IN THE PAST two years, the administration has persuaded Congress to cut about $2 billion a year from the food stamp program and to deny benefits entirely to about 1 million people who would otherwise have gotten them. Now the administration wants Congress to cut another $1 billion or so. Still, administration officials would have you believe that they have done little but crack down on abuse and that the poor are "getting more food assistance . . . than ever before." Don't believe it.

Administration officials rely for their conclusions on the undisputed fact that food stamp rolls have been growing longer and the average benefit to people on the rolls has risen faster than inflation. What they fail to point out is that rolls have grown longer because poverty has been increasing. Average benefits are up for two reasons, neither of which redounds to the credit of the administration.

First, the working poor, who got lower-than-average benefits because of their modest earnings, have been denied assistance altogether. When you cut out people with below-average benefits, simple arithmetic (not new generosity) makes the average paid to the remaining recipients rise.

Second--thanks to rising unemployment and falling welfare benefits--the people left on the food stamp rolls are poorer than ever. When a person's income falls, food stamps are adjusted upward to offset part of his income loss. The average benefit rises, but, since food stamp benefits--averaging 47 cents a meal --are not in themselves sufficient to buy even a minimally adequate diet, the person has less money--and less food--than before. Some improvement.

Looking to next year's cuts, the administration has been claiming that 80 percent of the savings are administrative. House nutrition subcommittee chairman Leon Panetta, however, notes that a Congressional Budget Office study done at his request says that administrative savings are more on the order of 3 percent. The discrepancy arises because the administration's number does not take account of the future year effect of a proposed benefit freeze and because it counts as savings an error sanction provision that would actually transfer substantial costs to state governments. The administration also fails to note that its plan to cut benefits for people--most of them elderly or disabled--whose rent payments are high relative to their incomes would cause substantial losses for perhaps 60 percent of current participants.

Under questioning last week by both Republican and Democratic members of the nutrition subcommittee, an administration official acknowledged that future year effects would be much larger than the administration had advertised. But he justified the cuts as necessary for better administration. That's a case Congress may well find unpersuasive. But it ought at least to be argued on its own merits--not on the basis of false claims of altruism backed up by juggled numbers.