Deciding Who’s Poor

By Barbara R. Bergmann

For decades, now, right-wing think tanks have urged a reform of the official method of defining poverty and counting the poor. They have quite rightly complained that, in deciding whether a family should be considered poor or not, the Census Bureau pays no attention at all to valuable government benefits the family may receive. These include the Earned Income Tax Credit, food stamps, housing aid, and Medicaid, whose total worth can run to thousands of dollars a year. Right-wing commentators have gone on to assert what seemed to them an obvious corollary — that if those benefits were properly counted, we would find that there was a lot less poverty in America than the official statistics had led us to believe.

As we all know, you should be careful what you ask for, because your request might be granted. The Census Bureau is moving (perhaps) toward a method of deciding who is poor that would take account of all those previously ignored benefits. But surprise, surprise — its preliminary estimate is that if it did so properly, the poverty rate would go up, not down.

How can that be? To understand that, we have to look at the bad old (but still used) way of counting the poor, and contrast it with some of the new ways that have been proposed. When we do, we see that these revised figures have deep implications for policy and for where the government should spend its money if it truly wants to fight poverty, particularly among women and children.

The Flawed Way We Decide Who’s Poor

Official U.S. poverty lines were set up in the early 1960s based on how much the U.S. Department of Agriculture calculated it would cost to feed an adequate diet to a family of four (working father, housewife mother and two children). The poverty line was set by multiplying the cost of this "economy" food plan by three, since low-income families were thought to spend about a third of their income on food. The only change in the poverty line through time has come from adjustments for inflation. In classing a family as above or below the designated poverty line, the Census Bureau considers only the family’s before-tax cash income.

There is a lot to complain about in that methodology, besides its neglect of important government benefits. The food plan on which it is based was intended only for temporary or emergency use. A USDA survey found that 90% of the families spending the amount allowed under the food plan were not obtaining a nutritionally adequate diet. The average family of four living close to the poverty line devotes only 18% of its expenditures to food. It makes no sense to triple that inadequate food budget to come up with the total a family must spend to meet all their needs, including housing and clothing.

There is a further important objection. The official method does not set minimum standards for any element of the budget other than food, to which a seemingly "natural" nutritional standard can be applied. But food is not the only budget item that merits serious concern; housing, medical care, child care, transportation also qualify. A valid "poverty line" needs to include appropriate provision for these. Taxes — particularly Social Security payroll taxes — affect a family’s ability to consume, but are ignored in the official method.

The only family characteristics that the official food-times-three methodology takes into account are the number of family members and their ages, which are used to set the food budget on which the poverty line is based. But other family characteristics and location also matter. The price of housing, for example, varies considerably by area of the country. Child care and transportation needs, which are large items, depend on parents’ jobholding. People living in rural areas might need a car. Without listing a set of standards for what families consume besides food, we can’t take account of these differences among families. We have no way to ensure that the poverty-line budget for a particular family in a particular place is sufficient to provide a standard deserving of the description "not poor."

Polling-based Methods

A number of Dutch economists, reviving a line of thought that originated with the Harvard sociologist Lee Rainwater, suggest that the population be polled about how much income they think is necessary to keep a family decently. Then the poverty line should be set according to the results. Since setting poverty lines and counting the poor are acts that are intended to have consequences for public policy, it is certainly reasonable that public opinion about such matters should be taken into account. Further, polling would get rid of another troubling feature of the U.S. official method: the real standard of living achievable with a poverty line income never changes, no matter how the average living standard of the country advances. The polling method would allow the poverty line to change through time as the public’s ideas change about what constitutes a style of life for fellow citizens that the public can contemplate without disquiet.

When the public is polled about where to set the poverty line, the average response turns out to be a dollar figure that is about half of whatever the median family income is at the time of the poll (half of the population’s income falls above and half below the median income). This suggests a simple way of setting the poverty line each year — compute the median income and divide by two. As the median income advances over time, so will the standard of living designated as the poverty line. Like the polling method, the half-the-median method abandons the "absolute" view of what poverty that is enshrined in the current U.S. official methodology: namely, that a family is poor if it lacks a certain unchanging (although unspecific) set of commodities. The half-the-median method embodies a "relativistic" view of poverty: that people are to be designated poor if their income is far below the norm of the society they live in. This method of defining poverty was humanely endorsed by Adam Smith in The Wealth of Nations, who said, "By necessaries I understand, not only the commodities which are indispensably necessary for the support of life, but whatever the custom of the country renders it indecent for creditable people, even of the lowest order, to be without."

The half-the-median method has another virtue. It makes it easy to compare poverty rates across borders, even though the real bundle of goods and services that can be bought with a poverty-line income differs from country to country.

The polling method and the half-the-median method share important weaknesses with the official U.S. method, however. They don’t permit a detailed check that needed items of all kinds are covered, and don’t take account of differences in the situation and location of families. For example, some parents need money to pay for child care (which can run above $15,000 for a family with two children under 6) while other parents don’t.

The Detailed "Basic Budget" Methods

To make sure a poverty-line budget provides for everything a family is deemed to need, one has to do the messy work of making a long list of goods and services, with fairly precise specifications as to quantity and quality. It is not a cut-and-dried matter: for example, in housing standards, should a family be allowed separate rooms for their male and female children to sleep in? Should they be allowed a living room? In 1993, Trudy Renwick and I proposed a method of designating poverty lines which we called the "Basic Needs Budget" (BNB) approach based on such a long list. It set up basic standards separately for food, housing, health care, transportation, clothing, child care, and personal care/miscellaneous. We derived basic standards for each commodity group almost entirely from U.S. government-set standards, and took into account whether the parent or parents were in the labor force, the ages of the children, and the place of residence. We also took into account the taxes the family had to pay, and the noncash benefits available to the family from government, an employer, or a relative.

The BNB approach took full account of the government benefits that the right-wing commentators wanted to include as income, and, in addition, accounted for some nongovernmental benefits, such as employer-provided health insurance or child care provided by a relative. Yet it produced a significantly higher poverty rate than the official method. Among single-parent families, the poverty rate given by the BNB method was 47%, while the official method gave 39% for that group. Most notably, the poverty rate among full-time year-round workers who were single parents was 23% by the BNB method, as compared with 9% by the official method.

How could that happen? The difference in the poverty rate shown by the two methods was due largely to how child care and health care expenditures were treated. If we take the BNB budget, and subtract expenses for those two categories, what remains is quite close to the official poverty line. The official method, of course, is silent about what the family is expected to spend on child care and health care. However, if we think back to conditions in the 1960s, when the official poverty lines were set up, we can see that the real answer is that the government expected families to spend absolutely nothing on those items. Married mothers of young children were expected to stay home with their children while their husbands worked. So were unmarried mothers, thanks to welfare. As a result, family spending for child care was assumed to be nil. As to health care, the hospitals and the medical profession had a centuries-old tradition of giving free services to the poor and near-poor, a system that was largely still in place in the 1960s.

In the 1990s, with many more married mothers in the labor force, and new requirements that single mothers get jobs, the assumption that child care costs are nil is no longer tenable. As to medical care, hospitals and physicians have markedly reduced the provision of free services in recent decades, except in emergencies. So in considering whether a family has enough resources to be considered nonpoor, child care and health care needs can’t be ignored, as they were in the past.

When we compare the increase in the poverty rate on account of the newly recognized needs and the reduction because of the newly counted benefits, the new needs turn out to be larger than the benefits. For families with Medicaid benefits, the cost of the medical care they need is included in their poverty-line budget, while the benefits can be thought of as added to their income. So the two cancel out. But lots of low-income families who are eligible for Medicaid don’t get it because they don’t know how to apply. There is little outreach to help them apply. When you put the medical care they need into their poverty-line budget, with no benefits to balance it, their resources cannot pay for that budget, and those families get added to the list of the poor in the BNB method. In the case of child care, only about one or two in 10 of the low-income families eligible for subsidies under federal and state programs actually receive them. For those lucky ones, the need and the benefit cancel out. For the 80% or 90% of families that don’t get that help, but do need the service, the poverty line goes up, and their resources are not enough to cover it.

A Shift to a New Official Method?

In 1995, a panel of experts appointed by the National Academy of Sciences (NAS) recommended that the Census Bureau consider adopting a method very close to that of the BNB approach. Now the Census Bureau has responded by calculating poverty rates using this kind of an approach. Its outside estimate is that under the NAS method, the poverty rate in 1998 would have amounted to 15.7% instead of the 12.7% that the current official method gives. This represents a striking increase of 24%, almost all of it concentrated in families with children under 5.

Already, there are rumblings that we can’t make a change that would give the poverty rate that big a leap. Sheldon Danziger of the University of Michigan and Gary Burtless of the Brookings Institution want to change to the new method, but avoid the controversy that such a leap would arouse. They imply it might be best just to ignore the fact that the new method tells us that the poverty rate is worse than we thought. They say that the level is far less important than the changes that occur over time. So they propose using the new method just to compute future percentage changes in the poverty level. The truth is, however, that a poverty standard does more than track changes — it tells us about the extent of the national problem that should be closest to our consciences. If a more cogent way of measuring poverty says the problem is worse than previously understood, we should respond with better policies, not by introducing fudge factors into the calculation.

Using the new method, it would become very clear that the surest and handiest way to reduce deprivation in the U.S., without reducing work and marriage incentives, would be to make sure that all low-income families are provided with subsidized medical insurance, and that those with children under 6 have access to subsidized child care. Hiding behind the old numbers allows the dire poverty of 15.7% of America to be swept under the rug, along with the consciences of those with the resources to meet all of their families’ needs.