In making the case for the supplemental income poverty measure proposed by the Administration last week, Matthew Yglesias, who calls the proposed measure an "excellent first step," explains that:

the official US government measure of poverty is based on a very crude estimate. In the 1960s, the average family spend one third of its income on food. So the way the poverty line works is that it calculates an “emergency food” budget for a family, and then triples the resulting number. Modern-day families, of course, spend much less on food but substantially more on housing, health care, and child care so there’s a need for an updated metric to better account for household spending.

An accompanying graphic in his post notes that food now "amounts to around one-seventh [of the median family budget] as the costs of housing, child care, and health care have all risen disproportionately."

This helps explain why they current federal poverty measure isn’t based on a sound conceptual basis anymore (if it ever was), but it also could be read to imply that the proposed measure does more than it actually does to take into account the costs of housing, health care, and child care, and will inevitably result in thresholds that are substantially higher than the current poverty thresholds.

This isn’t the case. Based on the most recently available Census estimates (from October 2009), under the approach proposed by the Administration the reference poverty threshold for a family of four in 2007 would have been somewhere between $23,500 and $27,744, compared to $21,500 under the current measure. Census has made clear that the proposed measure is "different along some dimensions from any estimates that have been produced to date," so these estimates will change, but they are unlikely to move significantly higher than this range. I think the most reasonable guesstimate right now is that the reference threshold comes out near the high end of that range, but that it will almost certainly remain more than $20,000 below the average nationwide amount that family budgets produced by the Economic Policy Institute and other organizations suggest is needed to "make ends meet" at a basic level (as well as the minimum amount that most Americans say in surveys is needed to make ends meet at a basic level).

Moreover, since the proposed approach would also count benefits like the EITC and food stamps that aren't currently counted, you need to discount higher thresholds somewhat for comparison purposes, especially with groups like low-income children who may receive several thousand dollars a year in those benefits. For example, the average EITC for a family with children was $2,488 in 2007.

While the new approach would subtract amounts parents actually spend out-of-pocket on child care, it wouldn't make any adjustment for parents who need quality child care but can't afford it; the same is likely to be the case for health care (although Census suggests it's open to looking at an adjustment for people who are uninsured and simply can't afford to get the care they need). 

Finally, the supplemental poverty threshold could be lower in some states and regions than even the current federal poverty thresholds (depending on whether and how geographic adjustment is done, an issue discussed below), and for certain groups (such as homeowners without mortgages, who will have their own threshold under the proposed approach).

For these reasons, I think progressives should be cautious about implying that the proposed supplemental poverty measure is really a direct measure of the amount families need to meet basic needs, even the limited set of them included in the measure. It might be better to frame this as an “extremely low” income measure, for example, than as a “low” income measure.

Yglesias also faults the current poverty line for failing to take into account "considerable place-to-place variation in prices, especially in the price of housing." And he links to a CAP release on the proposed measure that includes this argument:

The traditional measure includes no adjustment for geographic disparities in cost of living. This means that two families with the same income—one in Tate County, Mississippi and the other in Seattle, Washington—are considered equally as well off despite the fact that [HUD's] fair market rent for a two-bedroom apartment is $574 per month in the former and $987 per month in the latter.

Here, again, some caution is in order. Rents are certainly higher in Seattle than in Tate County, Mississippi, but what this means for poverty measurement is considerably more complex than this discussion suggests. Whether or not these two families with same incomes should be “considered equally well off” depends on a number of real-world factors, most of which will lead us to conclude that the Mississippi family is objectively worse off as the Seattle family, even holding income equal.

First, there are problems with using HUD's Fair Market Rents to make comparisons like these. The actual typical difference in housing expenditures between the two families may be considerably less than HUD's Fair Market Rent (FMR) figures suggest. Some data suggests that the actual geographic difference in expenditures is less than the FMR differences, and at a 2005 National Academy of Sciences workshop on poverty measurement, HUD staff noted a dozen specific problems with using FMR data for geographic adjustment purposes. For these and other reasons, Commerce is not planning to use FMRs to make the adjustments (the economist Rebecca Blank, now at Census, has previously called them “crude”). Also, there may be systematic differences in housing quality that aren't captured in the data that's available.

Second, as was noted at the 2005 workshop, “rents reflect both amenities and disamenities of a geographic area.” These amenities and disamenities clearly impact how “well off” people are in various ways. So, for example, higher rents in Seattle reflect things like its public transportation system and other beneficial public structures that prosperous progressives communities provide to their citizens as well as greater employment and educational opportunities. These public amenities improve the quality of life for residents of progressive communities. Many if not most of these amenities are not available in conservative, economically depressed, rural counties in the Mississippi Delta.

Finally, considerable other evidence suggests that low-income Mississippians are considerably worse off in economic and social terms than low-income Washingtonians. Despite low rents, Mississippi has the nation’s highest level of food insecurity and hunger (17.4%), while the relatively high-rent Washington has a considerably lower rate of food insecurity (11.1%). Similarly, the Delta region of Mississippi that Tate County is located in scores lower on the American Human Development Index (“a numerical measure of well-being and opportunity made up of health, education, and income indicators”) than any other region of Mississippi, a state which itself scores dead last among the 50 states on the index. Finally, the Congressional District that Tate County is located in, Mississippi’s 1st, ranks 416th out of 436 districts on the American HDI; by comparison, the Congressional District that Seattle is located in, Washington’s 7th, ranks 28th out of 436 districts. You might want to ask yourself, if you were given the same amount of income and the choice of growing up in either Seattle or Tate County, Mississippi, which one would you choose: 28th or 416th?

Related to this is an important question about whether geographic adjustment should be part of any measure that is used for the purpose of a national poverty reduction target. Do we want to bring people in Tate County, Mississippi up to a national standard, or should we set the bar lower for because rents are lower there? I’m strongly in favor of a national standard for Tate County, Mississippi, not a dismal Mississippi one. Similarly, poverty estimates produced by CLASP in November 2009 using Census' own on-line calculator for producing alternative poverty measures found that West Virginia's poverty measure would be lower than Massachusetts' (12.3% vs. 14.2%). I have a hard time believing that the West Virginia social and economic model does a better job of fighting poverty than the Massachusetts one, and other direct measures of deprivation and economic don't support that conclusion. These numbers are likely to change somewhat as Census refines its measure, but they need to change quite a bit to be consistent with what most data tells us about geographic differences in deprivation, and what state-level social and economic policies work best in reducing poverty.