The Effects of Government Taxes and Transfers on Income and Poverty: 2004

 

I.                     Introduction

II.                   Income Definitions

III.                  Income

IV.               Poverty

V.                 Methodology and Alternative Income Components

VI.               References

 

 

I. Introduction

 

In August 2005, the Census Bureau released its annual report on income, poverty, and health insurance coverage in the United States. The income and poverty figures in that report were based on money income alone and did not include the effect of important public programs such as the Earned Income Tax Credit and noncash assistance such as food stamps and public or subsidized housing programs. As in previous years, the Census Bureau is now releasing a study that includes the effect of these and other government programs on economic summary measures, such as median household income, the Gini Index of income inequality, and the percentage of people below the poverty level. This release includes fewer alternative income definitions than previous reports to provide a more focused assessment of the effect of government programs (cash and noncash transfers and taxes, including the effect of the Earned Income Tax Credit) on income and poverty summary measures. Unlike previous reports, the poverty estimates shown here use a single set of thresholds that differ from the official poverty thresholds, see Section V for details. The resulting alternatives illustrate how poverty estimates are affected when various types of noncash benefits are treated as income and when taxes are taken into account, while holding constant the measure of need (the thresholds).

 

The rest of this release is divided into five sections. Section II describes the four income definitions used in the report. Section III reviews the impact of taxes and transfers on household income measures (median income and income inequality) using these four definitions of income. Section IV discusses the effect of government programs on the percentage of people below the poverty thresholds by using the alternative definitions of income. Section V provides additional details and background on the concepts used for this release. Finally, Section VI includes references for more technical details.

 

 

II. Income Definitions

 

Money Income: This includes all money income received by individuals who are 15 years or older. It consists of income before deductions for taxes and other expenses and does not include lump-sum payments or capital gains. It also does not include the value of noncash benefits such as food stamps. This income concept is the basis for the official U.S. poverty measure.

 

Market Income: Includes money income except government cash transfers; includes imputed realized capital gains and losses; includes imputed rate of return on home equity; and subtracts imputed work expenses.[1] Market income can serve as a starting point for examining the effect of government activity on income and poverty estimates. For example, comparing market income with post-social insurance income reveals the impact of non-means tested transfers like Social Security on median household income, income inequality measures, and the percentage of people below their poverty thresholds.

 

Post-Social Insurance Income: Includes money income except government means-tested cash transfers; includes imputed realized capital gains and losses; includes imputed rate of return on home equity; and subtracts imputed work expenses. A comparison of post-social insurance income to disposable income shows the net impact of means-tested government transfers (both cash and noncash) and taxes.

 

Disposable Income: Includes money income; includes the value of noncash transfers (food stamps, public or subsidized housing, and free or reduced-price school lunches); includes imputed realized capital gains and losses; includes imputed rate of return on home equity; and subtracts imputed work expenses, federal payroll taxes, federal and state income taxes, and property taxes on owner-occupied homes.[2] A side-by-side comparison of market income and disposable income the net impact of government transfers and taxes on income and poverty estimates.

 

 

III. Income

 

Table 1 shows the effects of government taxes and transfers on income measures using the traditional money income concept and three alternative definitions:  market income, post-social insurance income, and disposable income.

 

In 2004, median household money income was  $44,389 for the United States. Market income represents resources available to people and families based on market activities and does not include income from government sources. Income under this definition can serve as a reference point for evaluating the effects of those government sources. U.S. median household income was $41,648 under the market income definition, or 6.2 percent lower than median household money income.

 

Post-social insurance income is defined as market income with non-means-tested cash transfers, such as Social Security, added back. Including those transfers increased median household income by 10.4 percent to $45,968.

 

Disposable income represents the net income households have available to meet living expenses. It includes all resources based on market activities, the value of government transfers and deducts taxes. The result of these additions and subtractions lowered median household income 13.5 percent to $39,754 or 10.4 percent below the money income concept. 

 

The money income definition and two of the alternative definitions showed no change in overall real median household income between 2003 and 2004. However, market income declined by 0.8 percent. Table 1 shows year-to-year changes for each income definition for selected demographic groups.

 

In three regions real median household income showed the same year-to-year changes for all definitions. The Midwest declined, while the Northeast and South did not change. In contrast the West showed a decline in post-social insurance income, but no change in the other definitions.

 

By type of household, real median income between 2003 and 2004 showed no change for family households across all four definitions. Median income for nonfamily households showed declines under post-social insurance income and disposable income.

 

By work experience of the householder, all four definitions showed a decline in real median household income between 2003 and 2004 for households in which householders worked. In comparison, households in which the householder did not work showed no change in any definitions.

 

Table 2 shows measures of income inequality—the shares of aggregate household income by income quintile and Gini Index—for each income definition.  As with money income, there was no change in the Gini Index between 2003 and 2004 for two of the three alternative income definitions.  The Gini Index declined between 2003 and 2004 for disposable income accompanied by an increase in share of aggregate income in the lowest quintile.  Even though the Gini Index showed no change between 2003 and 2004 for the other income definitions, there were some shifts in shares of income by quintile--money income showed a decline in shares in the second quintile, market income showed declines in shares in the third and forth quintile, and post-social insurance showed declines in shares for the second and fourth quintile, with an increase the highest quintile.

 

When looking at how government programs redistribute income, the distribution of income under the market definition is more unequal than money income alone. The Gini Index for market income is 10.2 percent higher than for money income. This fact is further reinforced by the shares of aggregate income increasing in the two top quintiles and declining in the bottom three. Moving to the post-social insurance definition and the disposable income definition results in a more equal income distribution. The Gini Index declines 9.5 percent between market income and post-social insurance income and declines another 10.9 percent going from post-social insurance to a disposable income concept. Again, the quintile measures support this trend with a decline in shares of aggregate income for the highest quintile and increases in shares for the lower four quintiles.[3]

 

 

IV. Poverty

 

Official poverty rates for 2004 were released in August 2005. In order to assess the impact of government taxes and transfers on poverty estimates, this data release uses revised income definitions. In addition, the thresholds used to determine whether a person or family is in poverty are not the official thresholds, but thresholds based on a three-parameter equivalence scale described in Section V.

 

Table 3 shows the effects of government taxes and transfers on estimates of poverty by using four definitions of resources: money income, market income, post-social insurance income, and disposable income. All four measures use the same poverty thresholds. See Section II for the income definitions.

 

In 2004, using the money income definition and the three-parameter thresholds resulted in a poverty estimate of 12.6 percent for the United States. The income definition used in this measure was identical to that used in the official poverty measure (which was 12.7 percent in 2004). Thus, using the three-parameter equivalence scale lowered the estimate of the poverty rate by 0.1 percentage points overall.

 

Market income represents resources available to people and families based on market activities, including realized capital gains or losses resulting from investments. Because market income does not include income from government sources nor does it deduct taxes, poverty rates under this measure can serve as a reference point for evaluating the effects of those government policies. The estimate of the overall poverty rate was 19.4 percent under this definition in 2004.

 

Post-social insurance income is defined as market income with non-means-tested cash transfers, such as Social Security, added back in. The definition is referred to as “social insurance” because the programs included here were designed to target everyone, not only people with low income. Adding those transfers back in reduced the poverty estimate to 12.9 percent.

 

Disposable income, in this data release, is a representation of the net income people have available for living expenses. Like market income, disposable income includes all resources based on market activities, but also includes the value of government transfers. Disposable income excludes taxes since they are mandated by law and the money used to pay those taxes is not available for other purposes. The net effect of these additions and subtractions is a poverty rate of 10.4 percent—which is lower than the poverty rates estimated using the other income definitions. 

 

By comparing poverty rate estimates across definitions, a pattern emerges. The estimated percentage of people below the poverty thresholds is higher when using market income than when using money income. That higher percentage reflects the impact of taking away government transfers and some work-related expenses. Since the market income definition includes returns on home equity and net capital gains (and the money income measure does not), the higher poverty estimate for market income indicates that those two income sources taken together are not as large as government transfers and work expenses combined. In turn, since non-means-tested transfers are added in the post-social insurance income definition, poverty estimates under that definition are lower than when using market income. Under the disposable income definition, the value of noncash benefits and means-tested transfers are added, and taxes are subtracted, resulting in an even lower percentage than under post-social insurance income. The lower estimate under the disposable income definition results not only from the noncash benefits and government transfers, but also from the Earned Income Tax Credit, which, though classified under “taxes,” does not reduce income but instead functions as a means-tested transfer for people with low income.

 

Child poverty estimates across income definitions illustrate the same pattern of the income components’ effects. Under the money income definition, 17.4 percent of children under 18 were below the poverty line in 2004. Under the market income definition, the estimate was 20.2 percent. Adding non-means-tested cash benefits lowered the child poverty estimate to 18.1 percent. Under a disposable income concept that includes the effects of noncash benefits, cash transfers, and taxes, the child poverty estimate in 2004 was 13.1 percent.   

 

None of the alternative income definitions exhibited a change in estimates of overall poverty between 2003 and 2004, unlike the official poverty measure, which rose from 12.5 percent in 2003 to 12.7 percent in 2004. 

 

For many demographic and geographic groups, all four alternative definitions told the same story (see Table 3). All four definitions identified no change in poverty rates between 2003 and 2004 for people in families, Blacks, children under 18 years, 55-to-59-year-olds, the foreign-born, people living in the South and the West, people who did not work at least one week, married-couple families, and female-householder families. All four definitions measured poverty rate increases between 2003 and 2004 for non-Hispanic Whites, 18-to-24-year-olds, 45-to-54-year-olds, and people living in the Midwest. All four definitions found lower poverty rates for Asian families in 2004 than in 2003.

 

For some demographic and geographic groups, year-to-year changes were not uniformly identified by all four definitions. For instance, the percentage below poverty for unrelated individuals (people living alone or with non-relatives only) increased between 2003 and 2004 under the disposable income measure but not any of the others. Conversely, the percentage below poverty for Asians (measured for people as opposed to families) fell under all definitions except for disposable income. 

 

The percentages shown in Table 4 are based on poverty thresholds adjusted with the CPI-U-RS. The CPI-U and the CPI-U-RS measure inflation differently; thus, the choice of price index affects the thresholds—the dollar values used to determine poverty status (see section V for details on poverty thresholds and on price indexes). The percentages are lower than those shown in Table 3 because the thresholds are lower. Under the money income definition, 10.6 percent of people were below the CPI-U-RS-adjusted poverty thresholds in 2004. Under the most comprehensive definition of income shown in this release (disposable income), the corresponding percentage in 2004 was 8.3 percent. Despite the lower percentages, none of the four definitions identified any change in the poverty estimate for the overall population between 2003 and 2004 for the CPI-U-RS-based estimates—a finding consistent with the CPI-U-based estimates. 

 

 

V. Methodology and Alternative Income Components

 

Unit of Analysis: The unit of analysis for the income measures in this report is the household. The units of analysis for the poverty measures are families and people. That is, the poverty status of people is based on their family’s income if they live in a family and their own individual income otherwise. For groups of two or more people living together who are related by birth, marriage, or adoption, poverty thresholds are defined for family units. For those living alone or with non-relatives, poverty thresholds are defined for individuals. Thus, two unrelated people living together are considered two separate “units” for poverty determination purposes, meaning that their combined poverty thresholds would be the same whether they lived in the same housing unit or two different ones. While it is clear that some allowance should probably be made to reflect the fact that it is less expensive for unrelated people to share living quarters, there is little research that points to exactly what type of adjustment is appropriate. But it is clear that the effect of such an adjustment is not trivial. In Weinberg 2005, there are tabulations based on a “household” definition of poverty. Under this definition, two unrelated people living together would be considered as exactly the same as a two-person family for poverty determination purposes. These tabulations showed that the effect of moving to a household-based poverty definition would have reduced the overall poverty rate by approximately 1.5 percentage points in 2002.

 

Work-Related Expenses: Previous Census Bureau alternative poverty reports that included the effect of work-related expenses on income have included both the effect of child care expenses and “other” (non-child care) work-related expenses (see Short 2001 and Dalaker 2005). This report only includes the effect of non-child care work expenses. The reason for this change is that the Census Bureau is considering making changes to its child care expenses imputation procedures and is deferring their inclusion until either the current method can be validated or an improved method can be found.

 

Government Cash Transfers: Government transfers include payments from the following sources: 1) Unemployment Compensation, 2) State Workers’ Compensation, 3) Social Security, 4) Supplemental Security Income (SSI), 5) Public Assistance, including Temporary Assistance for Needy Families (TANF), 6) Veterans’ Payments, 7) government survivor, disability, and pension payments, and 8) government educational assistance.

 

Government Means-Tested Cash Transfers: The means-tested portion of transfers includes payments from the following sources: 1) Public Assistance, including Temporary Assistance for Needy Families (TANF), 2) SSI, and 3) means-tested Veterans’ Payments.

 

Government Non-Means-Tested Cash Transfers: The non-means-tested portion of transfers includes payments from the following sources: 1) Unemployment Compensation, 2) State Workers’ Compensation, 3) Social Security, 4) non-means-tested Veterans’ Payments, 5) government survivor, disability, and pension payments, and 6) government educational assistance.

 

Government Noncash Transfers (also called noncash benefits): Non-cash transfers include those government benefits that are distributed as services or vouchers, and for which the recipient does not get cash. These include 1) food stamps, 2) housing subsidies, and 3) free or reduced-price school lunches.

 

Inequality Measures:  Two widely used measures of income inequality are shares of aggregate income and the Gini Index. Shares of aggregate income are computed by ranking households from lowest to highest income and then dividing them into groups of equal numbers of households, typically quintiles (20-percent groups). The aggregate income of each group divided by the overall aggregate income is each group’s share. In a perfectly equal society, the cumulative share of income should equal the cumulative share of households. The Gini Index summarizes the dispersion of income shares. The Gini Index ranges from 0, which indicates perfect equality, to 1, which denotes perfect inequality (all income is received by one household).

 

Poverty Thresholds:  The poverty thresholds used in this release differ from the official poverty thresholds described in Income, Poverty, and Health Insurance Coverage in the United States: 2004. Official poverty thresholds are based on the work of Mollie Orshansky, and they vary by family size and number of children (and age for unrelated individuals and 2-person families). Because the official thresholds were based on food costs and spending patterns, rather than an overall assessment of how needs vary by family size, the relationship between thresholds for different families has been criticized as non-systematic and ad hoc, and has been cited as one of the weaknesses of the current official poverty definition (see Citro and Michael 1995). The thresholds used for this release start with the official 4-person, 2-child threshold and compute the thresholds for other families through the use of a more systematic three-parameter equivalence scale. The first scale parameter reflects that children, on average, consume less than adults; the second parameter reflects that as family size increases, expenses do not increase at the same rate; and the third parameter allows the first child in a single-adult family to represent a greater increase in expenses than the first child in a two-adult family. For details on the derivation of this equivalence scale, see Appendix A of Short 2001.

 

Consumer Price Indexes: Official poverty thresholds are updated each year based on the CPI-U (Consumer Price Index for All Urban Consumers), which is computed by the Bureau of Labor Statistics (BLS). The BLS also produces another index, the CPI-U-RS (Consumer Price Index Research Series Using Current Methods). The CPI-U-RS applies most of the methodological improvements made to the CPI-U since the beginning of this series in 1978. The poverty thresholds used in this release were adjusted using both methods. First, the 4-person, 2-child threshold (the starting point for the three-parameter equivalence scale) was updated for inflation using the CPI-U. Table 3 presents these estimates. Second, the set of thresholds used to create the data in Table 3 was adjusted using the CPI-U-RS. Table 4 displays the estimates based on the thresholds updated with the CPI-U-RS. For more information on the Census Bureau’s use of the CPI-U-RS to adjust poverty thresholds, see Dalaker 2005. Household income is adjusted over time using the CPI-U-RS. For more information on the CPI-U-RS, see http://www.bls.gov/cpi/cpirsdc.htm.

 

Income Underreporting in the CPS: The collection vehicle for the estimates shown in this release is the Annual Social and Economic Supplement (ASEC) to the Current Population Survey. The fact that respondents don’t report their incomes with perfect accuracy on the CPS ASEC is well documented (see Roemer 2000, for example). A recent study by analysts at the Census Bureau and the Bureau of Economic Analysis (BEA) compared BEA State Personal Income (SPI) aggregates with those from the CPS for income year 2001 (Ruser, Pilot, and Nelson 2004). They found that once the necessary adjustments were made to make the two datasets conceptually the same, the CPS ASEC aggregate was about $806 billion less than the SPI aggregate—a gap of around 11 percent. About one-half of this gap is due to adjustments BEA makes to its SPI for unreported earnings (wages and salaries and self-employment income). The study also found that the gaps are not consistent by type of income. For example, the wage and salary gap was around 3 percent while the gap for transfer incomes was around 23 percent. Clearly there needs to be more research on the effect of underreporting of key income types on important summary measures such as the poverty rate and median household income. In Weinberg 2005, there were tabulations based on files created by the Urban Institute with support from the Department of Health and Human Services Office of the Assistant Secretary for Planning and Evaluation. These files include underreporting adjustment models for three transfer programs: Temporary Assistance for Needy Families (TANF), Supplemental Security Income (SSI), and Food Stamps. Tabulations from this file illustrate the potential importance of underreporting adjustments. They showed that the effect of using the file that incorporated imputations for unreported TANF, SSI, and Food Stamp benefits was to reduce the overall poverty rate by around 1 percentage point in 2002.

 

 

VI. References

 

Citro, Constance F., and Robert T. Michael (eds.), Measuring Poverty: A New Approach, Washington, DC:  National Academy Press, 1995.

                                   

Cleveland, Robert W., U.S. Census Bureau, Current Population Reports, P60-228, Alternative Income Estimates in the United States: 2003, U.S. Government Printing Office, Washington, DC, 2005.

 

Dalaker, Joe, U.S. Census Bureau, Current Population Reports, P60-227, Alternative Poverty Estimates in the United States: 2003, U.S. Government Printing Office, Washington, DC, 2005.

 

DeNavas-Walt, Carmen, Bernadette D. Proctor, and Cheryl Hill Lee, U.S. Census Bureau, Current Population Reports, P60-229, Income, Poverty, and Health Insurance Coverage in the United States: 2004, U.S. Government Printing Office, Washington, DC, 2005.

 

O’Hara, Amy, “New Methods for Simulating CPS Taxes,” U.S. Census Bureau Technical Paper, 2004.

 

Roemer, Marc, “Assessing the Quality of the March Current Population Survey and the Survey of Income and Program Participation Income Estimates, 1990-1996,” Mimeo, U.S. Bureau of the Census, 2000.

 

Ruser, John and Adrienne Pilot and Charles Nelson, “Alternative Measures of Household Income: BEA Personal Income, CPS Money Income, and Beyond,” presented at the Federal Economic Statistics Advisory Committee, 2004.

 

Short, Kathleen, U.S. Census Bureau, Current Population Reports, P60-216, Experimental Poverty Measures: 1999, U.S. Government Printing Office, Washington, DC, 2001.

 

U.S. Census Bureau, Current Population Reports, P60-186RD, Measuring the Effect of Benefits and Taxes on Income and Poverty 1992, U.S. Government Printing Office, Washington, DC, 1993.

 

Weinberg, Daniel W., “Alternative Measures of Income and Poverty and the Anti-Poverty Effects of Taxes and Transfers,” CES-5-08 (2005) U.S. Census Bureau.