Every year at this time, the Census Bureau announces the official poverty rate of the United States. And every year, the same fruitless debate takes place.

Some will point to the successes and urge that we stay the course. Some will point to the failures and call for more spending on anti-poverty programs. Unfortunately, the analyses on both sides are based on faulty data -- because our measures are critically flawed and overstate the number of Americans in poverty.

The official poverty measure counts only monetary income. It considers anti-poverty programs such as food stamps, housing assistance, the Earned Income Tax Credit, Medicaid and school lunches, among others, "in-kind benefits" -- and hence not income. So, despite everything these programs do to relieve poverty, they aren't counted as income when Washington measures the poverty rate.

We're talking about big bucks here. In 2002, the federal government spent $522 billion on low-income assistance programs. But $418 billion of that was not considered to be cash income and not included in calculating any family's income. Did that $418 billion do nothing to alleviate poverty?

It's time to scrap this outdated definition of income. After all, government has changed how it combats poverty: Direct-cash subsidies are out; benefits that can be used only for essentials, such as food, shelter and health care, are in. But because of how we measure poverty, progress goes unreported, even if families are doing better.

Since 1995, the National Research Council has recommended that the Census Bureau include programs that distribute in-kind benefits, such as food stamps, that are the equivalent of cash, and include the effect of taxes and tax refunds such as the EITC.        

And why not? Non-cash assets such as houses and cars are routinely used to assess economic worth. Taxpayers consider an IRS tax refund as monetary income and income taxes as lost income. Yet the Census Bureau ignores the effect of taxes and doesn't count the EITC refund as income.

Studies that do take into account all income and transfer payments to low-income individuals have found a decline in the number of people in poverty. A 2006 study in the Journal of Economic Perspectives reported that if in-kind benefits are included in income, poverty rates in 2003 would have declined from 12.7 percent to 9.9 percent. By counting all income and taxes, the poverty rate falls by more than 20 percent.

The bad accounting in the current system can lead to bad public policy. The misleading figures make it difficult for policymakers to accurately judge anti-poverty programs.

For instance, the EITC, one of the costliest anti-poverty programs, provides an income subsidy to low-income workers to offset taxes and encourage work. Yet the latest poverty figures discount its worth as an anti-poverty program. In reality, the EITC has cut the child poverty rate by 16 percent since 2003.

It's important to have an accurate measure of progress in our war on poverty. Americans want to help those in need, but they want to do it intelligently. That's difficult to do when the data are inaccurate. The current measure assumes direct-cash transfers to be the only effective way to reduce the poverty rate. Lawmakers rightly avoid direct-cash transfers because of the lack of accountability. In-kind benefits, such as food stamps, ensure the money is spent on needs, such as milk and food, and not vices, such as alcohol and tobacco.

The Census Bureau needs to update its measurement of income and poverty. At a minimum, it should emphasize the poverty rate after all government transfer programs and taxes are counted. This will allow Americans to see how effective low-income assistance is in reducing the poverty rate and what types of relief work best.

Rea Hederman is a senior policy analyst in the Center for Data Analysis at The Heritage Foundation (heritage.org).