Measuring Poverty: A New Approach

Our panel first met in June 1992 and, over two-and-a-half years, worked to come to grips with the range of conceptual and statistical issues involved in defining and measuring poverty and in setting standards for assistance programs. We were very aware of the importance of the poverty measure, which serves as a key social indicator and also determines eligibility for benefits for many government assistance programs. We were also cognizant of the intense interest in the poverty measure among the policy and research communities. Hence, we took steps to educate ourselves as fully as possible about the issues and to ensure that we heard a broad range of views. We held numerous meetings to which we invited staff from many executive and congressional agencies, as well as researchers and analysts with expertise in particular areas. We sent letters to more than 150 researchers and analysts asking for their views on key issues. We reviewed the large body of literature on poverty measurement both in the United States and abroad. Finally, with help from federal agencies, we conducted extensive data analyses of our own.

This report of our work is organized into three distinct parts of disparate lengths. First, a summary highlights key findings and lists all our recommendations. Second, Chapter 1, titled "Introduction and Overview," provides both background on the topic and the arguments for our recommendations; it is designed for a nontechnical audience. Third, Chapters 2-8 (and Appendices B-D) provide detailed reviews and technical analyses of many of the issues related to poverty measurement and the determination of program benefit standards.

On the basis of our deliberations, we recommend a new official poverty measure for the United States. Our recommendation is to retain the basic notion of poverty as material deprivation, but to use a revised concept for setting a threshold and a revised definition of the resources to be compared with the threshold to determine if a family or individual is or is not in poverty. Equally importantly, we recommend procedures for devising an equivalent poverty threshold for families of different sizes and for families in different geographic locations and for updating the poverty threshold over time.

The current poverty measure has weaknesses both in the implementation of the threshold concept and in the definition of family resources. Changing social and economic conditions over the last three decades have made these weaknesses more obvious and more consequential. As a result, the current measure does not accurately reflect differences in poverty across population groups and across time. We conclude that it would be inadvisable to retain the current measure for the future.

In deciding on a new measure to recommend, we used scientific evidence to the extent possible. However, the determination of a particular type of poverty measure and, even more, the determination of a particular poverty threshold are ultimately subjective decisions. "Expertise" can only carry one so far. To help us choose among alternatives, we developed a set of criteria, namely, that the poverty measure should be understandable and broadly ac ceptable to the public, statistically defensible (e.g., internally consistent), and operationally feasible. Finally, for the most judgmental aspect of a poverty measure, namely, setting the level of the threshold, we recommend a specific procedure to follow—but we do not recommend a precise number. We suggest a range that we believe provides reasonable limits for the initial poverty threshold, but we leave the ultimate choice of a specific value to the policy arena.

We also considered the possible relationship of the proposed poverty measure to eligibility and benefit standards for government assistance pro grams. The issues in this area are complex. For many reasons, there is no necessary relationship between a statistical measure of need and the extent to which programs can or should be devised to alleviate need. We do not offer specific recommendations, but we hope that our discussion of the issues will provide some helpful insights for the ongoing policy debate. We note that our discussion, of necessity, refers to assistance programs as they operated in 1992 -1994.

One member of our panel, John F. Cogan, dissents from the panel's decision to recommend a new poverty measure for the United States. He believes that it is inappropriate for a panel of the National Research Council to make such a recommendation, and he questions some of the panel's analysis in his dissent (Appendix A). Although Professor Cogan raises some important issues, we are confident that careful readers of the report will find that we have dealt thoroughly with all of them.

Professor Cogan also questions the scientific basis for our recommendations. There is, indeed, judgment as well as science informing many of the decisions that underlie the recommendations in this report. That is why the panel has taken great care to make clear at each step in the report the character and status of the scientific evidence and the role of judgment. Again, we are confident that careful readers of the report will see clearly how we have dealt with the interplay of science and judgment at every step.

But the panel concluded that it would not serve the public interest for our report simply to lay out the many possible alternatives to the current poverty measure or simply to call for more research on the topics where that might advance our knowledge or reduce the range of possible alternatives. The current U.S. measure of poverty is demonstrably flawed judged by today's knowledge; it needs to be replaced. The panel believes that the measure recommended in our report is a significant improvement over that current measure, and we urge its adoption.

Over time, we know that the nature of scientific evidence will change and the subjective judgments of what seems appropriate today will probably change as well. That was surely one important reason for convening this panel, since the current poverty measure was informed by early 1960s-vintage knowledge and perceptions. It is also the reason we recommend that a process be established for periodic review of the poverty measure (as is done for other key social indicators, such as the Consumer Price Index).

I know that I speak for all the members of this panel in expressing gratitude for the privilege of serving on it. Its purpose is an important one, and we have each learned much from our work over the past two-and-a-half years.

Robert T. Michael, Chair
Panel on Poverty and Family Assistance

NAS Home Page | NAP Home Page | Reading Room | Report Home Page