Thursday, November 10, 2011

An Alternative Poverty Measure from the Census Bureau

When the Census Bureau released its annual estimates of the poverty statistics in September, I mentioned some of the main themes in U.S. Poverty by the Numbers. I also mentioned that the Census Bureau was going to follow up with a report offering an alternative measure of poverty, which has now been published. Kathleen Short describes "The Research Supplemental Poverty Measure: 2010" in Current Population Reports P60-241.

As a starting point, here's my one-paragraph overview of the genesis of the current poverty line, taken from Chapter 16 of my Principles of Economics textbook available from Textbook Media:

"In the United States, the official definition of the poverty line traces back to a single person: Mollie Orshansky. In 1963, Orshansky was working for the Social Security Administration, where she published an article called "Children of the Poor" in a highly useful and dry-as-dust publication called the Social Security Bulletin. Orshansky's idea was to define a poverty line based on the cost of a healthy diet. Her previous job had been at the U.S. Department of Agriculture, where she had worked in an agency called the Bureau of Home Economics and Human Nutrition, and one task of this of this bureau had been to calculate how much it would cost to feed a nutritionally adequate diet to a family. Orshansky found evidence that the average family spent one-third of its income on food. Thus, she proposed that the poverty line be the amount needed to buy a nutritionally adequate diet, given the size of the family, multiplied by three. The current U.S. poverty line is essentially the same as the Orshansky poverty line, although the dollar amounts are adjusted each year to represent the same buying power over time."
It has been argued for at least a couple of decades that while a poverty line defined in this way is workable, it can be improved. Back in 1995, a National Academy of Sciences Panel made some recommendations for a new approach to measuring poverty. Kathleen Short summarizes the main concerns of the NAS panel near the start of her report: 
  • "The current income measure does not reflect the effects of key government policies that alter the disposable income available to families and, hence, their poverty status. Examples include payroll taxes, which reduce disposable income, and in-kind public benefit programs such as the Food Stamp Program/Supplemental Nutrition  Assistance Program (SNAP) that free up resources to spend on nonfood items.
  • The current poverty thresholds do not adjust for rising levels and standards of living that have occurred since 1965. The official thresholds were approximately equal to half of median income in 1963–64. By 1992, one half median income had increased to more than 120 percent of the official threshold.
  • The current measure does not take into account variation in expenses that are necessary to hold a job and to earn income—expenses that reduce disposable income. These expenses include transportation costs for getting to work and the increasing costs of child care for working families resulting from increased labor force participation of mothers
  • The current measure does not take into account variation in medical costs across population groups depending on differences in health status and insurance coverage and does not account for rising health care costs as a share of family budgets.
  • The current poverty thresholds use family size adjustments that are anomalous and do not take into
    account important changes in family situations, including payments made for child support
    and increasing cohabitation among unmarried couples.
  • The current poverty thresholds do not adjust for geographic differences in prices across the nation, although there are significant variations in prices across geographic areas."
Over the last few years, the Census Bureau has been rethinking what it means to be "poor," and developing an alternative measure of poverty that addresses many of these issues. It starts with a dollar threshold of what is needed in a certain geographic area to buy a basic set of goods, including food, housing, shelter, and utilities. It attempts to include in income the value of anti-poverty programs that offer in-kind benefits, like food stamps, and those that work through the tax code, like the earned income tax credit. It also includes expenses for income taxes, payroll taxes, childcare expenses, transportation-to-work expenses, and out-of-pocket medical care costs. Instead of looking at a "household" as defined by family dies of birth, marriage, and adoption, the new measure basically looks at everyone living in a "consumer unit" at the same address, regardless of whether they are related.

When all this is done, what picture of poverty in the United States emerges? How does that picture of poverty differ from the official existing poverty rates? Here are some main themes:

The absolute number of people below the poverty line is much the same, but slightly higher. In 2010, there were 46.6 million people below the official poverty line, for a poverty rate of 15.2%; with the new Supplemental Poverty Measure, it would have been 49.1 million people below the poverty line, for a poverty rate of 16.0%. In this sense, my quick reaction is that the existing poverty line has held up fairly well. However, the Supplemental Poverty Measure identifies a somewhat different group of people as poor.

One striking difference is poverty rates by age. Under the official poverty rate, it has long been true that poverty rate for those age 18 and younger is much higher than the poverty rate for those 65 and older: in 2010, the official "under 18 years" poverty rate was 22.5%, while the "over 65" poverty rate was 9.0%. However, under the new Supplemental Poverty Measure, the "under 18" poverty rate would be lower at 18.2%, while the "over 65" poverty rate would be 15.9%. Part of the reason here is that the official poverty rates have a different standard for the over-65 group, while the SPM does not. Food stamps and the earned income tax credit and looking at shared "consumer units" tends to reduce poverty rates among children, while taking out-of-pocket medical care expenses into account tends to increase poverty rates among the elderly.

Other differences emerge as well. Although the overall poverty rate would be higher under the Supplemental Poverty Measure, for certain groups the Supplemental Poverty Measure rate would be lower. For example, the official poverty rate for blacks in 2010 was 27.5% under the official measure, but 25.4% under SPM. The poverty rate for renters was 30.5% under the official measure, but 29.4% under the SPM. The poverty rate for those living outside metropolitan statistical areas was 16.6% with the official measure, but would be 12.8% under the SPM. The poverty rate in Midwestern states was 14.0% with the official measure in 2010, but would be 13.1% under the SPM.

At present, the Census Bureau treats the Supplemental Poverty Measure as "a research operation," and says that it will "improve the measures presented here as resources allow." The official poverty line will remain the line that is used in legislation and as a basis for eligibility for various government programs. This seems wise to me. One great virtue of the existing poverty line is that it isn't changing each year in response to research or political calculations, so it serves as a steady, if imperfect, standard of comparison over time.

But the Supplemental Poverty Measure as it develops seems sure to become part of our national conversation about poverty, because the way it is calculated raises questions about what it means to be for a "consumer unit" to be poor, and what it means to define poverty across a large country with many local and regional differences.