THE BUDGET DEBATE has now entered its middle stage. Earlier, President Reagan opened it with a long and specific lost of cuts -- a first step in the searching reexamination of federal spending that most Americans consider necessary and overdue. Now the country is beginning to take stock of the precise meaning of those proposals -- what they would hit, and whom. The Congressional Budget Office has completed its preliminary analysis of the impact on the poor, another valuable contribution to this process of assessment. Where the federal budget touches people's lives, legislators and their constituents have a responsibility to consider the meaning of each reduction and to weigh it against all the other possibilities.

The CBO study shows that roughly half of all low-income families would face an income loss under the Reagan program. For most families the losses are not large -- less than 5 percent of income on average. Even a small loss, however, can cause privation, particularly among the very, very poor and, the study shows, 57 percent of families with incomes less than half the poverty line would face losses.

Even this analysis covers only about one-third of the $15 billion in proposed cuts most likely to affect the poor and almost-poor. Several large changes -- like the $1 billion cut in federal funds for Medicaid, the $2.6 billion in Social Security and the several hundred million dollars in fuel aid and other emergency assistance -- were not included.

The CBO's figures also highlight the particularly harsh effect of the administration's plans on one group that has struck us as particularly deserving of help -- welfare families trying to work their way out of poverty. Over 80 percent of welfare families with working mothers, would suffer losses of income; over half would lose more than 5 percent of their annual income.

The CBO should not stop here. The missing two-thirds of the proposed cuts need to be taken into account. It will be much harder to guage the effects of some of these cuts, particularly in areas like Medicaid where state action is an important unknown, and in health, education and social-service programs where the benefits are widely but unevenly distributed among the poor. But simply going through the exercise, with all its uncertainties, will be an important reminder that what we are talking about is not just adjusting numbers on a balance sheet, but changing the circumstances in which people actually live.

This is not to say that no change can be considered if it violates the "hard-harmless" rule. The nation needs to reexamine all the various things that the federal government in recent years has taken upon itself to do, whether by choice, ritual or simple inadvertence. Some of them, Congress may decide, shouldn't be done or can't be done well enough to be worth the effort. Others may be done better elsewhere. But later this year, when the country has finished sorting through the pile, what it is left with is what it should be willing to pay for.