Income, Poverty and Health Insurance Coverage
2006 American Community Survey
David S. Johnson, Ph.D.
Chief, Housing and Household Economic Statistics
August 28, 2007
Good morning and thank all of you for being here today. The Census Bureau today presents its annual compilation of data on the nation’s income, poverty and health insurance coverage, including the release of data from the American Community Survey.
I would particularly like to thank the many Census staff who are involved in collecting, processing, compiling, preparing and disseminating the data for these important statistics – many of whom are here today. It is only through these dedicated staff that the Census Bureau can produce timely, accurate and relevant information about the nation’s economy.
I would also like to thank the millions of you – the respondents – who have talked to us on the phone, invited us into your home, or sent us your completed questionnaire. It is only because of you that we have these results today.
Before we begin, I also want to take a moment to remember someone who passed last year who left an indelible mark on the work we’re presenting here today. Mollie Orshansky died last December at the age of 91. In 1963, while serving at the Social Security Administration, she developed the official measurement of poverty in the United States, the Orshansky Poverty Thresholds. She was a true pioneer to whom we owe a debt of gratitude. All of us working in statistical agencies should be encouraged by the work of Ms. Orshansky – with hard work and diligence she was able to change the way the country measured poverty and her methods have stayed with us for over 40 years.
The figures we are presenting today are the income, poverty, and health insurance coverage data from the Annual Social and Economic Supplement of the Current Population Survey (or CPS), and the income, earnings, and poverty data from the American Community Survey (or ACS).
As in past years, the CPS is the survey used for national income and the official poverty estimates; however, if you are interested in state estimates in comparison to the national estimates we urge you to use the ACS estimates. If you are interested in the relationship between local areas and states, such as the relationship between Fairfax and Loudoun Counties and Virginia, you should also use the ACS. Under no circumstances, should you mix and match one survey to the other.
Let me first summarize the national findings about income, poverty and health insurance coverage from the Current Population Survey. The CPS is the nation’s longest running annual government survey and the official source of monthly unemployment statistics and annual poverty statistics. The CPS has been collecting income data since 1948 and health insurance data since 1980.
Here are some highlights.
In 2006, median household money income in the nation rose to $48,200; an increase of 0.7 percent from 2005 in real terms – the second consecutive year of increase.
At 12.3 percent in 2006, the official poverty rate for the nation fell from the rate in 2005. The total of 36.5 million people living below the official poverty thresholds was not statistically different from 2005.
The number of people without health insurance coverage increased from 44.8 million in 2005 to 47.0 million in 2006. The uninsured rate (the percentage of people without health insurance coverage) also increased to 15.8 percent in 2006.
Turning to a time series chart on real – inflation-adjusted -- median household income, we can see that median household income rose in 2006.
For the second consecutive year, households in the United States experienced an increase in real annual median income. Even though overall household income has not yet recovered to its 1999 pre-recessionary peak of $49,200, the gap is narrowing. In 2004, real median household income was 3.9 percent less than its 1999 level; in 2006, it was only 2.1 percent less. Compared with 1967, the first year for which household income statistics are available, real median household income has increased 30.8 percent.
[This increase in real median income has also been accompanied by a decrease in average household size, from 3.3 people per household in 1967 to 2.6 in 2006, which could imply an even larger increase in economic well-being for households. While the statistics on household income are not adjusted for differences in household size, the poverty statistics are adjusted for differences in family size.]
The time series for the poverty rate reflects the time series for median income. These time series charts show that periods of rising median income are usually accompanied with periods of falling poverty rates, and vice versa. In fact, during the past 39 years, the median household income and the poverty rate have never experienced a significant change in the same direction.
From the most recent trough in 2000, the poverty rate rose for 4 consecutive years, from 11.3 percent in 2000 to 12.7 percent in 2004, and has declined to 12.3 percent in 2006, a figure that is 45 percent lower than the rate in 1959 – 22.4 percent – the first year for which poverty estimates are available.
Focusing on median income by race and Hispanic origin, White households (including Hispanics) were the only group to experience an increase in real household income in 2006, rising by 1.1 percent—the first real change in median income for this group since 1999. Whites also experienced a decrease in poverty – from 10.6 to 10.3 percent. This is mainly due to the fall in the poverty rate for Hispanics; their poverty rate fell from 21.8 percent in 2005 to 20.6 percent in 2006. Black households had the lowest median income in 2006 among the race and Hispanic-origin groups at $32,000.
The poverty rates for children, adults and older adults show different trends over the past 47 years – yet similar trends to the overall poverty rate in recent years. The poverty rate for people aged 18 to 64 was not statistically different from 2005 at 10.8 percent in 2006. In contrast, the poverty rate for people 65 and older in poverty decreased to 9.4 percent in 2006 from 10.1 percent in 2005 – which is among their lowest poverty rates since 1959 (it was 9.7 percent in 1999, 9.8 percent in 2000 and 2004).
In 2006, children under 18 showed no statistical change in their poverty rate or the number in poverty (17.4 percent and 12.8 million children).
[While children represented 24.9 percent of the total population, they represented 35.2 percent of the people in poverty.]
Another demographic group discussed in the report is the native and foreign-born populations. The poverty rate and the number in poverty for natives were not statistically different from 2005 at 11.9 percent and 30.8 million in 2006. The foreign-born poverty rate decreased from 16.5 percent in 2005 to 15.2 percent in 2006, while their number in poverty remained statistically unchanged at 5.7 million in 2006.
While median household income in 2006 rose by 0.7 percent, the real median earnings of both men and women who worked full-time, year-round declined between 2005 and 2006. The median earnings of men declined 1.1 percent to $42,300. The median earnings of women declined 1.2 percent to $32,500. (The apparent difference between the declines in the earnings of men and women was not statistically significant.) This is the third consecutive year that men and women experienced a decline in earnings. And the ratio of women’s to men’s earnings was not different from the 2005 ratio at 77 percent.
The different trends for household income and individual earnings are partially the result of increasing numbers of year-round, full-time workers, which has resulted in an increase in the number of year-round, full-time workers per household. This trend can have a positive impact on household income, even if real earnings for individual year-round, full-time workers is declining. This is apparent by the increase in median household earnings between 2005 and 2006, which increased 2.2%.
So far we have presented statistics about the median household income for various demographic groups. This chart depicts an entire distribution of household income, showing the percentage of households at each income level. The median represents one point on the distribution of household income --
– the point at which half of the households have income below it and half above it.
Other points along this distribution, however, provide additional information about the nation’s household income.
For example, the 20th percentile, which was $20,000 in 2006, denotes the lowest quintile and the income at which 20 percent of the households have incomes below this income level.
The highest quintile – the 80th percentile, which was $97,000 -- is at the other end of the distribution. Notice that the distribution of income in the highest quintile is extremely disperse (95th percentile at $174,000). While the median represents the dividing points for households in terms of income, for the past few years, the 80th percentile was essentially the dividing point for total household income. That is, the income for households in the highest quintile accounts for about half of the nation’s income, while income for those households in the bottom quintile accounts for only 3.4 percent.
Using this information, we can produce a Gini index -- the most widely used measure of inequality. The Gini index indicates higher inequality as the index is closer to one, and this figure shows an increase of inequality from 1967.
The Gini index in 2006 is not significantly different than that in 2005. During the last 10 years there was never a statistically significant annual change in the Gini; however, the Gini index has increased 3.3 percent over the past ten years (from 0.455 to 0.470), and it has increased 1.7 percent since 2002. (The apparent difference between the increases in the Gini index was not statistically significant.)
These income distributions, along with poverty thresholds, are similarly used to determine the poverty rates. The Census Bureau, using the CPS, determines the distribution of money income, and constructs families of related people, singles and unrelated individuals.
By placing people in related families this means that there are many people who live in households, but are not related to anyone in the household, and hence, their income is treated separately and their poverty status determined independently from the other people in the household.
Then we compare the family and individual income to the poverty thresholds defined by the Office of Management and Budget (for example, the 2006 threshold for a 4-person family with 2 adults and 2 children was $20,444, and about half of that for an unrelated individual under 65 years old – $10,488).
This produces the percentage of people living in families (and unrelated individuals) with incomes less than these thresholds - the poverty rate.
Thanks to Ms. Orshansky, this is basically the same method that she developed in 1960s -- providing a consistent time-series measure of poverty.
This figure shows that the poverty rate for people in families is about half the rate for unrelated individuals (10.6 vs 20.0 percent). In addition, while the poverty rate of people in families in 2006 is not statistically different than the rate in 2005, this rate is lower than the rate in 2004, when poverty for people in families was 11.0%.
[In addition, the poverty rate for related children (children living with a relative) is lower than the poverty rate for all children (16.9 vs 17.4 percent). This is partially due to the fact that there are about 172,000 children between the ages of 15-17 that are not living with relatives and hence, are treated as unrelated individuals – and their poverty rate is higher.]
Using family and individual income and the poverty thresholds produces a family size adjusted distribution of income, which provides a broader measure of the economic well-being of families and individuals. While the household income distribution treats single person households with $30,000 similarly to a couple with $30,000, this adjustment of income by family size (or equivalence adjusting) takes into consideration the number of people living in the household and how those people share resources and take advantage of economies of scale.
This figure illustrates the distribution of people according to their income-to-poverty ratios.
The curve graphically depicts the proportion of people with given income-to-poverty ratios. Hence, this chart depicts the percentage of people in poverty as the area under the curve to the left of the vertical line at 1.0, represents 12.3 percent of people in 2006.
[Another way to measure income inequality is using equivalence-adjusted income. A new feature of this year’s report is the Gini coefficient for equivalence-adjusted income (which is lower than the Gini for household income).]
This distribution of income to poverty thresholds illustrates how changing the poverty threshold (and alternatively examining the percentage of people living below various percentages of income) affects a poverty rate.
For example, in 2006, 5.2 percent of people had an income below one-half their poverty threshold.
In 2006, the percentage of people with an income-to-poverty threshold ratio of 1.25 – or with income below 125 percent of their thresholds - was 16.8 percent – not statistically different than the rate in 2005.
Recall that these statistics use the measure of money income, which is basically regular money income from a variety of sources including employment, government, pensions and interest.
The Census Bureau recognizes that the measure of money income may not completely capture the economic well-being of individuals and families. Families also derive economic well-being from noncash benefits, such as food stamps and housing subsidies, and have reductions in disposable income due to taxes. The Census Bureau computes a number of other measures of income and poverty that attempt to account for those factors. Last March, we released these measures, one of which was a measure of disposable income (which included various in-kind benefits, subtracted taxes, and included the rental value of homeownership).
This chart - from our March report - illustrates that in 2005, the distribution of disposable income is less disperse (more people in the middle) than money income, and hence the Gini coefficient is lower.
The estimates for alternative income measures for 2006 will be released later this year.
Today, on our website, we are also releasing a preliminary version of our new poverty measurement table generator, with which users can construct their own alternative income measures and compare these to alternative poverty thresholds – using data from 2004.
Next, I will provide statistics about health insurance coverage in the US from the CPS. As I mentioned in the beginning, the uninsured rate increased to 15.8 percent in 2006 – the second consecutive year of increase.
The percentage of people covered by private insurance (either through an employer or privately purchased) declined from 68.5 percent to 67.9 percent in 2006, while the percentage of persons covered by government health programs declined from 27.3 percent to 27.0 percent. The decrease in the private insurance rate was driven by the decline in employer-provided coverage that dropped from 60.2 percent to 59.7 percent between 2005 and 2006.
In looking at the uninsured rates for children, the percentage of children without health insurance increased between 2005 and 2006 from 10.9 to 11.7 percent.
The percentage of children covered by private coverage (employment-based and direct purchased) decreased from 65.8 percent to 64.6 percent in 2006. The increase in the overall uninsured rate for children can be attributed to this decline. [With an uninsured rate of 19.3 percent, children in poverty were more likely to be uninsured than all children.]
Finally, we can use the CPS to calculate 3-year average estimates of the uninsured rate by state. This figure shows that the uninsured rate ranges from 8.5 percent (Minnesota) to 24.1 percent (Texas). Most of the states with uninsured rates higher than the national rate are in the South and West. We have also computed health insurance rates at the county level using a model-based approach, and next year we will begin collecting health insurance data on the ACS.
I now turn to the income, earnings and poverty statistics from the American Community Survey.
Which we release today.
And on September 12 we will release social, economic, and housing data from the 2006 ACS, as well as the Public Use Microdata File. Finally, on September 27th we will release Group Quarter Profiles and other Selected Population Profiles.
This year, as in 2006, the Census Bureau is collecting information in every county in the nation. Data collected in 2006 allow us to provide information for about 7000 places, including all States, congressional districts, and counties with populations over 65,000. There are over 3,100 counties in the US, and today we are releasing information for about 780 of these counties – those with populations over 65,000 (representing 83% of the total population).
2006 was the first year in which the ACS sample included the group quarter population. Both institutional (such as prisons) and non-institutional (such as college dorms) group quarters are included. Thus, from 2006 and forward, the ACS provides a more representative picture of the entire resident population of the United States.
Since group quarters were not included in the ACS prior to 2006, one must use caution in making comparisons between 2005 and 2006. This will affect some subject areas more than others, and we are providing guidance to users on this on our website.
As I stated earlier, this year the Census Bureau is focusing on the annual state estimates and sub-state estimates for income and poverty from the ACS, while focusing on the national income and poverty rates, and the changes over time, from the CPS. As with the CPS results, I’ll begin with median household income.
In the 2006 ACS, median household income in the past 12 months in 9 states were above $55,000 (and above the national median), while in 7 states the median incomes were below $40,000.
The states in the northeast tended to have median incomes above the US median, while states in the South tended to have median incomes below the US median.
Maryland was one of the states with the highest median incomes of $65,100, which was almost twice the median income of one of the lowest states, Mississippi, with a median income of $34,500.
The advantage of the ACS is that one can examine income below the state level – for all counties with population over 65,000. While this represents only 25% of all counties, the ACS is collecting data in all 3100 counties. We can only publish estimates for counties with sufficient sample, however. In 2010, we will have sufficient sample sizes to produce estimates for all counties, and census tracts.
Looking at the median income by county can help understand the results by state. Similar to the state results, many counties in the NE have household incomes higher than the national median. Even though states in the south have lower median incomes, there are still many counties with higher incomes – such as Collin County (outside of Dallas) in Texas, with a median household income of $74,100.
This chart illustrates that there is dispersion of income across the country and even within states.
With the ACS, we can examine the dispersion of incomes within many states. Maryland, with its 24 counties or county equivalents, has 16 counties of 65,000 or more. They range in median income from Howard county with $94,300 to Allegany county with $33,000 (which is not significantly different from the median income for Baltimore city, a county equivalent ($36,000)), almost a 3 to 1 ratio – compared to the almost 2 to 1 ratio for the state medians, demonstrating that the dispersion of incomes within states can be larger than the dispersion across states.
This year, the report produces Gini indexes for all states, and the tables present the Gini for all places with over 65000 in population. The national Gini index was 0.464, and it varied from state to state, ranging from .537 for the District of Columbia to .410 for Utah (which is not significantly different from the Gini indexes for Wyoming (.413), New Hampshire (.417), Arkansas (.417), and Vermont (.420)).
Turning to the state poverty rates, similar to median income, states in the northeast tended to have poverty rates below the US poverty rate.
As with median income, the state poverty rates varied. Maryland had one of the lowest poverty rates, at 7.8 percent, while Mississippi had one of the highest rates, at 21.1 percent – the state that also had one of the lowest median incomes. (Maryland also had one of the highest median incomes).
New York was the only state in the northeast with a poverty rate above the US poverty rate. In fact, New York also had a median income higher than the US median. And as a result, as we saw from the last slide, NY had one of the highest inequality measures.
We can use this information to again drill-down to the dispersion of poverty rates within a state. In New Jersey all of its 21 counties have populations over 65,000. Their poverty rates range from 3.5 percent in Hunterdon County (which is not significantly different from the poverty rates for Morris County (3.9 percent), Somerset County (4.4 percent), and Sussex County (4,8 percent)) to 15.3 percent in Cumberland County (which is not significantly different from the poverty rates for Hudson County (15.2 percent), Passaic County (15.0 percent), and Essex County (14.5 percent)). Camden City – a small place within Camden County, had one of the highest poverty rates in the country at 35.6 percent.
As with the CPS data, we can evaluate the median earnings of men and women. With the large sample sizes in the ACS, however, we can also produce estimates of median earnings for men and women by occupation group, and by state. (We focus on 22 occupation groups.)
Here we show the five occupation groups for men and women with the highest median earnings. Four out of five of the groups are the same for men and women; however the median earnings and the ranges are different for men and women. For example, the median earnings for women in legal occupations is only 49.3 percent of men’s earnings, while for computer and mathematical occupations the percentage is 86.7.
There is more parity between women's and men's earnings among the occupation sub groups, within the legal occupations category. For example, among lawyers (within legal occupations), women's earnings were 76 percent of men's earnings. In fact, all four legal occupations (lawyers, judges and judicial workers, paralegals, and other support workers) have a higher ratio than that for the larger legal category. Hence, much of the divergence within these large groups can be due to the composition of men and women in these occupation sub-groups.
We have shown you just some of the data available from the Annual Social and Economic Supplement of the CPS and the American Community Survey. Much more data for the national poverty rates are available in the reports and on the Census website, at www.census.gov. and the links provided there.
For the ACS, data on income, earnings and poverty are also available on the website for most metropolitan areas, every Congressional District, and every place with a population of 65,000 or more.
That concludes my presentation. Thank you. I turn it back to Jay who will coordinate questions.