of underreporting cancel each other out and leave the conclusions of this study unaffected.

This definition is also identical to estimates of poverty published by the Committee on Ways and Means in various editions of the Green Book. This definition is also consistent with the proposal advanced by the National Academy of Sciences panel on the measurement of poverty. Compared to the NAS poverty definitions, poverty estimates presented here are probably lower because they do not account for work expenses, out-of-pocket medical expenses and the somewhat higher poverty thresholds that the NAS has recommended.

All adjustments for price changes are made using the CPI-U consumer price index.

Guide to Understanding Appendix Tables A-1 through A-15

Tables A-1 through A-15 are identical in format. The population count for the specific group analyzed in the table is presented in row 1 for selected years between 1979 and 1995. The first column is 1979, the first year for which the Census Bureau gathered food stamp and housing information in its population survey. It also is the last year before the recessions of 1980 and 1982. The poverty rate reaches a peak in 1983. It steadily declines until 1989 then increases until 1993, and declines from 1993 to 1995. The last two years in the analysis are 1994 and 1995, the most recent years for which data exists.

Rows two through five were supplied by the Census Bureau and correspond to tables presented in the 1993 Green Book on pages 1,354 through 1,366. The last year presented in that analysis was 1991. All other information in the table including the last three columns are derived from rows 1-5. Examples are from Table A-1 using 1979 information.

Cash income before transfers — the number of individuals who are poor before counting any government transfers. Income includes earnings from wages and self-employment, dividend, interest, rent, child support received and all other forms of cash non-governmental income. In 1979, there were 43.4 million poor individuals before any government benefits are added.

Plus social insurance — the number of individuals who remain poor after adding in all forms of social insurance to income in the previous row. The main sources of income added are social security, including disability benefits, unemployment compensation, workers' compensation and veterans benefits. In 1979, the number of individuals who are poor fell to 28.8 million, a reduction of 14.6 million, when all forms of social insurance benefits were added.

Plus means-tested benefits — the number of individuals who remain poor after adding in all means-tested benefits to income in the previous row. This would include Aid to Families with Dependent Children (AFDC), General Assistance (GA), food stamps, housing benefits, school lunch, Supplemental Security Income (SSI), and means-tested veterans pensions. The number of poor individuals who remain poor after these means-tested benefits are added to income falls to 22.1 million.

Less federal taxes — the number of individuals who remain poor after adding in federal tax policy to income in the previous row. The taxes included are the FICA employee payroll taxes, income taxes and the earned income tax credit. Taxes can either add to poverty or reduce poverty depending upon the size of the earned income tax credit to taxes paid. Numbers in parenthesis indicate that taxes added to the number of individuals in poverty. In 1979, the number of poor individuals increases to 22.7 million when federal taxes are subtracted from income.

Poverty Rate — This block expresses the number of poor individuals as a rate and is calculated by dividing the previous four rows by the total population and multiplying by 100 to express the rate as a percentage. In 1979, some 19.5 percent of the population was poor before any government benefits are counted. This falls to 12.9 percent when social insurance benefits are added, falls still further to 9.9 percent when means-tested benefits are added and increases to 10.2 percent when federal taxes are subtracted from income.

Number removed from poverty due to — In consecutive order, this block of numbers is obtained by subtracting row 2 minus row 3, row 3 minus row 4 and row 4 minus row 5. Conceptually this is the number of individuals removed from poverty by the addition of the various types of government benefits. Thus, in 1979, some 14.6 million individuals are removed from poverty when social insurance benefits are added, another 6.65 million are removed when means-tested benefits are added and 0.6 million individuals additional individuals are counted as poor when federal taxes are subtracted.

Percentage of individuals removed from poverty — This concept is the percentage of individuals removed from poverty divided by the total number of poor individuals before any government programs are added. For 1979, the 14.6 million individuals removed from poverty by social insurance benefits are 33.7 percent of the entire 43.4 million people who are poor before any government benefits are counted. In all, 47.6 percent of the pre-transfer poor are removed from poverty by the addition of government benefits.

Reduction in poverty rate due to — expresses the number of individuals removed from poverty by each major source of income as a percentage of the total population. The 14.6 million individuals removed from poverty in 1979 by social insurance benefits can also be expressed as 6.6 percent of the entire population simply by dividing 14.6 million by the total population of 222.9 million.

Guide to Understanding Tables A-16 through A-19: Poverty Gap Data

Tables A-16 through A-19 are identical in format and represent poverty gap data for all persons, elderly, children, and individuals age 18-64. Row one of each table presents the population count for the specific group analyzed in the table for selected years 1979-1995.

Poverty gap is the total amount of income necessary to lift all persons who are below the poverty line up to that level for a defined income concept.

Rows two through five show how the poverty gap changes as different government benefits such as social insurance, means-tested benefits and federal taxes as defined by the previous appendix are added to the pre-transfer poverty gap in row 2. By subtracting or adding these figures we get a change in the poverty gap. Rows two through five come directly from tables supplied by the Census Bureau reporting poverty gap data for selected years. All data from previous years has been converted to 1995 dollars.

Example: Looking at poverty gap data for All Persons in 1979, it would take $139.5 billion to raise to the poverty line all poor families who are below the poverty line before government transfers. Adding social insurance benefits reduces the gap to $71.6 billion, while adding means-tested transfers benefits lowers the poverty gap further to $40.1 billion. Taking federal taxes into account increases the poverty gap to $40.7 billion. After government transfers the poverty gap of $40.7 billion means it would take that amount of income to bring all those remaining in poverty to the official poverty line.

Poverty gap per person — These rows were calculated by dividing the poverty gap by the population of that particular year. This adjusts the gap for changes in population. The result is the average amount of income per person in the U.S. needed to pull all persons in that particular category out of poverty. After means-tested programs are subtracted, for example, it would take an average of $180 per person to raise all families to the poverty line. After federal taxes are added, however, that number increases to $183 per person.

Reduction in poverty gap per person by program — This section measures the reduction in the per person poverty gap caused by each particular program. It is calculated by subtracting the poverty gap per person from the previous section after each government program is accounted for. In 1979, for instance, social insurance decreased the poverty gap per person from $626 to $321 or by $305. On the other hand, federal taxes contributed to the poverty gap by adding an average of $3 per person to that gap. After all government programs were taken into account, the poverty gap was reduced by an average of $443 per person.

Percentage reduction in poverty gap — This section measures the reduction in the poverty gap caused by each program as a percentage of the total gap. It is calculated by multiplying the reduction in the poverty gap per person per program (the previous block of numbers) by the poverty gap per person prior to including any government benefits. Social insurance in 1979 decreased the poverty gap by $305 per person. Dividing this number by the poverty gap before transfers of $626, we get a 48.7 percent reduction in the poverty gap.

1. The programs include Social Security, Unemployment Compensation, AFDC, Supplemental Security Income, food stamps, school lunches, and housing assistance. The effects of federal income and payroll taxes and the Earned Income Tax Credit also are included. Medical insurance programs, such as Medicare and Medicaid, are not included because these programs are insurance programs and do not provide income or benefits used for basic living expenses like food or rent. In addition, when the poverty line was developed, it did not take the costs of medical care into account. If medical coverage were to be included as income, the poverty line would have to be adjusted to make it consistent. The definition of income employed here, counting major non-cash benefits other than health insurance, is similar to that recommended last year by a distinguished panel of the National Academy of Services and to the definition used by the Urban Institute and the Office of Management and Budget in studies they issued over the past year on the effect of various welfare and budget proposals on poverty.

2. These figures understate the effectiveness of safety net programs somewhat in lifting people out of poverty because they are based on Census data that reflect most but not all of the government benefits that the low-income population receives. A comparison of Census data with data on the actual number of households receiving assistance under these programs shows that the Census data understate the number of households receiving some benefits, particularly means-tested benefits such as food stamps and Aid to Families with Dependent Children. This occurs because not all households correctly recall or report to the Census Bureau on all of the benefits received during the course of the previous year. If the Census data reflected all benefits received, the number and proportion of households lifted from poverty by these benefits would be somewhat higher. Thus, the estimates in this report should be regarded as conservative.


3. The percentage of the "pre-transfer" poor who were moved out of poverty by government benefits was identical in 1995 to the percentage moved out of poverty by these benefits in 1979. The year 1979 is the first year for which these data are available.In 1979, government benefit programs lifted out of poverty 47.6 percent of those who were poor before receipt of government benefits. In 1995, government benefits lifted out 47.4 percent of those who would have been poor without government benefits. The years 1995 and 1979 thus are tied as the years in which government benefits lifted the largest proportion of the pre-transfer poor out of poverty. (The difference in the percentages for these two years is too small to be statistically meaningful.)There also is another way to gauge the anti-poverty effectiveness of benefit programs — to examine the percentage of the total U.S. population lifted out of poverty by government benefits (rather than the percentage of those poor before government benefits who were lifted from poverty by those benefits). If measured in this way, 1995 is the year in which the anti-poverty effects of the programs were strongest. In 1995, government benefit programs lifted out of poverty 10.4 percent of the total U.S. population. This is the largest percentage on record.

4. These figures somewhat understate the effect of the programs in moving people from poverty for the reasons explained in footnote 2. Most of this understatement occurs in the area of the means-tested programs.

5. The poverty gap concept is a household concept. If a household of four consisting of two adults and two children has income that falls $2,000 below the poverty line, the household's poverty gap is $2,000. To determine the poverty gap for children in the household, the $2,000 poverty gap for the household is divided by the total number of individuals in the household — in this case, four — to convert it to a per capita poverty gap of $500. The per capita amount is then multiplied by the number of children in the household, which is two in this example. As a result, in this example, $1,000 of the household's $2,000 poverty gap is associated with children.

6. The effects of different types of government programs in reducing the depth of poverty also can be measured. In 1995, benefits from the social insurance programs closed $89.6 billion — or nearly half — of the $194.5 billion poverty gap that existed before government benefits are counted. The benefits provided by means-tested programs reduced the poverty gap another $43.4 billion. The net effect of income and payroll taxes, including the EITC, was to reduce the poverty gap $2.1 billion.

Among children, the means-tested programs had the largest impact in shrinking the poverty gap. Benefits from means-tested programs reduced the child poverty gap by $19 billion in 1995, substantially surpassing the $5.7 billion impact of social insurance programs or the $1.8 billion impact of the combination of federal taxes and the EITC. As would be expected, among the elderly, social insurance programs had the greatest impact on the poverty gap. These data underscore the finding that social insurance programs have a larger effect than any other type of program in reducing poverty among the elderly, while means-tested programs have the largest effect in reducing poverty among children.

7. During the early 1980s, the economy officially experienced two back-to-back recessions, the first from January 1980 to July 1980 and the second from July 1981 to November 1982. In the early 1990s, the recession officially began in July 1990 and ended in March 1991. Recessions begin when the economy experiences two consecutive quarters of negative economic growth and end when economic growth becomes positive again. Thus, the official end of a recession is the low point of the economy and is often followed by a period of high unemployment and high poverty rates. This part of the analysis compares poverty in the final years before each of the last two recession (1979 and 1989, respectively) to poverty in the years when poverty rates peaked following each recession (1983 and 1993).

8. The child poverty rate before receipt of government benefits climbed from 20.1 percent in 1979 to 25.9 percent in 1983, an increase of 5.8 percentage points. During the recession of the early 1990s, it rose from 22.8 percent in 1989 to 26.3 percent in 1993.

9. A similar trend is seen in the percentage of individuals of all ages — rather than just children — removed from poverty by government benefits. The percentage of people poor before government benefits who were lifted from poverty by those benefits fell from 47.6 percent in 1979 to 36.1 percent in 1983, edged up to 40.3 percent in 1989, increased further to 41.7 percent in 1993, and climbed to 47.4 percent in 1995.

10. This analysis, which is based upon data in the 1993 Green Book, includes the District of Columbia but excludes Alaska and Hawaii. Thus, 49 is the total number of jurisdictions examined.

11. The poverty rate after receipt of government benefits declined from 14.6 percent in 1983 to 13.5 percent in 1985. During the 1993-1995 period, the rate fell from 13.6 percent to 11.5 percent.