In a post yesterday, I wrote that the new poverty measure proposed by the Obama Administration—known as the Supplemental Poverty Measure (SPM)—likely increases the poverty rate by a modest amount compared with the official rate (less than 1 percent), while producing lower poverty rates in most (34) states. A story in today's New York Times by Jason DeParle and Sabrina Tavernise goes considerably further, saying that the "bleak portrait of poverty" painted by official poverty statistics "missed the mark" with the implication being that poverty is "less bleak" than official statistics suggest.
Oddly missing from the article is any clear mention of the fact that the Supplemental Poverty Measure actually paints a modestly "bleaker" portrait of poverty in 2009, a fact confirmed by Census officials in a media webinar this morning. In fact, as the table below shows, some 2.4 million more people lived in poverty in 2009 than under the official measure.
|People in Poverty in 2009: Official vs. Obama SPM|
|Obama Supplemental Poverty Measure||46,471,000||15.30%|
|Difference (Obama minus Official)||2,442,000||0.8|
|Source: Table 1 in Short, Census (2011)|
The article's implication that poverty is less bleak is based on the NYT's analysis of alternative poverty measures that show that "as much as half of the reported rise in poverty since 2006 disappears" But here, again, the NYT isn't telling the whole story.
The NYT analysis is based on a Census table that compares the official poverty with eight alternative poverty rates from 1999-2009. Four of these measures uses poverty thresholds that are mostly consistent with the Obama measure and the National Academy of Sciences; the other four use thresholds that are not consistent with the Obama measure or the NAS in a fundamental way that affects how they change over time. Unlike the four consistent measures, the four non-consistent measures do not adjust the poverty thresholds for changes in consumer spending on housing and certain other basic needs expenditures as the NAS recommendations and the Obama measure do. The table below shows the differences in both the consistent and non-consistent measures between 2006 and 2009.
|Poverty Rate||Poverty Increase in Percentage Points: 2006-2009|
|Official Poverty Measure||12.3||14.3||2|
|Alternative Poverty Measures||Adjusted for Increase in Certain Basic Needs Expenditures (Recommended by NAS and proposed by Obama)||Adjusted for Geographic Differences in Housing Costs||Medical Needs in Thresholds||13.6||15.7||2.1|
|Out of Pocket Medical Subtracted||14.1||17.3||3.2|
|Not Adjusted for Geographic Differences in Housing Costs||Medical Needs in Thresholds||13.7||15.7||2|
|Out of Pocket Medical Subtracted||14.2||17.3||3.1|
|Adjusted Only for Inflation (Not Recommended by NAS or Proposed by Obama)||Adjusted for Geographic Differences in Housing Costs||Medical Needs in Thresholds||12.2||12.9||0.7|
|Out of Pocket Medical Subtracted||12.6||13.2||0.6|
|Not Adjusted for Geographic Differences in Housing Costs||Medical Needs in Thresholds||12.4||12.8||0.4|
|Out of Pocket Medical Subtracted||12.8||13.3||0.5|
As you can see, the four measures most consistent with the proposed Obama measure all show increases that are equal to or higher than the increase in the official rate.
I'm not a fan of the Obama measure, largely because it, like the current poverty measure ($22,113 for a family of two adults and two children in 2010), sets the poverty line at a level that is far too low compared with mainstream incomes. When established in the early 1960s, the poverty line was equal to nearly 50 percent of median income. Because it has only been adjusted for inflation since then, and not for increases in mainstream living standards, the poverty line has fallen to just under 30 percent of median income. As a result, to be counted as officially “poor,” you have to be much poorer today, compared with a typical family, than you would have in the 1960s. Unforunately, this fact gets no mention in the NYT story, and there is acknowledgement of the reasonable view that both the FPL and SPM are too low as measures of basic income adequacy.
A better standard and one that is commonly used in other counties is half of median income. As even Census has noted, some 22.1 percent of Americans fell below this very-low income standard in 2010, compared with the official rate of 14.3 percent. Similarly, both public opinion research conducted by Gallup and other pollsters, and basic budget analyses conducted by the Department of Commerce and various non-governmental research organizations, suggest that the minimum amount needed to “make ends meet” at a basic level is around $45,000 to $50,000 for a family of four. While the Census report does not report data for such a standard, it does provide data on the number of people with incomes below 200 percent of the federal poverty line, which is a roughly equivalent income level. These figures show that just over one out of every three Americans (33.9 percent; almost 104 million people) in 2010 had income below 200 percent of the poverty line, and that the percentage of such people increased by .9 percentage points between 2009 and 2010, and by 4.6 percentage points since its lowest recorded level (29.3 percent) in 2000.
On the bright side, today's NYT story made included no mention of "poverty rolls."
Update (11/15/11): In a Economix entry posted later on the day of the NYT story, the authors of the story cite my post and acknowledge that "the Supplemental Poverty Measure ... uses the Consumer Expenditure Survey [to adjust poverty threshholds] and there differs from the method we used." To justify their use of an alternative measure that differs from the SPM they argue that:
One of the questions we wanted to ask was whether an alternative measure — by including many billions in increased safety-net spending — gave a different view from the official count of how much poverty has risen since prerecession days.
That question cannot be answered with the measures in the Consumer Expenditure Survey because of a change in its methodology in 2007, contaminating comparisons with earlier years. Therefore, under guidance from the Census Bureau, we chose the Consumer Price Index measure that most closely approximates a new alternative the bureau will release on Monday — the Supplemental Poverty Measure.
I have no doubt that the NYT authors acted in good faith, but they made the wrong decision here (as did Census if they told the authors that the Consumer Expenditure Survey-adjusted numbers could not be used to gauge the effect of safety net spending on poverty over the last several years). The explanation of this is technical and long-winded, but worth putting down here for the record.
- One of the most fundamental ways in which the SPM poverty measures differs from the current official poverty measure is in the way it adjusts poverty thresholds each year. The official poverty measure adjusts the poverty threshold using the CPI. The SPM (and all of the alternative measures actually based on the NAS recommendations--I'll call these SPM/NAS measures for short) adjust the poverty threshold for changes in spending on a basket of four specific "basic needs" items: food, clothing, shelter and utilities. To ensure that no single-year change has a disproportionate impact on the thresholds, the adjustment is made using either the last three years of consumption data (NAS) or the last five years (SPM).
- Thus, to use CPI-adjusted poverty thresholds rather than CEX-adjusted thresholds to judge poverty trends over the last several years is simply not consistent with the SPM/NAS approach. All else equal, under the SPM/NAS approach, if housing increases at a faster rate than inflation generally, the SPM poverty threhshold and poverty rate will increase at a faster rate than the official poverty threshold and poverty rate. This, in fact, appears to be what happened during the pre-Great Recession period in the 2000s—consumption expenditures on the SPM/NAS basic needs basket generally increased at a faster rate than inflation. As a result, the SPM/NAS poverty threshold and rate both increased at a faster pace than the official poverty measure's threshold and rate. After the housing bubble burst and the Great Recession commenced, this process started unwinding and went in the opposite direction.
- The NYT said they used the CPI-adjusted thresholds to assess the impact of social benefits on SPM poverty since the start of the recession. Because the trends in poverty using the SPM can be affected both by changes in consumer expenditures on basic needs and by income (including benefits), the best way to do this is to statistically separate the effects of both of these changes on poverty. New York City's most recent report on SPM/NAS poverty trends there uses this approach. If the NYT had wanted to provide the best assessment of the impact of social benefits on SPM poverty trends, they should have used this same approach.
- Finally, the NYT points to a methodological change made in 2007 as "contaiminating" the CEX numbers. However, this change is a minor one having to do with the wording of a question about how much consumers spent on food away from home. The change is simply too small to "contaminate" the CEX-adjusted poverty numbers in a way that makes them less accurate as a measure of poverty trends over the last several years than ones used by the NYT. Food away from home only accounts for a modest share of the poverty thresholds. In addition, the thresholds are based on three years of consumption data, not one, so the impact of any single year change in a minor component of the thresholds is further diminished. If the NYT had concerns about the CEX-adjusted data, they should either have avoided making any claim about 2006-2009 trends in the first place, or noted the difference in the trends depending on what measure is used.