What's Wrong with the Poverty Numbers

Changing Government Policies Would Do Much to Alleviate Poverty


Last fall the U.S. Census Bureau released its annual report on poverty in the United States. The report indicated that the number of people below the official poverty line had risen from 32.9 million in 2001 to 34.6 million in 2002. Worse, the official poverty rate had risen from 11.7 percent in 2001 to 12.1 percent the following year.1

The response to the grim news was predictable. “Everyone’s taking a bump down, and you haven’t seen the worst of it,” declared Syracuse University economist Tim Smeeding. (I just got my degree and landed a job teaching economics. My salary is more than double my stipend in graduate school; I guess I’m not a somebody to Professor Smeeding.)

Self-proclaimed advocates for the poor pounced on the statistics—particularly the estimated 400,000 more children living in poverty—to prove the need for bigger government budgets. Others were quick to point fingers over the news: “The Bush administration has continued to pander to the wealthy through millionaire tax cuts while having no real plan for low- and middle-income Americans,” declared AFL-CIO president John Sweeney.

Although the Census report and the associated media commentary have the appearance of scientific rigor, one cannot help but suspect that the entire enterprise is driven more by politics than by disinterested curiosity. The annual release of the Census data provides a wonderful avenue for media statements by those whose entire profession involves “advocacy” for the poor—those who make a living off the growth industry in wealth redistribution. No matter what the Census reports, such professional Robin Hoods can and will always proclaim the urgent need for additional tax money. If the official poverty rate goes up in a given year, then obviously the press releases point to this dire fact as proof of the emergency.

However, even if the official poverty rate drops in a given year, the advocate for the poor still benefits from the Census announcement. This is because he can always claim, “Although we are making progress in the War on Poverty, we can’t relax our vigilance.” Furthermore he can criticize the statistical method used by the Bureau (see below) and claim that the official poverty rate underestimates the problem. To give a concrete illustration: According to the data contained in the Census report (p. 22), the official poverty rate declined every single year from 1983 to 1989 (15.2 to 12.8 percent). Does anyone think that during those years the AFL-CIO spokespeople issued annual statements of praise for President Reagan’s compassionate supply-side tax policies?

The Census figures are dubious on purely technical grounds. Most obvious, the Census collects its data on income by asking a (hopefully representative) sample of Americans to fill out a survey. Although the report itself claims that “[r]espondents provide answers to the best of their ability” (p. 1), one wonders how the bureau can possibly know this. (Presumably there is not an additional box declaring, “Check here if the above estimates of your income are truthful.”) In any event, those relying on the Census data should keep in mind that these income figures are estimates given by the respondents themselves, not some official measure, such as their W2 forms.

Even if we assume that the self-reported levels of income are accurate, such statistics are still not sufficient to get a true measure of poverty by any common-sense definition. As a Heritage Foundation report documents, many of the officially “poor” in America enjoy expensive durable consumer goods.2Thus a household’s income in any given year may not reflect its members’ actual consumption in the broadest sense. For example, the Heritage report cites the Department of Commerce’s Housing Survey for the United States in 1995,which showed that 41 percent of “poor households” owned their homes (the median value of which was $65,000), 69.7 percent had a car or truck, 99.3 percent had a refrigerator, and 66.3 percent had air-conditioning. In addition, 97.3 percent had a color television, while 49.1 percent had two or more color televisions.

Income Statistics Incomplete

Even setting aside the problem of durable consumption goods (which may provide a flow of benefits even when money income is low), the Census income statistics are objectionable on their own terms, because they measure only pre-tax income, and exclude noncash government entitlements (such as public housing, Medicaid, and food stamps). To hear that someone only makes $5,000 a year conjures up images of the barest survival, until we realize that the taxpayers may be footing his room and board.

Of course, the poverty advocates may cite such cases as proof of the importance of food stamps, public housing, and other government provisions. But this simply assumes that people do not respond rationally to the existence of such generous handouts. After all, far more parents would patronize private schools if there were no “free” government alternatives; current retirees would have invested far more in pension plans if they had not counted on Social Security payments; and private donations to charities would be far higher if the government did not spend (and tax) billions of dollars annually to “help the poor.” In the same way, simply because many people choose to earn little money when the government makes that way of life feasible does not prove that such people are incapable of working to support themselves.

In addition to the omission of noncash benefits, another flaw is that the Census figures do not reflect geographical differences in the cost of living. To its credit, the Census report discloses these and other shortcomings (p. 15), and includes an entire section on alternative measures of poverty.

There are dozens of plausible measures of income, each including some factors while excluding others, and differing in the treatment of medical expenses, child-care expenses, and so on. There is really no way to single out one of the measures as “correct,” and that’s the point: If one wants to paint a picture worse than the official statistics, one can construct measures that imply a higher rate of poverty. This is certainly the case for Arloc Sherman, a senior researcher for the Children’s Defense Fund, who said, “When we mismeasure not only how many Americans are poor, but who they are, we misunderstand their plight . . . [which] makes it easier to ignore them.”

On the other hand, someone can just as easily argue that the official poverty rate overstates the problem. I found the income measure “MI-Tx+NC” to be compelling; this statistic computes the relevant income as money income plus capital gains (or losses), minus income and payroll taxes, plus the value of all noncash transfers (such as food stamps). Using this definition of income (rather than the official measure of pre-tax income) with the same official thresholds of poverty, the poverty rate in 2002 was only 9.4 percent, compared with the “official” rate of 12.1 percent (p. 19).

Bruce Bartlett, senior fellow at the National Center for Policy Analysis, has made similar observations regarding the shortcomings of conventional poverty statistics.3 For example, Bartlett writes that “consumption by households in the lowest 20 percent of the income distribution averaged $13,957 in 1993, while their income averaged only $6,395. Insofar as consumption is a truer measure of living standards, many low-income Americans are far better off than their income alone would suggest.”

Low Incomes, Substantial Assets

How can these people consume so much more than their official income? Besides the in-kind benefits (such as food stamps and public housing) and the possibility of underreporting that we have already considered, Bartlett raises another issue: “[A] large percentage of those with low money incomes are the elderly. Many have low expenses and may have substantial assets.” For example, Bartlett points out that many elderly own their own homes, and thus have no mortgage or rental payments to make. Because of this, they can still “consume” a flow of housing services without reducing their money income. Someone who owns his home can live comfortably on a far lower money income than a younger person who has not had time to accumulate such assets.

Beyond the technical problems described above, the Census approach invites misleading reports by the media. To calculate the number of people in poverty in a given year, and then compare the figure with previous years, gives the impression that we are referring to the same group of poor people. For example, the October 22, 2003, New York Teacher stated, “More than 1 million Americans sank into poverty . . . the U.S. Census Bureau reported in September.”

To understand the problem with such statements, imagine that we are measuring the number of people in a public swimming pool. Suppose that at noon on June 15, we count 50 people in the pool. We then return at noon on August 15, and now count 250 people in the pool. Would it really be accurate to then report, “In the last two months, 200 Americans sank into pool water”?

The pool analogy is not as silly as it first sounds. The official poverty rate merely gives us a snapshot of how many people were earning less than the official income threshold at the time of the survey. But if we turn to the Survey of Income and Program Participation (SIPP; the results of which are summarized on pp. 14–15 of the Census report), we have access to data from longitudinal surveys that periodically interview the same respondents over several years.

Although the following quotes are contained in the Census report itself, you will not see them trumpeted in the media: “According to the 1996 SIPP panel, a little over half of the [poverty] spells lasted 4 months or less (51.1 percent), and about four-fifths (79.6 percent) of spells were over within 1 year.” The Census report goes on to say, “About 34.2 percent of all people were in poverty for at least 2 consecutive months from 1996 through 1999, but only 2.0 percent were in poverty every month of that 4-year period.”

In short, people have variable incomes; someone might be officially in poverty when he fills out a survey, but that doesn’t mean he is condemned to membership in the underclass. The media obsession with static figures causes us to fret about ways of helping “the poor,” when a more accurate picture would have us concentrating on ways to accelerate the upward mobility of those who leave the ranks of “the poor” all the time.

Lest I give the wrong impression, let me assure the concerned reader that poverty is indeed a vexing social problem, even in the wealthy United States. There are undeniably people whose material standard of living is shocking, and the government can definitely take steps to ameliorate the situation. Most obvious, governments at all levels could cut (or abolish!) income and sales taxes, which would allow the poor man’s dollar to go that much farther. They could also reduce (or eliminate!) the contradictory and pointless regulations that stifle entrepreneurship and thus retard economic development of “depressed” areas. More than any official (and ineffective) “jobs program,” the radical move to legalize drugs, abolish the minimum wage, and at long last end the failed government experiment in so-called public schooling would revitalize the nation’s inner cities almost overnight.

There is one last governmental policy reform we should mention—and one perhaps that would prove more effective in raising the incomes of America’s poor than all of the above items put together: Perhaps our society would experience a tremendous reduction in poverty if only our federal government would stop spending billions of dollars subsidizing it. Currently the government tells its citizens: “Tell you what. If you agree to make under $10,000 or so a year, we’ll give you a bunch of money and other goodies.” As Murray Rothbard put it, we can have as many poor people as we’re willing to pay for.

Alas, it is not likely that the public—not to mention the professional advocates for the poor—will support such bold reforms anytime soon. Consequently, those in poverty would do well to take matters into their own hands. As the Census report itself declares, “Those who worked in 2002 had a lower poverty rate than those who did not—5.9 percent compared with 21.0 percent. . . . Among full-time year-round workers, the poverty rate was much lower. . . .” (p. 8).

And there you have it, backed up by cutting-edge statistical investigation: If you want to reduce the likelihood of finding yourself in poverty, get a job. And if you really want to minimize the probability of being poor, get a full-time job and keep it.


  1. Bernadette D. Proctor and Joseph Dalaker, U.S. Census Bureau, Current Population Reports, P60–222, Poverty in the United States: 2002, U.S. Government Printing Office, Washington, D.C., 2003.
  2. Robert E. Rector, et al., “The Extent of Material Hardship and Poverty in the United States.” Heritage Founda-tion, WebMemo no. 187, www.heritage.org/Research/Welfare/wm187.cfm, September 1, 1999.
  3. Bruce Bartlett, “How Poor Are the Poor?” National Center for Policy Analysis, Brief Analysis no. 185, undated, www.ncpa.org/ba/ba185.html.