If the government counted as income the full market value of such in-kind benefits as food stamps, housing assistance and medical care, the number of people officially classified as poor in this country would drop by as much as two-fifths, the Census Bureau said in a new study yesterday.

The long-awaited study could have important policy implications. The Reagan administration is currently under attack for cutting programs that aid the poor. But conservatives have argued for some time that the official statistics exaggerate the extent of poverty in the United States, in that they count only cash income and not the non-cash programs that have been among the fastest-growing in the federal budget in the last 15 years.

The official federal poverty line, which is raised each year to allow for inflation, is now about $9,290 for an urban family of four. It is less for rural and smaller households, more for larger ones.

The new study, ordered by Congress in 1980, outlines nine different ways of including non-cash benefits such as Medicare, Medicaid, food supplements and housing aid in the poverty calculation. It found that:

* Under the broadest method of evaluating in-kind benefits, based on the market value of all such benefits received, only 6.4 percent of the population would have been below the poverty line in 1979, instead of the 11.1 percent reported under the all-cash method of calculation. Only 13.6 million persons instead of 23.6 million would have been classifed as poor. The rate for the elderly would have been only 4.5 percent instead of 14.7 percent, and for blacks, 15 percent instead of twice that.

* Under the narrowest method, which excludes medical benefits because their value is difficult to assign and calculates the value of food and housing differently, the group put below the poverty line would have dropped less sharply, to 9.8 percent or 20.7 million people. Other methods produce figures between those two.

The government's in-kind programs, which came to only a few billion dollars in the early 1960s when the present system of poverty measurement was devised, ate up about $72 billion in 1980, according to the new report. The two largest welfare programs, Medicaid and food stamps, are both of the in-kind variety. Such Reagan administration policy setters as former White House aide Martin Anderson and Health and Human Services Undersecretary David Swoap say it is silly to ignore these large sums in the poverty estimates.

Census Director Bruce Chapman, in releasing the study, said it is basically technical, and the bureau is making no recommendations to Congress or the Office of Management and Budget as to how the present poverty definition might be changed.

The study showed that there are still major theoretical disputes over how to evaluate medical benefits and about what cash values to assign to different services. Chapman said OMB is the agency with the legal authority to set the poverty definition.

OMB spokesman Edwin L. Dale Jr. said his agency "has no plans at this time to revise the official definition. Any changes in the future would be made only after a full airing of the statistical issues involved. This narrow technical study provides no basis for drawing the conclusion that the standard should be changed."

The study was conducted with the help of Dr. Timothy Smeeding of the University of Utah. Chapman and aides said there are problems with all nine proffered methods of measuring in-kind benefits.

Among the most basic is how to count medical benefits, since if everyone's actual Medicare or Medicaid benefits were included on an individual basis, it would be akin to saying that the sicker you are, the richer you are. Thus, a welfare client who spent two months in the hospital would be credited with thousands of dollars in free Medicaid services received; if his medical benefits were indeed counted as income he would be so well off as to be ineligible for any other form of assistance--food stamps, welfare or housing payments--even if he didn't have a penny of cash coming in.

One way to minimize the problem would be to use only the insurance value of Medicare or Medicaid eligibility, not actual benefits received. But this also presents some problems.

Another fundamental problem is how to estimate the actual cash value of the benefit to the recipient. One method is to use the market value of the benefit. Another method, called the "cash equivalent value," tries to measure how much cash an individual might accept instead of the in-kind benefit. For example, instead of the right to Medicaid coverage (valued at its insurance value), an individual might choose to accept a smaller amount of cash, which he could spend on anything, and worry about medical bills if he had them.

The third method, "poverty budget share," seeks to determine how much an average person at the poverty level would spend on food, housing and medical care if all his income were in cash. Amounts of in-kind benefits in excess of these amounts are then ignored.

For each of the three methods, the study calculated what the poverty rate would be under three conditions: (l) if food stamps, child nutrition and housing aid, but not Medicare and Medicaid, were counted; (2) if food, housing and medical aid except for long-term institutional care were counted; (3) if food, housing and medical aid including long-term institutional care were counted. The end result was nine different methods for counting in-kind income.

Smeeding, in a telephone interview from Utah, said he prefers the "cash equivalent" method, excluding institutional care. Under this method, with everything counted but institutional care, the portion of the nation in poverty in 1979 would have been 8.7 percent.