February 23, 2009: POOR MEASUREMENT Series: A Truly New Approach to Measuring Economic Inclusion, By Shawn Fremstad, Director of the Bridging the Gaps project at the Center for Economic and Policy Research
<?xml:namespace prefix = o ns = “urn:schemas-microsoft-com:office:office” />
The federal government bases its poverty measure on a formula that was established in the 1960s and has not been updated since. Many experts and elected officials alike have made repeated calls for the measure to be changed, especially in light of a changed economy that has altered substantially in the nearly half-century that has passed since the federal poverty measure was first set.
Spotlight on Poverty and <?xml:namespace prefix = st1 ns = “urn:schemas-microsoft-com:office:smarttags” />Opportunity is pleased to announce a series of commentaries entitled “Poor Measurement” to discuss this issue. Over the next several months, Spotlight will bring together experts, advocates and policy makers to address how, why, and whether to update the federal poverty measure.
Rebecca M. Blank and Mark H. Greenberg previously introduced the series with the piece “Poor Measurement.”
We need a truly new approach to measuring poverty. Both the current measure and the most commonly touted alternative developed in 1995 by an academic panel of the National Academy of Sciences (NAS) have serious limitations.
A new approach should draw upon important recent developments in poverty measurement and existing public consensus about the resources families need to “get along” at a basic level. To develop a new measure, the Obama Administration should establish a public consultation process with the goal of announcing a preliminary alternative this year, and, after getting additional input, an official version next year.
A new measure should meet two fundamental criteria. First, it should represent the minimum income a family needs to “get along” or “make ends meet” at a basic level. Second, while retaining a focus on income deprivation, it should encompass other forms of deprivation, including hunger, substandard housing, and lack of health insurance. In doing so, it would cut across the various conceptual and issue “silos” that characterize anti-poverty policy and advocacy.
Meeting both goals may seem like a tall order. But we know it۪s doable: both the United Kingdom and Ireland have recently adopted poverty measures that are consistent with these goals, measures that provide the best starting place for us. Before detailing what that measure should look like, I۪ll say a bit about the two biggest problems with the current one, what I۪ll call the “too-low” and “too-narrow” problems.
First, the poverty threshold currently about $22,000 for a family of four is far too low. How do we know? The most common answer draws on basic family budgetsbudgets constructed by experts who make judgments about the goods and services a family needs to live at a basic level, and the income needed to purchase them. These budgets typically result in income thresholds between 150 and 250 percent of the current poverty line.
While the opinions of experts are important, I think we should give more weight than has commonly been the case to the views of another group with direct knowledge in the arts and sciences of making ends meetthe public.
Public-opinion researchers have frequently surveyed people about the minimum amount of income they think is needed to “get along” at a basic level. The typical response in these surveys amounts to between 50 and 60 percent of the median income for a family of four ($37,000 to $44,000 in 2006). Responses to this question have been consistent over time, rising at roughly the same rate as median income.
By comparison, the current poverty threshold is equal to only 28 percent of median incomeless than half of what the public thinks is needed to get along at a basic level. It hasn۪t always been this way. When the official measure was unveiled in the 1960s, it was closer to the public understanding of what it takes to make ends meet, but it has fallen steadily behind over time.
<?xml:namespace prefix = v ns = “urn:schemas-microsoft-com:vml” />
Source: Fremstad (2008).
Of course, the public isn۪t always right. But in this case, public opinion and the basic budgets constructed by experts generally agree. Thus, we have the luxury of being both principled and politically pragmatic when constructing a new measure.
The second problem with the current measure is that it defines poverty in a way that is far too narrow when compared with historical and contemporary understandings of poverty, ones that view poverty as involving the deprivation of a range of basic needs or capabilities and not just income.
If the goods and services needed to meet basic needs were broadly available for purchase at set dollar amounts and if government efforts to help people meet them were limited to writing checks an income-only measure would be sensible. But, of course, this isn۪t the case.
As a consequence, an income-only poverty measure produces odd results. Consider two different children: one in a family with an income of $18,309 and public health insurance; the other in a family with $2 more in income, $18,311, but no health insurance. Under the current measure, the former is poor, the latter is not, even though, all else being equal, the latter is certainly worse off.
Does the NAS alternative to the official measure of poverty better address the too-low and too-narrow problems? For the most part, no, and it seems to create some new ones.
The NAS approach doesn۪t address the too-low problem directly. As a result, in the vast majority of states, effective poverty thresholds and the number of people considered poor would change little or decline. It does address the problem in an indirect way for a few states by taking into account differences in housing costs across the country. As a result, poverty rates would increase substantially in a handful of high-rent states.
However, simply looking at housing costs without addressing the too-low problem for all states also means that poverty thresholds will end up being effectively lower in a number of states, including several in Appalachia and the South. For example, as Census Bureau researchers have found, the poverty rate will increase by 5 points in California, but fall by as many in Arkansas.
As a matter of principle alone, we shouldn۪t be too concerned about this kind of change if a new measure is really a more accurate one; if, that is, we can be confident that the current poverty measure undercounts poverty in California and over-counts it in Arkansas. (The politics of getting such a measure adopted are another matter).
I think, however, it is more likely that the current poverty measure undercounts poverty in both states, albeit by more in California than in Arkansas. We know from basic family budgets which take a variety of geographic cost differences into account that the percentages of California and Arkansas families living below basic budget thresholds is higher than the percentage living below the official poverty line in each state.
The NAS approach also doesn۪t adequately address the too-limited problem.. The NAS approach improves on the current measure by taking certain work and health expenses into account. But it continues to measure poverty solely in terms of income.
Going back to our example, this means that the first child would still be counted as poor under the NAS measure (unless, possibly, she lived in one of the states that ended up with a lower poverty threshold), but the poverty status of the second child would depend on how much her parents spend on health care. If they avoided doctors and preventative care, they would continue to be counted as non-poor; if they paid for such services, they would likely be counted as poor. Having one۪s poverty status depend on whether they spend or skimp on health care seems an odd result.
Why did the NAS panel develop a measure limited to income? In part, I think, due to timing. When the NAS panel developed its measure, relatively little quantitative research had been conducted on multidimensional measures of poverty, and no wealthy nation had adopted one.
A lot has changed since then. Growth in research on multidimensional measures has been exponential and both the United Kingdom and Ireland have adopted them. As a result, we۪re now in a much better position to develop a broader measure.
All that said, as long as income remains central to our economy, it must remain central to understanding poverty. So we should add other relevant dimensions of poverty into a new measure, but income should continue to play the leading role.
Even if one doesn۪t agree that a new measure should address the too-low and too-narrow problems, the NAS measure raises new problems related to accuracy. This can be seen by comparing it with other measures of deprivation and well-being. Examples include food insecurity and the new American Human Development Index, which includes health and education indicators.
If the NAS measure is an improvement, state-level NAS poverty rates should be more strongly associated with state-level indicators of deprivation and human development than the current poverty measure. In a recent paper, I found this wasn۪t the case; in fact, the NAS measure was less associated with such indicators than the current measure.
A related problem with the NAS measure is that it generally produces a much higher poverty rate for the elderly and a lower or relatively unchanged poverty rate for children. Again, if it really is the case that we are undercounting elderly poverty and over-counting child poverty, this shouldn۪t concern us. But we know that children are much more likely to suffer from food insecurity and other forms of hardship than the elderly, so it doesn۪t seem right that the child and elderly poverty rates end up being roughly the same.
What would a measure look like that addresses the too-low and too-narrow problems? The United Kingdom۪s child poverty measure, adopted in 2003 after a 20-month consultation process, provides the best starting point. The measure has three tiers that, according to the government, “captur[e] different aspects of poverty whilst respecting the finding…that income is at the core of people۪s conception of poverty.”
- Income-Poverty Measure Adjusted for Inflation: 60 percent of median income in a baseline year. Updated annually for inflation.
- Income-Poverty Measure Adjusted for Overall Income Growth: 60 percent of median income. Updated annually to remain at that percentage.
- Material Deprivation and Low Income: 70 percent of median income and “materially deprived,” i.e., suffering from certain material hardship.
In adapting this measure for use here, we should retain the second and third tiers largely as they are. For the first tier, a poverty measure not tied initially to a percentage of median income could be used.
Pegging the second and third tiers to a percent of median income is particularly important. It ensures that the measure is updated annually in a manner consistent with the basic value of shared prosperity. That is, when our economy grows, we should all grow together, rather than further apart, as has happene