Each year, with considerable fanfare, the Census Bureau announces its estimate of how many Americans have incomes below the government's official poverty line.

The latest report last month, covering 1987, found that 13.5 percent lived below the poverty line, a bit lower than the early 1980s but higher than at any time during the 1970s. Under the guidelines, for example, any family of four with income under $11,611 in 1987 was in poverty.

The poverty line was developed in the 1960s by the government as a way to measure how many people are in need and require federal aid. The indicator is used not only for general social measurements but also as eligibility guides (along with other factors) to determine federal benefits, such as food stamps or Medicaid. Some programs use other figures related to the poverty line in defining eligibility.

The poverty estimate is the nation's best measure of economic privation and need. Despite the seeming solidity of the numbers -- printed in crisp columns in a comprehensive booklet -- the poverty figures have always been subject to controversy and challenge.

Some specialists, such as Stuart Butler of the Heritage Foundation, a conservative think-tank, believe the poverty estimates are far too high because they are based only on cash income reported by 60,000 households. Butler has complained that no adjustments are made for unreported income or for receipt of noncash benefits like food stamps and subsidized housing.

Others are critical of the way the income cutoffs were initially constructed by the Social Security Administration's Mollie Orshansky a generation ago, saying the government set the formula too high. Other experts maintain the poverty line is distorted because an inadequate housing cost measure in the Consumer Price Index helps overstate the number of poor.

Orshansky also has performed studies that, with certain perfectly plausible statistical adjustments, show the poverty cutoff could have been set much higher initially, meaning that today millions more would be considered poor.

The original poverty line was based on the least generous Agriculture Department food budget for low-income families, and on assumptions about household spending for other items that many experts now consider too low. It looks at before-tax income, not the cash people have in hand after paying taxes. The annual survey also excludes one group of people who are undoubtedly almost all poor -- the homeless living on the streets -- since it is a survey of households.

In the 1960s, when President Lyndon B. Johnson was launching his war on poverty, the Council of Economic Advisers, on the basis of preliminary studies, proposed that any family with less than $3,000 a year income should be considered poor. In that period, median family income (half of all families are higher than the median, half lower) was about $7,000 a year, so the poverty line was 43 percent of the median.

Orshansky, in a series of studies, developed a more precise methodology for determining poverty. Her method, she said, was not intended to measure absolute physical privation, not the bare minimal level on which people could subsist, but a standard somewhat more in consonance with American experience.

She looked at the Agriculture Department's two food budgets adequate to sustain life decently. The economy budget "designed for emergency use when funds are especially limited" and a somewhat more generous plan, costing about 25 percent more, called the low-income budget. Then, she determined from government consumer surveys that the average family spent about one-third of its income on food.

She concluded that to cover its needs, a family would need income three times as high as the minimum food budget. Eventually, the poverty line was set at triple the cost of the economy budget, with variations worked out for families of different sizes.

Criticisms of the poverty line from the conservative point of view have been described by economist June O'Neill in the just-published book "Thinking About America," edited by Annelise Anderson and Dennis Bark.

O'Neill noted that economist Rose Friedman has criticized Orshansky's use of a 3-to-1 multiplier in calculating the poverty line. While the average family may have spent about one-third its cash income on food, conservatives said, the ratio for poor families was lower. Some analysts put it at only 2-to-1 or 2.5-to-1, saying the poverty cutoff should be only twice the cost of the economy budget. According to this line of criticism, Orshansky had inflated the total number of people considered poor.

Another criticism concerns the Census Bureau survey: It does not adjust for an estimated 10 percent under-reporting of income by families, again exaggerating the proportion in poverty.

The last key point centers on the differences in noncash income. It was relatively small in the 1960s and substantial in the late 1980s. Last year, major federal benefits such as Medicare, Medicaid, food stamps, housing subsidies and other items amounted to at least $145 billion for the nation's population. Employer-provided health insurance and pension benefits for workers exceed $200 billion annually. None of this has been counted as income in the family income and poverty statistics because it was not direct cash payments.

A rebuttal to this argument is that most of the noncash benefits go to the more well-to-do in the nation.

Plausible arguments can also be made that the poverty measure is too low. In a 1979 article, Orshansky and Carol Fendler, of the Census Bureau, wrote that the original 3-to-1 multiplier had been based on 1955 government consumption surveys showing that average families spent a third of income on food, but a 1965

The poverty line was developed in the 1960s as a way to measure how many people are in need and require federal aid.

survey indicated that the proportion the average family spent was 29 percent. Other surveys show the food proportion at 20 percent.

On this basis, an argument can be made for using a 4-to-1 or 5-to-1 multiplier, which would place millions more under the poverty line. Similarly, the use of a more generous food budget, as the base, would have the same effect.

Meanwhile, the Census Bureau has sought on an experimental basis to estimate the poverty population if noncash benefits were counted. Depending on how they are valued, the proportion of Americans in poverty might be as low as 8.5 percent, instead of 13.5 percent.

But there are disputes over how to value in-kind benefits, especially whether to include medical benefits and, if so, what they are really worth.

For example, the poverty cutoff for a single elderly person in 1987 was $5,447. As Sheldon Danziger, former director of the Institute for Research on Poverty at the University of Wisconsin, has pointed out, if an elderly person had cash income of $2,000, plus Medicaid coverage with an insurance value of $3,774 a year (the market value in New York) but never got sick, it could not really be argued that this person's real income was above the poverty line.

At one time, the Bureau of Labor Statistics estimated the income needed for an urban family of four to live at a decent level. To live at the lowest reasonable standard in 1981 -- the last time these estimates were made -- required $15,323. That year, the poverty line was $9,287 for such a family.

Such a difference, many experts say, points up the need for more statistical work on how to measure poverty. The Census Bureau and other agencies are at work, but major changes in how the poverty line is determined may take years.