Widely-Reported Flat Poverty Rate May Be Deceptive

February 17, 2011

A New York Times editorial cites one of the Census Bureau’s alternative poverty estimates as evidence that “the safety net, fortified by stimulus” kept the number of people in poverty from rising in 2009.

For this, it relies on an analysis by the Center on Budget and Policy Priorities — the same one I used to arrive at a similar, though more cautious conclusion.

“Sorry,” says Shawn Fremstad, Director of the Inclusive and Sustainable Economy Initiative at the Center for Economic and Policy Research. “Poverty really did increase in 2009.”

True, the expanded food stamp benefits and tax credits that were part of the economic recovery act may have kept poverty from increasing as much as it would have otherwise. But they didn’t offset the impacts of massive job losses and related losses of health insurance.

According to Fremstad, the alternative poverty rate didn’t go up in part because the alternative poverty threshold that produced the no-increase result went down. This, he says, was also true for the threshold used to produce the official poverty rate, but the decline was smaller — slightly over a third of a percent, as compared to 1%.

The gap reflects differences in the data sets Census uses to establish the thresholds.

As I’ve written before, the official threshold is set at three times the early 1960’s cost of the U.S. Department of Agriculture’s Economy Food Plan, adjusted for inflation. The alternative threshold at issue is instead tied to the amount that moderate-income households spend on housing, utilities and food.

When the housing market tanks, as it certainly has, the alternative threshold won’t keep up with the overall inflation rate — even actually decline, as it did in 2009. This could boost some people above the cut-off, though they were as income-poor as those who fell below it were in 2008.

But if their housing costs were actually lower, wouldn’t their resources come closer to covering their basic needs? For the purposes of the poverty measure, that depends on what counts as a basic need.

Which brings us to Fremstad’s second point. The no-increase alternative measure doesn’t fully account for medical costs. Instead, it adjusts only for out-of-pocket medical expenditures, e.g., deductibles and co-pays.

Sounds reasonable enough until you consider what can happen when people lose health insurance, as 4.4 million did in 2009.

Some will be well enough off to pay for essential health care costs, notwithstanding the bigger drain on their resources.They’ll seem to be poorer because the measure picks up their costs.  Others will forgo care. They’ll seem to be relatively better off, though they could well be poorer than those who continue to pay for care.

Fremstad says the Census Bureau actually did publish some alternative poverty measures that include medical expenses, rather than just out-of-pockets. These produced higher thresholds than in 2008 and somewhere between 1.1 million and 1.8 million more people in poverty.

Still less than the 3.74 million in the official estimate, but enough to suggest that the poverty rate didn’t stay flat — if the test is whether people could afford essential expenditures.

Lastly, Fremstad notes that the Census Bureau counted the full value of refundable tax credits as 2009 income, even though “nearly all” the families who gained from the expanded Earned Income Tax Credit and Child Tax Credit got their benefits as a lump sum in 2010, i.e., after they filed their 2009 tax returns.

So they were no less poor in 2009 than they would have been with no refunds at all.

None of this is to say that CBPP erred in finding that major safety net programs, including the expansions effected by the recovery act, kept some millions of people out of poverty. Nor that the Times is wrong in saying that Congress should “take a good look at those numbers … before it commits to any more slashing and burning.”

But it does, I think, show how urgently we need a single, reliable poverty measure to tell us how many poor people there are — and who they are — at any given time and over time.

As the Times editorial indicates, this is not just of interest to economists and others of a wonkish bent. It’s got real world consequences for policymaking and for a still-unknown number of poor people affected by the policies made.


New Angles On How Many Poor People There Are In The U.S.

January 20, 2011

I remarked some time ago that we didn’t know how many poor people there were in the U.S. We still don’t because the Census Bureau is still working on a measure that would take account of many factors the official measure ignores.

As part of the process, it’s been releasing annual alternative poverty estimates based on recommendations the National Academy of Sciences made back in 1995. The latest set came out in early January — three multi-columned spreadsheets, each with many, many figures.

I couldn’t make heads or tails of them, though I could see that the poverty rate for 2009 might be as low as 12.8% or as high as 17.1%, depending on which NAS recommendations were applied. So  there could have been as relatively few as 39 million people in poverty or as many as 52.5 million.

Fortunately, a new brief from the Economic Policy Institute gives us non-economist the big picture — though not an answer to how many poor people there are.

As EPI explains, the alternative estimates make different kinds of adjustments in the poverty threshold, i.e., the dollar cut-off for counting people as poor, and/or in what’s counted as income.

The official threshold is three times the food budget at the time the official poverty measure was developed, with adjustments for inflation based on the Consumer Price Index for All Urban Consumers.

The Census Bureau produces alternative thresholds by adjusting for out-of-pocket medical expenses, cost-of-living differences in different parts of the country and a different measure of consumer price inflation — the Consumer Expenditure Survey.

Looking only at the alternative thresholds, the share of the population in poverty seems higher than the official 14.3% rate the Bureau reported in September. Hence a high-end estimate of poor people so much greater than the official 43.6 million.

The income adjustments tell a different story.

The official measure counts only cash income, i.e., wages and cash benefits like Social Security and unemployment insurance.

The alternative measures take account of non-cash benefits like food stamps, housing vouchers and Medicaid and of tax credits like the Earned Income Tax Credit and the Child Tax Credit.

With these included, the poverty rate is lower than the official estimate, even when taxes are factored in. As with the thresholds, how much depends on which adjustments are made.

The Center on Budget and Policy Priorities also crunched the numbers. It came to basically the same conclusions about the income adjustments, though with a more political slant aimed at justifying the temporary new and expanded tax credits and benefits in the economic recovery act.

According to CBPP, the recovery act improvements kept 4.5 million people out of poverty. An additional 11 million were lifted above the poverty threshold by the regular versions of five of the programs — the Earned Income and Child Tax Credits, unemployment insurance and food stamps.

And, as EPI also shows, the biggest anti-poverty impact came, as it has in the past, from Social Security retirement benefits. CBPP says these kept more than 20 million people out of poverty. Looking at its table on program impacts as a whole, the number seems more like 21.4 million.

In short, the major federal anti-poverty programs are doing what they’re supposed to do. Without them, a vastly larger number of people would have been poor enough to be counted as such.

I don’t suppose I need add that these programs are at high risk — if not of annihilation, then of significant retrenchments.

Bills That Count

September 19, 2009

I’ve been having a running discussion with some fellow advocates about how to impact public policy. One says there’s no point in presenting hard data. Legislators “aren’t interested in the truth.” They’re interested in getting re-elected. So what we need to do is create a groundswell of constituent pressure.

No doubt about the value of grassroots activism. But I think data have a role to play too, especially when issues are complex and call for complex, evolving solutions. As the business management gurus say, “you can’t manage what you don’t measure.”

So it’s good to see that bills have been introduced in the Congress that would give us a better fix on one of the most complex issues we’ve got–poverty itself.

As I’ve written before, the current federal poverty measure gives us a very incomplete picture of the problem–how many people are poor, who they are and where. Experts, advocates, even policymakers have known this for a long time. But nothing’s happened to change it.

Last year, a bill was introduced to belatedly implement recommendations for an improved poverty measure that was developed by the National Academy of Sciences. This bill–the Measuring American Poverty–has been reintroduced in the House as H.R. 2909 and in the Senate as S. 1625.

Under the MAP Act:

  • The Census Bureau and the Bureau of Labor Standards would develop a “modern poverty measure” generally based on the National Academy of Sciences’ recommendations.
  • The Census Bureau would publish poverty rates based on this measure, as well as on the “traditional poverty measure,” i.e., the one in use now.
  • The National Academy of Sciences would be commissioned to develop recommendations for two additional measures–a measure of the income a family would need to afford the costs of a “safe and decent, but modest standard of living” and a measure of the extent to which individuals are at risk of being unable to afford needed medical care.

The new measures would be only a first step. Under the MAP Act, the “traditional poverty measure” would still be used to determine eligibility for federal benefits and the amounts and kinds of benefits provided. So there’d be no direct impact on poor people, unless state and local governments decided to adopt the “modern” measure for their own programs.

Congress could, of course, switch federal programs to the “modern” measure. But, as the Center for American Progress points out, the new measure is likely to change poverty rates–not only overall, but among populations and among states. So adopting it for federal benefits would be politically challenging–as changes always are when some stand to lose what others gain.

Still, the MAP Act would give us a better understanding of the extent of deprivation in this country and a better benchmark for the effectiveness of anti-poverty programs. It would also give us a radically different measure that could, over time, become the basis for efforts to ensure that everyone in this very wealthy country of ours has a decent standard of living.

But first Congress has to pass the MAP Act, and the prospects don’t look great. The bill has only 13 cosponsors in the House and one in the Senate. And, as columnist Alec MacGillis observes, poverty isn’t high on the Democrats’ agenda–let alone the seemingly wonkish issue of how we define and measure it.

Meanwhile, health care reform is absorbing virtually all the political energy. Congress will nevertheless have to complete work on Fiscal Year 2010 appropriations before members can go home to work on their real top priority–getting re-elected.

So revamping the federal poverty measure may, once again, get put off to another day. Or maybe not. It could get tucked into some other piece of legislation. You never can tell about these things.

How Many Poor People Are There in the U.S.?

March 15, 2009

Since I wrote this posting, the Census Bureau has issued two new official estimates of the number of poor people in the U.S. You can find my summary of the latest here. It has also issued two sets of alternative estimates. You can find my summary of the first set here and a discussion of the second set here.

According to the latest U.S. Census Bureau figures, the answer is about 37.3 million. But if we want to know how many people lack the resources needed for a decent standard of living, then the answer is nobody knows.

Economists, service providers, advocates and many policymakers have known for years that the current poverty measure is outdated–in fact, was flawed from the get-go.

It was developed in 1963, based on a 1955 survey. The survey found that the average family of three or more spent a third of its after-tax income on food. So the basic poverty threshold became three times the cost of the U.S. Department of Agriculture’s Economy Food Plan–a bare subsistence food budget.

This threshold was then spun off into specific thresholds for different family sizes. And certain adjustments were made based on factors related to food purchases. That’s basically what the thresholds are today.

If your income is at or below the relevant threshold, you’re counted as poor. If it’s over, you’re not.  Whether you’re supporting children not living with you doesn’t matter. Nor does it matter whether you have resources other than cash income. Whether you live in New York City or Biloxi, Mississippi doesn’t matter either.

The poverty thresholds are annually updated to reflect rising costs of living as indicated by the Consumer Price Index. But no adjustment has been made for major changes in consumer spending patterns that have significantly reduced the percent of household budgets spent on food.

And the thresholds still takes no account of geographic differences in cost of living. Studies by Wider Opportunities for Women indicate how big those differences can be. For example, in 2005, self-sufficiency for a D.C. adult with two young children required an annual income of $53,634 per year. The same family in Wheeling, West Virginia could have been self-sufficient with $24,321.

In 2005, the poverty threshold for that family of three was $15,735–less than 30% of the D.C. family’s minimal costs of self-sufficiency. And it’s the threshold that determines eligibility for assistance under more than 30 federal programs and some state programs as well.

Last December, the Brookings Institution issued a report recommending a poverty measure based on recommendations in a 1955 report issued by the National Academy of Sciences. It’s quite complex, but then so is the issue.

Basically, the measure would use current data on actual expenditures for a set of basic necessities and resources available to obtain them, after deductions for additional necessary expenditures. Adjustments would then be made for family composition, as well as size, and for geographic differences.

In 2008, legislation was introduced to establish a poverty measure based on the NAS/Brookings approach. It may–and should–be reintroduced in the new Congress.

As the Brookings authors say, it will be politically challenging to put a new measure in place. But we can’t provide needed assistance to poor people unless we can accurately identify them. And we can’t assess current policies and programs unless we know, over time, how many poor people there are.