Poverty Rates in Recent Years

In a recent commentary in the Washington Times, Alan Reynolds says:

No economist who hopes to avoid professional ridicule would try to deny that consumption is a better measure of long-term living standards than the most widely cited income distribution figures, which do not even add transfer payments or subtract taxes.

It's clear why the administration's defenders are pushing this point. Here's a graph of income and consumption based measures of poverty taken from a recent article from the Minneapolis Fed on measuring poverty:

The green line is income based poverty and it has been increasing since 2000. The consumption based measure shows more progress and that's why it is being pushed on some editorial pages. But even with the consumption based measure, poverty is little changed between 1998 and 2003 and the total decline since 1998 has been less than 1%. Thus, while the consumption based measure does not show the increase in the poverty rate that income based measures show, it is still evident that progress has stalled in recent years as compared to the decline from 1993 through 1998.

As Paul Krugman notes in comparing the change in poverty in the U.S. and in Britain:

And Britain’s poverty rate, if measured American-style — that is, in terms of a fixed poverty line, not a moving target that rises as the nation grows richer — has been cut in half since Labor came to power in 1997.

For the same time period, and using the best case consumption based measure, the rate has only fallen by a little over a percentage point over the same time period in the U.S. (see graph). Thus, while it's easy to see why the administration prefers the consumption based measure, even using this measure the U.S. has not done as well as Britain has over the same time period. As Paul Krugman also notes, this is partly due to a difference in the priorities of the two administrations.

I want to defend my colleagues against the claim made by Reynolds that they will face ridicule if they question Reynold's preferred consumption based measure of poverty.

Actually, I'll let who economists who work in this area speak for themselves. This is from the article containing the graph shown above. See if you think these economists ought to receive the "professional ridicule" Reynolds says they deserve rather than the respect accorded to colleagues engaged in serious research on important issues:

Poor by what standard?, FedGazette, Minneapolis Fed: ...Not foolproof Add it all up, and a different pattern emerges regarding poverty. A 2003 Census report on material well-being noted, “As many (studies) show, the levels of poverty and inequality tend to decrease using consumption-based figures, in comparison with income-based measures.”

Recent studies have reinforced that notion. In a 2006 working paper for the NBER, economists Bruce Meyer (University of Chicago) and James Sullivan (Notre Dame) pointed out that the official poverty rate “suggests that poverty has changed very little over the past three decades,” rising with recessions and then subsequently falling. In contrast, “Consumption-based poverty rates often indicate large declines, even in recent years when income-based poverty rates have risen” (see chart).

Responding to questions via e-mail, Sullivan said that consumption “is a more consistent measuring stick over time and that it is a better measure of the well-being of the worse off.” He added, “Over the past three decades, consumption poverty tells a more optimistic story than does income poverty ... suggesting we are winning the war on poverty.”

Some economists prefer to look at consumption because it is less volatile than income on an annual basis for most households. People smooth their consumption based on long-term income expectations. Such a phenomenon is readily apparent among those who lose a job. While their income might plummet, consumption tends to fall much less dramatically. Such households tend to either dip into savings or take on additional debt with the expectation that higher income will return in due time.

All this is not to say that consumption wins the best-measuring-stick debate hands down, even among advocates. Sullivan, for example, acknowledged “some important practical concerns with switching to consumption,” including the fact that consumption surveys are much smaller in scale than income surveys, making it difficult to analyze local patterns because of sampling problems.

The consumption model has other blind spots. For example, it can only measure total costs; it has no ability to distinguish the quality of purchases or the utility of different types of purchases to a household. For example, a 2005 working paper by Thomas Deleire of Michigan State and Helen Levy of the University of Michigan found that higher expenditures among single-mother households during the 1990s “can be explained by a shift from food at home to food away from home.” While that is positive in some senses—less work cooking at home and more food “leisure”—an alternative explanation is that more meals were eaten outside the home out of necessity and at higher cost to the household budget, as more single mothers worked, either voluntarily or because of changes to the welfare system in the 1990s. Better off? Hard to say for sure.

Sullivan and others also point out that income poverty has simple longevity on its side. “I think it is well understood that there are flaws in the official measure of poverty,” Sullivan said. “(But) we have been using the current measure for about 40 years, so we have a nice time series that is generally understood.” A 2005 article in the BLS's Monthly Labor Review noted that most studies of well-being are based on income data “partly because of history and also partly because of habit. Income data are accessible, comparable over time, and of high quality.” International comparisons are possible only through income because other measures like consumption are simply unavailable in most other countries.

Austin Nichols, a research associate at the Urban Institute, a nonpartisan economic and social policy research organization, has authored several recent reports on poverty trends. “I think a lot of folks use the official poverty line for the sake of convenience and comparability,” Nichols said via e-mail. That might sound like faint praise, but Nichols said that “convenience and comparability is not to be scoffed at.” Any new measure would not likely offer a view of poverty dating back to the 1960s and could have “equivalent or greater problems. ... At least the official poverty measure is understood by most people, as are some of its limitations.”

In the end, everything is relative. Not even researchers within the same organization agree on the best way to measure poverty. Gregory Acs is a senior research associate at the Urban Institute. Along with his counterpart Nichols, he has considerable experience with both poverty trends and the definition-measurement issue.

Acs and Austin tend to disagree over the utility of consumption-based poverty measures. According to Acs, “Ultimately, consumption is a better measure of well-being than income, but I think it is harder to measure, and income is not a bad proxy for consumption.” But Nichols responded, “I disagree that consumption is a better measure of well-being,” in part because researchers don't know how much consumption is financed by unsustainable borrowing. He added that consumption measures “have just as many problems as income-based measures.”

This scholarly head butting illustrates the general difficulty of pinning down who is poor and who is not. Said Acs, “I think Austin and I agree that there are pros and cons to all the poverty approaches,” both income and consumption.

The existing measure has stuck because “we have the most experience measuring income ... (and) researchers and policymakers are by now quite aware of its limitations,” according to Acs. “We want poverty to be an absolute measure of deprivation, and I think that's asking too much of any single statistic.”

However you measure it, we can do better.