Analysis of Census Bureau's Income and Poverty Report for 1999 -- Rev. 10/10/00

Census data released September 26 show that the percentage of Americans living in poverty declined to 11.8 percent in 1999, the lowest poverty rate since 1979. Poverty rates for people 65 and over, African-Americans, people living in the South, and families headed by single women fell to all-time lows. Among non-Hispanic whites and Hispanics, poverty rates fell to levels statistically equal to the previous all-time lows for these groups. In addition, the child poverty rate dropped to 16.9 percent, its lowest level since 1979, although this rate remained above its record low. (See summary table at the end of this report.)

The new Census data, which cover income as well as poverty status, also show that median household income reached a new high in 1999.

"This excellent news seems primarily to reflect the effects of the lowest unemployment rate since 1969, as well as rising wages," said Robert Greenstein, director of the Center on Budget and Policy Priorities. "The Census report demonstrates the power of the beneficial effects of a strong economy and especially of a low unemployment rate."

The Census data show that the reduction in poverty in 1999 was concentrated in the nation's cities, with 81 percent of the drop in poverty between 1998 and 1999 occurring in the cities, as distinguished from suburbs and non-metropolitan areas. The number of poor people nationally fell by 2.2 million, from 34.5 million to 32.3 million. Some 1.8 million of this 2.2 million drop occurred in the cities.

The picture the new data paint becomes still brighter in certain respects when a broader measure of poverty favored by many analysts is used. The broader measure includes the Earned Income Tax Credit and government non-cash benefits such as food stamps and housing subsidies and nets out income and payroll taxes. Under the broader measure, the child poverty rate for 1999 fell below its 1979 level. (The overall poverty rate, as distinguished from the child poverty rate, was statistically the same in 1999 as in 1979, under the broader measure.) The year 1979 is the first year for which this more-comprehensive poverty measure is available.

Amidst this positive news, some areas of concern emerge from the Census data. Poverty rates remain high for certain groups of children and families. For example, one in every two children under age of six who lives in a female-headed family — 50.3 percent of such children — lived in poverty last year. In addition, while poverty rates among minority children have declined significantly, one-third of African-American children (33.1 percent) and nearly one-third of Hispanic children (30.3 percent) still were poor in 1999. International comparisons also indicate that the overall child poverty rate in the United States remains higher than the rate in many western European countries and Canada.

Also troublesome are Census data showing that while the number of poor Americans has fallen markedly in recent years, those who remain poor have, on average, grown poorer. Using the broader measure of poverty that includes non-cash benefits and the EITC and subtracts income and payroll taxes, the average poor person fell farther below the poverty line in 1999 than in any year since 1979, the first year for which these data are available.

The average poor person fell $2,416 below the poverty line in 1999. By contrast in 1993, the average poor person fell $2,104 below the poverty line; in 1996, the figure was $2,122. (These figures are adjusted for inflation so they can be compared across years.) The increase since 1996 in the amount by which the average poor person falls below the poverty line reflects the large decline in the proportion of the low-income population receiving means-tested benefits such as food stamps.

The finding that those who remained poor have grown poorer in recent years also applies to children. The average poor child fell farther below the poverty line in 1998 and 1999 than in any year since 1979, the first year for which these data are available.

A similar trend is seen in Census data on the "poverty gap." The poverty gap is the total amount by which the income of all poor households falls below the poverty line. It reflects both the extent of poverty (i.e., how many poor people there are) and the depth of poverty (how far below the poverty line these people fall). When the number of poor people declines, the poverty gap should decline with it. This has not occurred, however, in recent years. Despite a drop between 1995 and 1999 of 3.4 million in the number of poor people, the poverty gap failed to decline. The poverty gap stood at $65 billion in both years.

The Census data show that the reason no progress was made in reducing the poverty gap during this period was that the decline in the number of poor people was offset by an increase in the depth of poverty among those who remained poor. This occurred, the data indicate, because reductions in means-tested benefits offset increases in earnings.⁽¹⁾

The Census data appear to show that income disparities in the United States, which widened sharply between the mid-1970s and 1993, have remained unchanged since 1993. These data do not, however, provide a complete picture of changes in the income distribution. While the Census data include the income received by most households, the Census figures miss a substantial portion of the income that the highest-income households receive.

The official Census income data do not include capital gains income, which has risen dramatically in recent years and is concentrated among those at or near the top of the income spectrum. Recent data from the Internal Revenue Service show that after adjusting for inflation, capital gains income grew more than two-and-one-half times between 1993 and 1998, rising from $163 billion in 1993 to $427 billion in 1998. The IRS data also show that in 1998, some 72 percent of capital gains income went to the 1.7 percent of tax filers with the highest incomes, those with adjusted gross incomes exceeding $200,000.

In addition, the Census data do not record earnings above $999,999. If an individual makes $5 million, the Census records the individual's earnings as $999,999. Other sources of income are similarly capped. While these limitations in the Census data do not significantly affect the data on median income or most other income measures in the Census statistics, they do have a large effect on the data for very high-income households and consequently on measures of income inequality.⁽²⁾

Many experts have noted that the Census data are not very well-suited to measuring the incomes of those at the top of the income scale. When incomes rise much faster for those at the top of the income scale than for other Americans, the Census data can show no change in income disparities when income disparities are indeed widening.

This is what appears to have happened in recent years. Recent IRS data that cover years through 1997 and are based on actual tax returns tell a different story than the Census data. The IRS data show that between 1995 and 1997, the average after-tax income of the one percent of tax filers with the highest incomes jumped 31 percent, or $121,000 (from $397,000 in 1995 to $518,000 in 1997). By contrast, the average after-tax income of the bottom 90 percent of tax filers — that is, everyone except those in the top 10 percent — rose just 3.4 percent.⁽³⁾

In percentage terms, the average income of the top one percent of tax filers thus rose nine times faster than the average income of the bottom 90 percent of filers. Moreover, the average increase of $121,000 in the after-tax income of the top one percent of filers between 1995 and 1997 is substantially larger than the total incomes of the vast majority of American households and three times the income of the median household, which was $40,816 last year.

The IRS data also show that over the somewhat-longer 1993-1997 period, the top one percent of tax filers secured after-tax income gains of 41 percent, on average, while the bottom 90 percent of filers experienced a modest gain of five percent.⁽⁴⁾

The striking improvements in poverty and median income in 1999 appear to be due primarily to the continued robust performance of the economy. The unemployment rate of 4.2 percent was the lowest since 1969. In addition, between 1996 and 1999, the average hourly wages of the lowest-paid workers increased, after declining for much of the previous two decades.

Data from the Quarterly Wage and Employment Series produced by the Economic Policy Institute indicate that wage increases were particularly strong for very low-paid workers between 1996 and 1999.⁽⁵⁾ Among male workers with hourly wages at the 10^th percentile of all male workers, hourly wages increased 9.1 percent between 1996 and 1999, after adjusting for inflation. (Workers with wages at the 10^th percentile have wages lower than 90 percent of all wage-earners and higher than the remaining 10 percent of wage-earners.) Female workers with hourly wages at the 10^th percentile of all female workers saw a 9.7 percent increase in hourly wages during the same period. Wages for both male and female workers at these wage levels had fallen for most of the previous two decades, after adjusting for inflation.

The effects of very low unemployment and rising wages on the low-wage and minority workers highlight the significance of policies that maintain low unemployment rates as well as policies that help restore the value of the minimum wage. The minimum wage was increased in October 1996 and September 1997 but remains substantially below the levels of the 1970's in purchasing power.

**Summary of Poverty and Income Statistics**
	1979	1998	1999

Unemployment Rate	5.8 **	4.5 **	4.2

Official Poverty Rate
Race/Ethnicity
All Persons	11.7	12.7 **	11.8
White	9.0 **	10.5 **	9.8
White non-Hispanic	8.1	8.2 *	7.7
Black	31.0 **	26.1 **	23.6
Hispanic	21.8	25.6 **	22.8
Age
Children Under 6	18.1	20.6 **	18.0
All Children	16.4	18.9 **	16.9
White	11.8 **	15.1 **	13.5
White non-Hispanic	10.1	10.6 **	9.4
Black	41.2 **	36.7 **	33.1
Hispanic	28.0	34.4 **	30.3
Elderly	15.2 **	10.5	9.7
Metropolitan Status
Central Cities	15.7	18.5 **	16.4
Outside Central Cities	7.2 **	8.7	8.3
Non-Metro Areas	13.8	14.4	14.3
Marital Status
Married-Couple Families	5.4 **	5.3 **	4.8
Female-Headed Families	30.4 **	29.9 **	27.8

Alternative Poverty Measures
Poverty Rate Before Counting Government Benefits¹
All	19.5	20.1	19.2
Children	20.1	21.5 **	19.6
Elderly	54.2 **	48.2	47.7

Poverty Rate After Counting Government Benefits²
All	10.2	10.5 **	9.9
Children	13.6 *	14.3 **	12.9
Elderly	13.5 **	8.8	8.4

Median Household Income (in 1999 dollars)
All	37,059 **	39,744 **	40,816
White	38,856 **	41,816 **	42,504
White non-Hispanic	39,403 **	43,376 **	44,366
Black	22,813 **	25,911 **	27,910
Hispanic	29,362	28,956 **	30,735
1 Based on the official poverty level and a definition of income that excludes government benefits, federal taxes, and the EITC. ² Based on the official poverty level and a definition of income that includes cash assistance, food stamps, housing, federal taxes, and the EITC. * indicates a statistically significant change relative to 1999 at the 90% confidence level. ** indicates a statistically significant change relative to 1999 at the 95% confidence level.

The Center on Budget and Policy Priorities is a nonpartisan research organization and policy institute that conducts research and analysis on a range of government policies and programs, and specializes in issues related to fiscal policy and policies and trends affecting low- and moderate-income families. It is supported primarily by foundation grants.

1. This poverty gap measure corresponds to the broader measure of poverty; it counts non-cash benefits and the Earned Income Tax Credit and nets out income and payroll taxes.

2. There also are other reasons that Census data are not the best source for information on the incomes of high-income households. While there are substantial penalties for failing to report all income on tax forms, there is no such penalty for failure to disclose all income to the Census Bureau. Many household surveys have difficulty obtaining the cooperation of high-income households. For further discussion of these issues, see Appendix A to An Analysis of New IRS Income Data, by Isaac Shapiro, Robert Greenstein, and Wendell Primus, Center on Budget and Policy Priorities, September 2000. https://www.cbpp.org/9-4-00inc-app.htm

3. The IRS data reflect adjusted gross income, as reported on income tax returns, as well as the federal income taxes paid on that income. The IRS data do not include forms of income that are not included in adjusted gross income, such as certain government cash benefits. They include data on income taxes but not payroll taxes or other taxes. We focus here on adjusted gross income after federal income taxes, which we refer to here as "after-tax income." All income figures are adjusted for inflation and expressed in 1997 dollars.

4. While the standard Census data do not show a change in income inequality from 1993 to 1999, the alternative definitions of income the Census Bureau provides do suggest a rise in income inequality. These definitions are more comprehensive and include estimates of the effects of taxes, capital gains, and non-cash government benefits. These measures still miss some income for those at the top of the income spectrum; for example, these measures do not capture earnings above $999,999.

Although these alternative definitions still understate income gains at the top, they indicate that income inequality was greater in 1999 than in 1993. For example, under the standard definition of income the Census Bureau uses, the share of the national income that the top five percent of households receive was 21 percent in 1993 compared to 21.5 percent in 1999, not a statistically significant difference. But under the Census Bureau's alternative definition of income that includes capital gains income and the Earned Income Tax Credit and subtracts federal and state income taxes and payroll taxes, the share of the national income the top five percent of households receive rose from 19.6 percent to 21.8 percent over this period, a statistically significant increase.

For additional information on income disparities, see An Analysis of New IRS Income Data, by Isaac Shapiro, Robert Greenstein, and Wendell Primus. This report is available from the Center on Budget and Policy Priorities' web site at https://www.cbpp.org/9-4-00inc-rep.htm, or by calling 202-408-1080.

5. Jared Bernstein, "Annual Wage Growth Slows Despite Low Unemployment" Quarterly Wage and Employment Series, 1999:4.