Recently the Census Bureau published its annual tally of the American poor, touching off a predictable debate about the nation's progress in reducing poverty.

A standard feature of this debate is the argument between conservatives and liberals about the value of the official statistics. Many conservatives argue that census statistics exaggerate the number of poor. Liberals argue that poverty is grossly understated.The heated debate over technical issues should not obscure the significance of the latest poverty numbers, however. The number of poor Americans remains surprisingly high in spite of seven years of economic expansion.

The poverty rate was invented early in the 1960s. It measures the percentage of Americans with incomes below the poverty threshold, an income level defined as three times the cost of a nutritionally adequate diet.

In 1989 the official poverty line for a family of four stood at $12,675. This threshold is adjusted every year to reflect changes in the cost of living. The latest figures show that 12.8 percent of Americans have incomes below the poverty line.

Conservatives reject official poverty statistics for several reasons. They complain that the Census Bureau ignores non-money resources - in-kind income like food stamps, medical insurance or the rental savings from owning your own home. If these resources were counted, a smaller percentage of families would have incomes below the poverty line.

Liberals counter with some arguments of their own. They insist that the poverty line was originally defined using a family's after-tax income, on the assumption that money paid out in taxes cannot be used to buy food, clothing and other essentials. The Census Bureau measures a family's resources on the basis of its gross income before taxes, minimizing the number of families found to be poor.

Finally, many liberals believe that the poverty threshold should not be defined in terms of a fixed market basket of goods that remains unchanged from one decade to the next. The cost of being poor should rise in line with average living standards, not in line with the price of a consumption level that was minimally adequate a generation ago.

There is some merit in these objections to the official statistics. The Census Bureau should certainly calculate poverty rates using alternative procedures that reflect these criticisms. Census officials have in fact attempted to do so. They publish a variety of tabulations of poverty, permitting users to draw their own conclusions about its prevalence in the United States.

These tabulations do not reveal a picture of poverty that is very different from the one we see in the official statistical series. Poverty rates naturally appear higher if the tax payments of American families are subtracted from their incomes. Poverty seems lower if we count in-kind sources of income. None of these changes vitally affects our assessment of the trend in poverty over the past decade, however.

Recent progress against poverty has been depressingly slow. This conclusion holds up no matter which poverty series is used to measure recent trends.

According to the official series, the poverty rate jumped from 11.7 percent in 1979 to 15.2 percent in 1983. It fell gradually during the economic recovery, reaching 12.8 percent last year. This is significantly above the rate in 1979 when the level of unemployment was somewhat higher.

The same pattern is evident in every poverty series the Census Bureau has tabulated.

Recent progress against poverty has been slow because inequality is surging, not because official statistics lie.