Journal of Policy Analysis and Management

How to Improve Poverty Measurement in the United States


Although there is widespread consensus that the current official measure of poverty in the United States is badly flawed, three decades of discussion and debate have not resulted in any changes to this statistic. To a casual observer, this may seem puzzling in a nation with a long tradition of regularly updated national statistics. The first several sections of this paper discuss current poverty measurement and various efforts at improvement, attempting to answer the question “Why has it been so difficult to improve the official measurement of poverty in the United States?”

While the United States has been embroiled in its own debate over poverty statistics, the European Union has moved in quite a different direction. The next part of this paper summarizes some alternative poverty measurement approaches that are being used elsewhere, with an eye to seeing what we in the United States can learn from these efforts. The final part of the paper makes a series of recommendations about how to move forward with improved measures of economic need, as well as broader measures of deprivation in other key social areas.

Because this paper has necessary space limitations, I will discuss only a restricted set of topics. Hence, I will ignore some important but more technical debates within the literature. I am going to focus almost entirely on so-called “headcount” measures of poverty, which indicate a certain share of people who fall below some definable point that indexes poverty or deprivation. The problems with this approach are wellknown because it does not measure the depth of economic need. People who are poor could become poorer, with no change in a headcount measure of poverty. Headcount measures are widely used, however, because they are easily understood (important for a public and broadly used measure) and are often easier to implement than other more complex measures. Consistent with the use of the headcount, I will focus on a limited number of poverty measurement approaches that have been implemented in the United States or elsewhere. I mention a few alternative approaches below but spend little time discussing them.

The poverty measure in the United States is usually thought of as a measure of serious economic need or economic deprivation. Our historical definition of poverty focuses on income, the economic resources available to a family, rather than on outcome measures of consumption or well-being. “Living in poverty” suggests that a family has so little income that they are unable to purchase the things that we as a society think they need for a minimally decent life. In the United States, this typically means more than merely escaping starvation; it means being able to purchase the goods and services that are necessary to afford adequate and stable housing, find and hold a job (if physically able), participate as a citizen in the community, keep oneself and one’s family reasonably healthy, and provide the things that one’s children need to participate effectively in school.

An income-based measure of poverty requires agreement on at least four major definitional items. In this paper, I primarily address issues related to the first two of these. To begin, one needs to define a poverty threshold, the level of income or other resources below which a family is considered poor. Thresholds that are fixed over time in real terms (that is, they are entirely nonresponsive to economic growth or changes in living standards) are typically referred to as absolute. The official U.S. measure falls into this category. Thresholds that vary one-to-one with income growth (such as a threshold set at 50 percent of median income) are typically referred to as relative. The European Union and the OECD use such measures. It is, of course, possible to have a threshold that changes with income growth but has an elasticity of less than one.

The second necessary definition is a resource measure, defining which resources are counted for each family; the sum of these resources is then compared to the threshold to determine whether the persons in the family are poor. Discussions about the resource definition in the United States have included such issues as whether to include the value of in-kind programs in estimating income or whether to use after-tax rather than before-tax measures of income.

Third, it is important to agree about the level at which income is aggregated and over what time period. The current U.S. definition is based on the resources of all related and coresident family members over a calendar year. I will assume throughout this paper that we are interested in poverty rates based on annual income and that there is an agreed-upon definition of “family” whose annual income is being measured. In reality, there is substantial debate over whether poverty measures should be based on related individuals who live together (families), whether they should also include cohabiters, or whether they should include all coresidents (households). There is also debate over whether income should be measured for longer or shorter durations than one year.

Fourth, it is important to agree about how different family sizes are dealt with. One option is to develop a different threshold for each family size. An alternative option, used in both the United States and Europe, is to develop a threshold based on a modal family size and then calculate the threshold for other-sized families using an equivalence scale that determines the relative income level needed to keep families of different sizes at the same standard of living. Many papers have explored appropriate equivalence scales, and the scale proposed by Betson (1996) has been most used in alternative poverty calculations in recent years in the United States.