To donate by check, phone, or other method, see our More Ways to Give page.

×

Fighting Poverty: The First Step Is Counting It

The Center for American Progress issued a useful report last week on ways to reduce and eventually eliminate poverty.

The report proposed a national goal of cutting the number of people in poverty by half within a decade, and eliminating poverty altogether within a generation. While these are worthy goals, and the report describes many carefully designed policies for achieving them, there is a more basic problem that must also be considered in confronting poverty: We often don't count poor people.

The point here is simple. The Labor Department's Current Population Survey (CPS), the generally accepted basis for poverty data, doesn't reach many of the people who fall below the poverty line. In fact, it is likely that factors associated with poverty, such as unstable living situations and an alienation from government, are causing the CPS to exclude many poor people from the survey.

In the 1980s, the CPS covered approximately 93 percent of the US population. This effectively means that the CPS surveyors were able to contact people and receive answers in 93 percent of the homes they targeted. This coverage rate edged down in the 1990s, and has fallen to just 89 percent during the current decade. This means that they miss about 11 percent of the homes they target.

While missing 11 percent of the population should by itself be cause for concern, non-coverage is concentrated most heavily in the demographic groups most likely to be in poverty. In 2005 (the most recent year for which data is available), the coverage rate for black women was 83.2 percent, while the coverage rate for black men was 76.1 percent. For Hispanic women and men the rates were 91.3 percent and 83.5 percent, respectively. For young black men, the coverage rate was less than 70 percent.

The Labor Department adjusts its data for this undercounting, but it does so by assuming that the people who it does not talk to are just like the people who it does talk to. In other words, it assumes that when it comes to employment, wages and income, the 16.8 percent of black women who are not covered by the CPS are just like the 83.2 percent who are covered. Similarly, it assumes that the 30 percent of young black men who aren't counted by the survey are just like the 70 percent who are counted.

This is a very strong assumption and one that we have good reason to believe is wrong, based on a comparison with Census data. John Schmitt (my colleague at the Center for Economic and Policy Research) and I compared data on employment from the CPS with data from the 2000 Census. Because the Census is done once every 10 years (as opposed to once a month) and has a far larger budget, it is better able to devote the resources necessary to contact most of the population - including extensive outreach to economically marginalized neighborhoods. As a result, the Census reaches close to 99 percent of the total US population. It therefore provides a good basis for assessing the effects of the uncounted population on the CPS.

After making adjustments for differences in the survey questions, we found that the CPS overstated employment rates for the population as a whole (the percentage of the population that is employed) by 1.3 percentage points. It overstated employment rates for black men by 3 percentage points, and it overstated employment rates for young black men by more than 8 percentage points. These overstatements are substantial. Since people who are not employed are also far more likely to be in poverty and lack health insurance coverage, it is likely that the CPS is understating the extent of poverty and the number of people who lack health insurance coverage.

It would be very expensive for the Labor Department to change its survey methods so it could get coverage rates as high as the Census. However, there are technical adjustments that it can make to its data to ensure that it better captures the extent of poverty in the country. This seems like a really small step, but if we're going to get serious about tackling poverty, we first have to get serious about counting it.

Dean Baker is the co-director of the Center for Economic and Policy Research (CEPR). He is the author of The Conservative Nanny State: How the Wealthy Use the Government to Stay Rich and Get Richer ( www.conservativenannystate.org). He also has a blog, "Beat the Press," where he discusses the media's coverage of economic issues. You can find it at the American Prospect's web site.

Our work is licensed under Creative Commons (CC BY-NC-ND 3.0). Feel free to republish and share widely.
center for american progresscenter for economic and policy researchdean baker

The Center for American Progress issued a useful report last week on ways to reduce and eventually eliminate poverty.

The report proposed a national goal of cutting the number of people in poverty by half within a decade, and eliminating poverty altogether within a generation. While these are worthy goals, and the report describes many carefully designed policies for achieving them, there is a more basic problem that must also be considered in confronting poverty: We often don't count poor people.

The point here is simple. The Labor Department's Current Population Survey (CPS), the generally accepted basis for poverty data, doesn't reach many of the people who fall below the poverty line. In fact, it is likely that factors associated with poverty, such as unstable living situations and an alienation from government, are causing the CPS to exclude many poor people from the survey.

In the 1980s, the CPS covered approximately 93 percent of the US population. This effectively means that the CPS surveyors were able to contact people and receive answers in 93 percent of the homes they targeted. This coverage rate edged down in the 1990s, and has fallen to just 89 percent during the current decade. This means that they miss about 11 percent of the homes they target.

While missing 11 percent of the population should by itself be cause for concern, non-coverage is concentrated most heavily in the demographic groups most likely to be in poverty. In 2005 (the most recent year for which data is available), the coverage rate for black women was 83.2 percent, while the coverage rate for black men was 76.1 percent. For Hispanic women and men the rates were 91.3 percent and 83.5 percent, respectively. For young black men, the coverage rate was less than 70 percent.

The Labor Department adjusts its data for this undercounting, but it does so by assuming that the people who it does not talk to are just like the people who it does talk to. In other words, it assumes that when it comes to employment, wages and income, the 16.8 percent of black women who are not covered by the CPS are just like the 83.2 percent who are covered. Similarly, it assumes that the 30 percent of young black men who aren't counted by the survey are just like the 70 percent who are counted.

This is a very strong assumption and one that we have good reason to believe is wrong, based on a comparison with Census data. John Schmitt (my colleague at the Center for Economic and Policy Research) and I compared data on employment from the CPS with data from the 2000 Census. Because the Census is done once every 10 years (as opposed to once a month) and has a far larger budget, it is better able to devote the resources necessary to contact most of the population - including extensive outreach to economically marginalized neighborhoods. As a result, the Census reaches close to 99 percent of the total US population. It therefore provides a good basis for assessing the effects of the uncounted population on the CPS.

After making adjustments for differences in the survey questions, we found that the CPS overstated employment rates for the population as a whole (the percentage of the population that is employed) by 1.3 percentage points. It overstated employment rates for black men by 3 percentage points, and it overstated employment rates for young black men by more than 8 percentage points. These overstatements are substantial. Since people who are not employed are also far more likely to be in poverty and lack health insurance coverage, it is likely that the CPS is understating the extent of poverty and the number of people who lack health insurance coverage.

It would be very expensive for the Labor Department to change its survey methods so it could get coverage rates as high as the Census. However, there are technical adjustments that it can make to its data to ensure that it better captures the extent of poverty in the country. This seems like a really small step, but if we're going to get serious about tackling poverty, we first have to get serious about counting it.

Dean Baker is the co-director of the Center for Economic and Policy Research (CEPR). He is the author of The Conservative Nanny State: How the Wealthy Use the Government to Stay Rich and Get Richer ( www.conservativenannystate.org). He also has a blog, "Beat the Press," where he discusses the media's coverage of economic issues. You can find it at the American Prospect's web site.

The Center for American Progress issued a useful report last week on ways to reduce and eventually eliminate poverty.

The report proposed a national goal of cutting the number of people in poverty by half within a decade, and eliminating poverty altogether within a generation. While these are worthy goals, and the report describes many carefully designed policies for achieving them, there is a more basic problem that must also be considered in confronting poverty: We often don't count poor people.

The point here is simple. The Labor Department's Current Population Survey (CPS), the generally accepted basis for poverty data, doesn't reach many of the people who fall below the poverty line. In fact, it is likely that factors associated with poverty, such as unstable living situations and an alienation from government, are causing the CPS to exclude many poor people from the survey.

In the 1980s, the CPS covered approximately 93 percent of the US population. This effectively means that the CPS surveyors were able to contact people and receive answers in 93 percent of the homes they targeted. This coverage rate edged down in the 1990s, and has fallen to just 89 percent during the current decade. This means that they miss about 11 percent of the homes they target.

While missing 11 percent of the population should by itself be cause for concern, non-coverage is concentrated most heavily in the demographic groups most likely to be in poverty. In 2005 (the most recent year for which data is available), the coverage rate for black women was 83.2 percent, while the coverage rate for black men was 76.1 percent. For Hispanic women and men the rates were 91.3 percent and 83.5 percent, respectively. For young black men, the coverage rate was less than 70 percent.

The Labor Department adjusts its data for this undercounting, but it does so by assuming that the people who it does not talk to are just like the people who it does talk to. In other words, it assumes that when it comes to employment, wages and income, the 16.8 percent of black women who are not covered by the CPS are just like the 83.2 percent who are covered. Similarly, it assumes that the 30 percent of young black men who aren't counted by the survey are just like the 70 percent who are counted.

This is a very strong assumption and one that we have good reason to believe is wrong, based on a comparison with Census data. John Schmitt (my colleague at the Center for Economic and Policy Research) and I compared data on employment from the CPS with data from the 2000 Census. Because the Census is done once every 10 years (as opposed to once a month) and has a far larger budget, it is better able to devote the resources necessary to contact most of the population - including extensive outreach to economically marginalized neighborhoods. As a result, the Census reaches close to 99 percent of the total US population. It therefore provides a good basis for assessing the effects of the uncounted population on the CPS.

After making adjustments for differences in the survey questions, we found that the CPS overstated employment rates for the population as a whole (the percentage of the population that is employed) by 1.3 percentage points. It overstated employment rates for black men by 3 percentage points, and it overstated employment rates for young black men by more than 8 percentage points. These overstatements are substantial. Since people who are not employed are also far more likely to be in poverty and lack health insurance coverage, it is likely that the CPS is understating the extent of poverty and the number of people who lack health insurance coverage.

It would be very expensive for the Labor Department to change its survey methods so it could get coverage rates as high as the Census. However, there are technical adjustments that it can make to its data to ensure that it better captures the extent of poverty in the country. This seems like a really small step, but if we're going to get serious about tackling poverty, we first have to get serious about counting it.

Dean Baker is the co-director of the Center for Economic and Policy Research (CEPR). He is the author of The Conservative Nanny State: How the Wealthy Use the Government to Stay Rich and Get Richer ( www.conservativenannystate.org). He also has a blog, "Beat the Press," where he discusses the media's coverage of economic issues. You can find it at the American Prospect's web site.

center for american progresscenter for economic and policy researchdean baker